April 18, 2017: Release 5.0.0-beta

This release is the second testing version of a major update of ELAN and repairs a few bugs and issues present in the previous test release. Some new functions are included as well.

The main changes and new features are:

  • new additions provided by the Sign Language Linguistics group of the Radboud University:
    • an extension for setting up a connection to a Signbank lexicon
    • “inline” visualization of a short video connected to a lexical entry or an external controlled vocabulary entry (requires a specific syntax)
    • alphabetical sorting of tiers via buttons or table headers in several windows and viewers
    • easier access to the “Work with tier sets” flag, a “View by Tier Sets” option in the Density viewer
    • experimental support for spell checking based on Hunspell or a Gecco web service
    • batch wise updating of transcriptions after changes in external controlled vocabularies
  • in the Interlinearization Mode and its lexicon component:
    • in the lexicon properties’ window custom fields can be added to a lexicon, these fields will then be available in the “Edit Entry” window and in the visible columns context menu of the lexicon table
    • improved keyboard navigation between annotations, also between tiers
    • keyboard shortcuts for the actions available for the active annotation
    • added margins between text of an annotation and its ‘bounding box’ and around the tier names etc. Display of the bounding boxes will become optional in a next release
    • in the suggestion window tool tips have been added to each suggestion showing some information from the lexical entry
    • single file find and replace has been enabled in this mode
  • in Transcription Mode find and replace has been activated too, some improvements in the color coding of cells (editable/non-editable, auto create or not)
  • in Transcription Mode merging of annotations and undo of some actions were handled incorrectly (visualization out of sync)
  • the Signal Viewer can now can display wave files that are bigger than 2Gb (up to 4Gb)
  • in Merge Transcriptions the “Overwrite” option now also applies to annotations on Symbolic Association and -Subdivision tiers

The main fixes are:

  • the “Show/Hide More Tiers” window had some dependencies on the selected interface language, these are now removed
  • fixed writing of eaf files containing multiple locales (input methods) with the same language but different variants
  • when creating a copy of a tier with a different linguistic type, references from annotations to CV entries are now removed
  • the “Multiple File Edit” window would show an error when loading newer files, this is now fixed
  • after resorting the lexical entries in the table in Interlinearization mode, editing an entry would select the wrong entry
  • the drop down list of the inline edit box would fail in case of CV entries without a description
  • the .mp4 player on Windows 7 (and some Windows 10 systems) should work again after recompilation of a native library
  • in “elan.properties” a new property has been added (“JMMFPlayer.EndOfMedia.NumberOfFramesBuffer”, default = 5) for setting the number of frames that is used as a “buffer” at the end of the file. The buffer is used to prevent a jump to begin of media when reaching the end.

January 30, 2017: new zip for Mac OS 5.0.0-alpha

The first ELAN 5.0.0-alpha distribution for Mac OS appeared not to be fully functional on macOS Sierra. Logging didn’t work and recognizers and other functions that depend on extensions (like the analyzers in the new interlinearization mode) were non-functional as well. A new distribution has been uploaded on 30 January, Mac OS users are advised to replace a previously installed 5.0.0-alpha with the new one.

January 09, 2017: Release 5.0.0-alpha

This release is a first testing version of a major update of ELAN but it also repairs a few bugs and issues present in the previous release. The most important new addition is the Interlinearization Mode, a text oriented working mode with a lexicon component and analyzers for parsing and glossing.
A lot still has to be improved before the actual release of 5.0.0 in terms of functionality, usability, visualization and documentation. Users who are not interested in testing this new Interlinearization mode can have a look at the solved problems to decide whether or not to download this release.

The main new features and changes are:

  • the Interlinearization Mode:
    • a text oriented working mode with “interlinear style” layout of annotations
    • includes a lexicon component for creating and editing a simple lexicon to support semi-automatic parsing and glossing
    • combines with analyzer modules (extensions) that can perform specific tasks e.g. tokenization, suggesting morphological parses and/or glosses based on a lexicon
    • the manual is gradually updated, some additional practical information on this mode in this version can be found in this pdf
  • the concept of Linguistic Type has been renamed to Tier Type
  • a new version of the EAF schema, version 3.0, adding support for cross-references between annotations. There is no user interface yet for creation and visualization of such references.
  • in the Signal Viewer a zoom option has been added to show the entire .wav file (this can be slow in case of large files)
  • when saving a selection as eaf, the preferences of the current file are copied as well (so that the newly saved file has the same colors and fonts etc. as the original)

The main fixes are:

  • playing a selection in selection mode on Windows shouldn’t change the selection anymore
  • Transcription Mode in combination with a user interface language other than English works again
  • on high resolution displays the font of menus and buttons can appear small. There is no elegant solution but a preference setting in the User Interface tab now allows to set a scale factor for the main font. There are several issues, the most important one being that the row height in tables doesn’t automatically scale with the font. (Also see startup properties in the last item of this list.)
  • when switching between .wav files in the Signal Viewer the time-offset of the viewer is now updated
  • shifting an annotation would sometimes not work (or would appear not to work), it should work correctly again
  • when importing a Toolbox file containing time information in the records but maybe not in chronological order, there is now an option to let the importer sort the records before the actual conversion starts (also see startup properties in the last item in this list )
  • the “changed” flag is not correctly set when updating the location of secondary linked files
  • small improvements in positioning of the CV drop down list, opening a selection in Praat on Windows and in updating the input method language in the Edit Tier dialog
  • in “elan.properties” two new properties were added: “ELAN.UI.FontScaleFactor” for setting a UI font scale factor (see above) and “ToolboxParser.SortRecordsBeforeParsing” for activating sorting of Toolbox records at the start of the import

May 19, 2016: Release 4.9.4

This release repairs a few bugs and introduces one new feature.

The new feature is that of Tier Sets (contributed by the Sign Language Research group, Radboud University). A Tier Set is a user defined, custom group of tiers that can combine tiers independent of hierarchical relations and independent of tier attributes (like Type or Participant etc.).
It can be used to show or hide all tiers in a group in a single action and thus to more conveniently switch between groups of tiers.

  • define tier sets in a separate window (Edit->Edit Tier Sets…), the tiers can be extracted from a domain or from the current transcription
  • the tier sets are stored in an XML file which can be shared with team members
  • the use of tier sets has to be activated in the Preferences tab of the Edit Preferences window
  • the defined tier sets are listed in the context menu of the tier label panel where each set can be made visible or hidden

The main fixes and improvements are:

  • zooming in and out in the timeline viewer can now be done by commonly used key strokes (CTRL+=, CTRL+-, CTRL+0). The effect of these combinations can depend on the working mode.
  • on Linux ELAN could freeze if the VLC player was not installed
  • a few inconsistencies related to line wrapping have been solved in Export as Interlinear Text
  • an “elan.properties” file is now part of the distributions. It contains a few parameters that can be changed by the user. One of them is labeled “ELAN.EAF.Validate” which defaults to “true”. By changing this to “false” XML validation is switched off and (especially large) EAF files will load faster.

May 10, 2016: Release of Simple-ELAN 1.1

This is a bug-fix release. It appeared that some actions with a keyboard shortcut behaved unreliably, leading to unpredictable results. It is strongly advised to replace version 1.0 by this update.

May 04, 2016: Release of Simple-ELAN 1.0

This is the first public release of Simple-ELAN: a simple transcription tool and a companion tool to ELAN. It has limited functionality and (therefore) reduced complexity.
The primary goal is to allow easy, low barrier transcription of audio or video recordings on a single layer (i.e. on a single tier). It can be used for e.g. coarse grained scene-level transcription, per speaker speech transcription or for aligning transcripts of legacy materials after these have been digitized.

The main characteristics are:

  • simultaneous segmentation and transcription by means of the keyboard; split annotations and create new segments by hitting the Enter key
  • transcription on a single tier (in principle)
    • a new transcript contains only one tier unless it is based on an ELAN template
    • when an existing EAF file is opened a drop-down list allows to select one of the top-level tiers to work on
  • supports one audio or video file per transcript and a wave file for visualization of a waveform
  • there can only be one window, one document open at a time
  • annotations are displayed in a vertical layout, top-to-bottom
  • reads and writes ELAN’s EAF format
  • a concise help function is available in the Help menu

Apart from possible general usability improvements, a next version will likely provide support (in some form not yet decided) for simultaneous transcription and translation and for simultaneous transcription of multiple speakers.

February 18, 2016: Release 4.9.3

This is primarily a bug-fix release.

The main fixes and improvements are:

  • in Segmentation Mode it looked like it was impossible to drag-modify an annotation. In fact the screen wasn’t updated properly, this is fixed now
  • in specific cases deleting a time-subdivision annotation could lead to inconsistent data, e.g. if there is a time-subdivision of a time-subdivision tier with (partially) unaligned annotations
  • when changing the font size in the Timeline viewer, the font size of the tier names is now also updated
  • importing a Toolbox file with a custom marker file (.mkr) in which built-in markers are used (e.g. \ELANBegin), no longer fails
  • the Change Case dialog now has an extended tier selection tab pane and the behavior of the check boxes is improved
  • in some places in the user interface and in some export functions there is a PAL and NTSC time format choice. Now there is also a PAL-50fps option, for 50 frames per second videos
  • in the Preferences window there is now a separate panel for the Controlled Vocabulary related settings
  • in the Recent Files menu there is now a tool tip showing the full path for file paths that have been truncated

February 03, 2016: Beta version for Mac OS X with JavaFX player

On Mac OS X ELAN is still dependent on the (outdated) Java 1.6 version provided by Apple. The integration of a “native”, platform specific media player (read: QuickTime) depends on certain components that are no longer there in newer Java versions (provided by Oracle). While a real solution still has to be worked out, this is an effort to integrate a media player which is part of the JavaFX framework. This is a first test version of ELAN for Mac OS X (based on the 4.9.2 code!) with a JavaFX based player, compatible with (requiring) Java 1.8.

It is not an ideal solution, some of the known limitations are:

  • very limited support for media files (basically only H264 (.mp4) and .wav)
  • frame forward/backward only seems to work reliably in case each frame of the mp4 is a Key frame (a.k.a I-frame)*
  • starting and stopping the player (play/pause) sometimes shows little “jumps”
  • play selection sometimes shows a little overshoot
  • the JavaFX framework does not always give access to media “metadata” like the encoded frame rate (in case of .mp4), 25 frames-per-second is used as a default

This version includes a Java 1.8 runtime environment (no need to install it separately). This version has a different distribution “layout” than usual ELAN distributions; all libraries etc. are included in the .app folder. It is not entirely complete (clipping might not work and the extensions folder is not there).
*A short guideline on how to create I-frame only .mp4 files can be downloaded here.

December 21, 2015: Release 4.9.2

This is mainly a bug-fix release, but there are also some changes and improvements.

The main fixes and changes are:

  • modifying the alignment of annotations on Included In dependent tiers could in specific cases lead to invalid annotation structures
  • in single file search the “Add constraint” button didn’t work anymore
  • in single file search case insensitive matching now also works for non-Latin scripts
  • some problems related to highlighting (of the selection and of the media playhead) in the Text Viewer are solved
  • 2D annotation hadn’t been functional for a long time and has now been removed. The “References to Graphics Allowed” checkbox in the Edit Type dialog is no longer there
  • Merge Transcriptions in some cases converted external Controlled Vocabularies into internal CV’s
  • Korean has been added as a language for the user interface
  • 24-bits mono .wav files are now supported by the Waveform Viewer (contributed by Chris Cox)
  • the Copy Tier function now has an option to rename the original (the source) tier so that the copies can have the original name
  • a few changes to the export as Traditional Transcript Text:
    • an option to include line numbers
    • an option to use participant labels instead of tier names
    • the labels for silence duration and for annotation duration are now omitted
  • in the multiple file export as tab delimited text it is now possible to export the file name or file path in a row preceding the annotations of that file
  • in the single file export as tab delimited text it is now possible to include a “header” line with some information about the media files (duration, frame duration)
  • in Segmentation Mode clicking an annotation now also sets the selection
  • some changes in the output of the modified kappa inter-rater reliability function (to make it more in line with EasyDIAg):
    • NaN is written instead of 0 (where applicable)
    • negative kappa values are now written as 0
    • global kappa values excluding “unmatched” are not exported if the matrix is smaller than a 2×2 matrix
    • an “empty” filler value will be written for annotations without values
  • experimental support for opening a transcription and creating a new transcription by drag-and-drop (of either an .eaf file or of media files) into the ELAN window (use with care, some crashes have been observed during testing)

July 27, 2015: Update for MacOS X (4.9.1-b)

This is a minor update of the MacOS X version only; the video player on this platform now also has a zoom function. It is possible to zoom in (up to 400%) via the context (right mouse button) popup menu. Once zoomed in, the visible rectangle can be changed by click-and-drag with the mouse.
Apart from that, the copying to the clipboard of the coordinates of a mouse-click on the video changed; this is now more in line with the implementation on Windows. The coordinate values are now in the coordinate system of the original, encoded size of the video. The old implementation (which depends on the size of the view in the ELAN window) is still accessible via Shift+Alt+Click.

July 15, 2015: Release 4.9.1

This is a bug-fix release.

The main fixes and a few other changes are:

  • Annotations from Subtraction didn’t work anymore in the previous version
  • the Recognizer tab wouldn’t be visible if only video was linked (no audio)
  • exporting hits from the structured multiple file search ended with an error message and not all hits would be saved
  • switching the Suggest Panel on or off in the Controlled Vocabulary dropdown list didn’t work reliably; sometimes the list would appear empty
  • some problems related to playing a selection on Windows, in case of a master media player with a synchronisation offset, have been solved
  • an option has been added to Change Case of Annotations to only change the first character of annotations to upper case
  • the Clip Media using Script action (and other missing actions) has been added to the list of actions that can be assigned to a keyboard shortcut
  • in the Export as Quick Time text it is now possible to select xml as output format, in which case the subtitles will be written in TeXML format (quicktime-tx3g)
  • the implementation of the Recognizer API changed such that a list of linked data files other than audio or video can be provided to a recognizer. For recognizers operating on e.g. time series / sequential data. The API documentation and the demonstration package have been updated.

May 27, 2015: Release 4.9.0

This release contains a number of new features and fixes a few bugs.

The main new features are:

  • calculation of inter-rater reliability based on codings (annotations) in multiple files. Two existing algorithms have been implemented and/or included.
    The old Tier->Compare Annotators function moved to this new “wizard” style compare dialog.
  • interaction with the CLARIN-D WebLicht framework. Web services that expose Natural Language Processing type of tools within the WebLicht framework, can now be called from within ELAN. Possible tasks to perform are tokenization, part-of-speech tagging and lemmatization (mostly for well-described languages).
  • export of annotations from multiple files to the Theme data format (preliminary implementation, not yet documented)
  • two new preference settings concerning controlled vocabulary entries have been added
    • an option to specify the width of the dropdown box for controlled vocabulary entries (in the Timeline Viewer)
    • a flag to specify that the description of cv entries should be shown in the dropdown box as well
  • changes in the Audio and Video Recognizer tab
    • the Audio and Video Recognizer tabs have been merged into one Recognizers tab
    • the window showing the logging messages of the recognizer can now be opened immediately after starting the process, so that messages can be viewed “live” while they are generated
    • in the dropdown list of recognizers, the primary type of data (audio, video, timeseries/other) for each recognizer is now indicated by an icon
    • in the same list an icon can be displayed of the logo of the producer of the recognizer
  • in the Editing pane of the Preferences a new variant has been added to the Copy Annotation function (copies: the annotation, media file name, tier name, start time, end time to the clipboard)
  • in export as QuickTime Text an option has been added to apply the last used font and display settings

The main bug fixes and other changes are:

  • a problem with access to ISOcat (two-letter) language codes has been fixed by including them in the ELAN distribution
  • a small fix has been added for the language dependent display of CMDI metadata that link to the CLARIN Concept Registry rather than to ISOcat
  • it was not possible to change the offset of a timeseries data file in the linked files dialog; that is now working again
  • in Synchronization mode the “player” for timeseries data files did not use and display the correct media duration
  • in the annotation statistics tables, numbers in columns that can contain decimals, now always contain a decimal marker
  • there was a rounding effect in the calculation of annotation times in the Tokenize tier function (only if the destination was a top level tier or an Included In type of tier)
  • the Grid Viewer wouldn’t always correctly store and restore the selected tier and selected mode
  • a potential memory leak in the native media player on Mac OS (Cocoa QT Player) has been fixed
  • the frame forward/backward functions of the Java Direct Show player could cause a jump to the begin or end of media in case of an audio file as the master media (and depending on preference settings)

December 11, 2014: Release 4.8.1

This release fixes one bug:

  • after the recent transfer of ISOcat, the ISO Data Category Registry, to a new hosting provider it was no longer possible to retrieve information on data categories. This function has now been restored so that it is again possible to connect tiers or annotations to a data category.

December 04, 2014: Release 4.8.0

This release contains a number of new features and some bug fixes. Most of the new functions have been developed as part of two CLARIN-NL projects (ColTime and EXILSEA).

The main new features are:

  • a commentary framework that allows to add comments and share them with colleagues
    • the main user interface for this is a Comment Viewer in the upper right corner tab pane
    • comments link to a time segment of the media and optionally to a tier, the time ruler of the Timeline viewer indicates segments with a comment by means of colored bars.
    • the contents of a comment is plain text, a comment has various other attributes
    • the list of comments can be filtered, sorted and configured in several ways
    • the comments can be stored and shared in different ways:
      • comments are stored locally and optionally on a web server
      • a comment file is an XML file (.eafcomment), each comment file refers to one transcription file
      • sharing of comments can be organised in the following ways:
        • via file sharing tools that use a shared folder on the local file system
        • via email, the recipient can import the comment via the clipboard
        • via the web service that allows users with an account to share and synchronize comments
    • a search function allows to search in multiple comment files. An .eaf file with comment can be opened by double clicking a comment in the result list.
  • the list of languages relevant for the content of annotations (introduced in 4.7) can now be edited separately and globally via the Edit menu
  • a Content Language can now be specified for each tier. This attribute can be used to indicate the language used / annotated on that tier. The existing “Default language” attribute has been renamed to “Input Method”. This new attribute can be used for sorting and grouping of tiers in several places.
  • the Annotation Statistics windows for single file or multiple files now have an additional tab for Content Language
  • the Metadata viewer can now display CMDI metadata as well. The display of the CMDI metadata is content language sensitive, meaning that metadata elements can be shown in different languages if translations are available in ISOcat.
  • the ISOcat DCR reference of an annotation as shown in the tooltip is now language sensitive a well
  • linked metadata files are now managed via the Secondary Linked Files tab of the Linked Files window. The link is therefore now stored in the .eaf instead of in the .pfsx preferences file.
  • the Tokenize tier function has been improved in two ways:
    • it is now possible to specify a list of punctuation marks which will be treated as separate tokens
    • the destination tier no longer has to be a dependent tier of the source tier, it can now be any independent tier
  • when importing .csv it is now possible to specify that no annotations should be created for empty cells. As a side effect this allows to import .csv with multiple tiers without enforcing the same segmentation on all tiers.
  • the preview in the import .csv window now updates when selecting a different delimiter
  • two new Preference settings have been added to the Media tab:
    • automatically start play back (of the selection) when an annotation is activated
    • start the player when marking the start time of a new annotation with Shift+Enter

The main bug fixes and changes are:

  • the OK and Delete button in the multiple file selection window remained disabled in specific cases with only folders selected
  • in the Delete tier(s) window it is possible to select multiple tiers in the table. Selecting a tier in the dropdown box now selects only that tier instead of adding it the selected tiers.
  • the tier-related items have been removed from the right-mouse context menu of the Timeline Viewer. The annotation related items are now hidden when there is no active annotation.
  • in the annotation statistics tables a decimal delimiter and at least one decimal are shown in columns containing decimal values
  • Annotations from Subtraction wasn’t always working well, it is functional again
  • on Linux (VLC based player) the initial offset, as retrieved from the eaf, wasn’t restored correctly

September 09, 2014: Release 4.7.3

This release fixes a few problems present in other 4.7.x (or older) versions.

  • the colors of controlled vocabulary entries disappeared after closing and reopening a file
  • in the multiple file chooser window some buttons would remain disabled when only folders were in the list of selected files and folders
  • the default keyboard shortcut for opening the edit box for the active annotation, SHIFT+E, had the side effect of removing previously entered text when entering a capital E. The default now changed to CTRL+SHIFT+E (CMD+SHIFT+E on Mac).
  • minor change in activating the annotation above or below the current active annotation
  • an error that could occur with font sizes in the Interlinear Viewer is fixed

July 04, 2014: Release 4.7.2

This release fixes a few problems present in 4.7.1 and 4.7.0.

  • a problem in the Multiple Files Editor in case of adding dependent tiers: the list of available linguistic types wasn’t always updated correctly
  • when closing a file in Transcription mode a “Save” warning will pop up again if there are uncommitted edits (edits without a final Enter or wihtout moving to the next cell)
  • the SignalViewer in the Segmentation mode now also has the drop down menu for selecting the .wav to display. Some fixes in updating the View->Waveform menu.
  • some improvements in the painting and resizing behaviour in (especially) the Segmentation mode (probably not the last change in this area).
    A setting for redraw behaviour can impact performance considerably. It seems that the best “Painting strategy” settings in Edit Preferences->User Interface are as follows:
    • on Mac OS: Paint to buffer first
    • on Windows: Paint directly to the screen

June 12, 2014: Release 4.7.1

This release fixes two problems introduced in 4.7.0.

  • the import of some file formats (e.g. .csv) wasn’t functional anymore
  • some .wav files weren’t accepted anymore for visualisation in the SignalViewer as a result of too rigid consistency checks

June 06, 2014: Release 4.7.0

This release contains again a blend of new features and bug fixes. One important change is an update in the EAF schema (now version 2.8) connected to this new release. The XML structure of controlled vocabularies is changed such that it breaks backward compatibility. There is an option in the Preferences to always store in the previous version of EAF, version 2.7, and there is a Save as EAF 2.7 menu item for backward compatibility with previous ELAN releases.

The new features are:

  • multilingual controlled vocabularies
    • each entry in a vocabulary can have multiple values, one per language
    • annotations based on a CV entry will display one of these values, depending on the language selected by the user
    • annotations maintain a reference to the id of an entry, CV entries that are part of the EAF now have an id, just as external CV entries
  • it is now possible to specify a number of languages that are relevant for the content of annotations. The languages can be selected from a list of ISO 639-3 languages and added to a local cache. First use of these languages is in multilingual CV’s, other multilingual content will follow (display of Data Categories, Metadata). The preferred language can be set in the Preferences window. A language property will later be added to tiers and annotations (already added to the EAF schema).
    This is unrelated to the language that can be selected for the user interface of ELAN. It is also different from the current tier attribute “Default Language”, which is in fact used for input methods.
  • the user interface of the ISOcat Data Category selection windows has been brushed up and can display information on the categories in all available languages now
  • on Linux a new media player based on VLC player is implemented and is now the default. VLC (version 2.0.x or higher) needs to be installed on the system. There are many issues but it is in general an improvement on the JMF player (Initiated by Ingo Lütkebohle.)
  • in the Edit Multiple Files window it is now also possible to add dependent tiers
  • merging of tiers can now be performed on multiple files
  • the silence (or noise) level for the Silence Recognizer (in the Audio Recognizer panel) can now be specified as a value, with a slider, in addition to the existing example based approach
  • in the control panel with the volume and rate sliders, new volume controls have been added for all media sources with audio. These controls allow to quickly switch between audio tracks or to mix audio sources (this concerns the audibility). By default the audio of the first media file (the master media) is still leading. There is an option in the Preferences to hide the extra controls.
  • in the SignalViewer a dropdown menu now allows to switch between linked wave files (this concerns the visualization)
  • in Segmentation mode a new way of segmenting has been implemented: pressing the segmentation key marks the begin, releasing the key marks the end time of a segment.
    (Contributed by Bob Kuczewski)
  • multiple file N-gram analysis produces statistics on all N-grams found in selected tiers. The size of the N-gram can be set and the results can be saved as tab-delimitd text. The raw data containing the results of many applied algorithms can be exported as well. (Contributed by Larwan Berke and Rosalee Wolfe)
  • FASTSearch is a variant of the simple multiple file search, but should in general be faster. (Contributed by Larwan Berke and Rosalee Wolfe)
  • interaction with Praat should now work without installation of sendpraat. Included in the distribution on all platforms is now a small library that builds on the sendpraat sources and that sends the commands to Praat.
  • the distribution now includes small extensions that can call web services performing automatic recognition tasks on audio or video files

The main bug fixes and changes are:

  • the frame forward/backward function has been revised in order to make it more accurate. This especially applies to NTSC type of videos, videos with a frame duration that is not a whole number. The QuickTime based player now uses built-in next/previous frame functions.
    There is an option to stick to the old implementation in the Preferences Panel, for ongoing projects.
  • for the media player based on Microsoft Media Foundation on Windows a synchronous interaction mode has been added. It can be switched on in the Preferences window.
    It is an alternative to the default asynchronous mode and can be tried in case of instable start/pause response and general behaviour.
  • in export to subtitle text some changes have been made in the handling of, especially small, overlaps
  • the CHAT importer has been improved slightly; it now supports the @Media header, media times without %snd or %mov or without file_name
  • if a font is set for a tier it is now also used in the list of controlled vocabulary entries
  • the size and location of the ELAN windows is now stored more frequently to prevent annoying jumps when changing preferences via the Preferences window.
  • some problems with storing the last used media location and format have been fixed
  • short MPEG videos would sometimes freeze ELAN
  • setting the shortcut for Import Toolbox actually changed the shortcut for the multiple file variant of Import Toolbox
  • some small fixes in the export to Toolbox; the setting for the record marker is not always stored and appending the filename to the record marker now cuts off the extension at the last dot in the name
  • after running a recognizer the results were not always automatically loaded and the media descriptor for segmentations was not always set. This is now fixed.
  • when going from Annotation Mode to Transcription Mode the selected annotation is now maintained
  • some errors in producing the XML of the export tiers for recognizer have been repaired
  • in Transcription Mode when there are no tiers in the table it was impossible to get to the Show/Hide Tiers menu
  • subscribing to the mailing list from ELAN didn’t work correctly, it should now
  • sometimes an eaf would be loaded twice, leading to doubling of all annotations, this shouldn’t occur anymore
  • the default setting for buffered rendering on Mac OS X has changed. In case of general slowness in the user interface, Mac users might want to change the setting via Edit->Preferences->Edit Preferences->User Interface->Painting strategy.

November 07, 2013: Release 4.6.2

This release is mainly intended to fix some bugs and to improve existing functions. But there are also a few new features. The Webstart version of ELAN is no longer maintained.

The new features are:

  • an option to export one or more controlled vocabularies as an external controlled vocabulary file (.ecv)
  • an option to create tiers based on the annotation values of an input tier
  • in the Lexicon view a button has been added to also apply values to depending annotations of the active annotation, provided there are mappings between the depending tiers and lexical entries
  • in the Linked Files dialog it is now possible to edit the “Offset” column in the table, allowing to enter the offset per media file. In this window you can now either change (the order of) the set of media files or you can change the offsets by typing.
  • in the Remove Annotations… function there is now an option to remove all empty annotations
  • in addition to the page forward/backward buttons, it is now possible to jump to a specific hit in the search results panel of the structured multiple file search
  • there is now row highlighting in the structured multiple file search results panel

The main bug fixes and changes are:

  • FLEx export:
    • in steps 1 and 2 more (linguistic) types are now allowed for the phrase and lower levels so that the export process is less likely to get stuck
    • the export failed in step 2 in case there are no “item” tiers on the phrase level
    • now all media files are exported to the flextext file
    • in step 4 the language and item type attributes that were added would not always appear in the drop down lists
  • FLEx import:
    • in some cases multiple identical paragraph-annotations were created “on top” of each other
    • there were several problems with the handling of time information in flextext:
      • overlapping annotations on different tiers would no longer overlap (begin time of annotations were shifted)
      • when there was a mix of aligned and unaligned annotations the calculation of the duration of the unaligned annotations was not done on a per tier basis
      • all time information was lost with the “extract speaker from note field” option selected
    • multiple file import failed with an exception in the first step
  • Annotations from Overlaps with Constraints did not work, always zero annotations were created
  • Annotations from Gaps involving multiple tiers produced incorrect results
  • the Copy Tier function would sometimes fail after creating a new linguistic type
  • setting a font or highlight color for a tier didn’t work anymore (only in Transcription mode it was still functional)
  • the colors and shortcuts of controlled vocabulary entries were lost when a CV was imported
  • the colors of controlled vocabulary entries were lost when two eaf files were merged.
  • the link to a timeseries file wasn’t updated in the configuration file when it was updated in the eaf; track definitions were lost
  • the setting for the folder to store Activity Monitor logs wasn’t (re)stored
  • the preference for last used media folder wasn’t always stored
  • when creating multiple eaf files for media files based on a template, the preferences of the template were lost
  • it was possible to open multiple windows for the structured multiple file search, but queries in different windows could interfere. Now there can only be one search window at a time per ELAN instantiation.
  • jumping to an annotation from the multiple file search results failed if the eaf wasn’t open yet
  • changing the participant in the Multiple File Editor was dysfunctional
  • ELAN crashed when deleting the first annotation (the first row) in Transcription mode
  • save current frame as image would pop up a “overwrite” question message twice
  • the export as Filmstrip failed when it included a waveform and the background color for the last video frame was incorrect
  • the time selection in the Signal Viewer would be displayed in the wrong “location” if a time offset was specified for the wav file
  • after an Undo in Transcription mode the table wasn’t repainted (so that it seemed the undo didn’t work)
  • templates saved from within Transcription mode could freeze ELAN when used for a new annotation file
  • when importing .cha CHAT files invalid XML characters are now filtered out of annotation content
  • the implementation of the “Lookup” type of input methods (ipa-trt, chinese) has been updated. The numeric keyboard can be used to select a character as well. On Mac OS the input methods no longer have to be copied into the Java/Extensions folder.

May 29, 2013: Release 4.6.1

This release fixes a few problems introduced in the previous release:

  • some keyboard shortcuts didn’t work anymore after editing an annotation and until clicking somewhere in the window
  • in Transcription Mode changes in a cell would get lost when activating another annotation by a mouse-click. This behaviour should now be in sync with the preference for “deselecting commits”.
  • in Transcription Mode there were strange crosshair drawing effects in the “direct to screen” drawing mode
  • in Merge Transcriptions the secondary linked files of the first source were not always retained. Preferences and timeseries track settings were not always copied

May 17, 2013: Release 4.6.0

This release contains a combination of new features and bug fixes.

The main new features are:

  • FLEx import/export:
    • FLEx export to .flextext format
      • export with time and speaker information stored in dedicated attributes in the flextext format
      • straightforward mapping from tiers to FLEx elements in case the tier structure is the result of an import of a FLEx file
      • rich (but complex) user interface for defining tier (type) to FLEx element mapping in other cases
    • FLEx import of .flextext format
      • support for flextext format, including the attributes for time and speaker information.
        For each speaker a new set of tiers is created. Time information is only available on the phrase level.
      • changes in import options: morph no longer available for time alignment, new options to customize the creation of linguistic types
      • option to retrieve the speaker id from a “note” field
    • batch conversion to and from FLEx; export multiple eaf files as eaf, import multiple flextext files
  • improvements in the structured multiple file search:
    • saving and loading of queries (in single and multiple layer search)
    • in the tier selection box it is possible to select combinations of multiple restrictions.
      E.g. several tiers by their name, two tier types and one participant, a participant and an annotator etc.
    • new inter-layer constraint: must have same participant. Combinations are possible here as well
    • variable matching: support for the use of variables in queries e.g. X followed by X
    • Alignment view: visualization of temporal alignment of hits.
      With a matrix similar to the matrix in the query construction panel, graphical display of duration and time position of hit constituents at different zoom levels and tier attributes in columns next to the hits
    • the query construction panel can now be hidden in order to have almost the entire window available for the hits
  • when tiers are exported as tiers for recognizer there is now the option to save in the new multi-tier format
  • create a selection exactly spanning several other annotations by alt+clicking them
  • support for the short notation of Praat TextGrid files. Removed the default encoding option when exporting to TextGrid
  • improvements to the Merge Transcriptions function; preferences of the first file are copied
  • in multiple file export to Toolbox it is now possible to sort the tiers. The file name can be added to the record marker.
  • new preferences options have been added. One concerns the painting-to-screen behaviour; painting without buffer leads to less screen artifacts and better quality on e.g Retina display, with buffer is faster, especially on older computers
  • batch clipping of multiple media fragments based on tab-delimited exported annotations from multiple eaf files. (This function is preliminary, untested and undocumented.)

The main bug fixes and changes are:

  • copying of tiers with depending tiers would sometimes lose annotations in case the original was time subdivision and the copy symbolic subdivision.
  • some improvements in Shift mode; undo sometimes didn’t work
  • prevent unnecessary scrolling in the Transcription mode
  • improved handling of escapes and double quotes in import of Praat TextGrids
  • fix in csv export of tiers for recognizer, repair in the ui
  • tiers don’t lose the link to a CV anymore when the CV is replaced via the import CV option
  • on Windows the Native Windows Media Player, that depends on a commercial library, has been removed
  • the default settings for the inline box changed; the Enter key now commits the changes as well as deselecting the box

December 18, 2012: Release 4.5.0

This release contains a mix of new features and bug fixes.

The main new features are:

  • Transcription mode:
    • option to switch off the automatic creation of annotations
    • option to change the tier color from this mode
  • improvements to the user interface of the audio and video recognizer tabs. For recognizers that accept segments as input a more clear choice is now presented between manual selections, segments from a tier or from a file. Several changes in the specification for recognizer extensions are supported, like displaying the help or documentation of a recognizer and the updated xml format for multiple tiers.
  • improved horizontal scrolling in viewers with a timeline. On all platforms shift+scrollwheel now scrolls horizontally. Via the preferences window, category Viewers, the speed of scrolling can be set.
    On Mac OS systems with a trackpad horizontal scrolling and swiping with two or three fingers is now also supported.
  • in audio only situations it is now also possible to have two tab panes by ticking the option “Place video in the center” in the Media category of the preferences window.
  • in merge transcriptions an option has been added to add the secondary linked files to the target transcription
  • in the third step of annotations from overlaps or subtraction it is now possible to create a new linguistic type
  • an option has been added to exclude the participant label from the tab-delimited export
  • sorting of tables by clicking the column header is now functional in the annotation statistics tables, in the Add (etc.) Tier table and in the Add (etc.) Type table
  • several improvements in the search windows such as better layout of results in frequency view
  • added support for some changes in the format of Praat TextGrid files
  • the pause behavior of the JMMFPlayer (mp4 on Windos 7 or 8) can be changed via a Preference option in the Platform category. The player “jumps” to the correct frame when pausing, this can now be switched off (smoother)

The main bug fixes are:

  • multiple file import of Toolbox is now fully functional and handles unexpected situations in a Toolbox text file better
  • when merging two transcriptions the secondary linked files of the first source weren’t copied
  • in change tier attributes the list of possible parent tiers was not always correctly updated
  • some of the user preferences weren’t always saved and restored correctly
  • in the annotation from overlaps function
    • the list of linguistic types wasn’t always correct
    • values weren’t always correctly concatenated if this option was selected
  • subtraction of annotations in the annotations by subtraction function wasn’t always performed correctly (resulting in a wrong number of annotation)
  • the export to HTML function didn’t escape special (xml) characters correctly in tier names and annotation content
  • possible “null” errors when exporting multiple files to tab delimited text
  • input of certain characters with diacritics in annotation mode wasn’t possible, depending the keyboard setting of the system
  • in the TextViewer the “dotted” visualization option was not functional anymore
  • the JMMFPlayer (mp4 on Win 7 and 8) could freeze when rapidly starting and stopping or changing the media time
  • the multiple file search doesn’t disable the tooltips in the rest of ELAN anymore. A bug in saving the hits has been fixed
  • a bug causing the application to hang in merge tier group is fixed
  • it wasn’t possible to remove a link to a data category once it was set for a linguistic type

July 20, 2012: Release 4.4.0

This release contains a mix of new features and bug fixes.

The main new features are:

  • multiple file import (i.e. conversion) from Toolbox and Praat
  • improvements to ‘Traditional Transcript’ export:
    • option to set the number of decimal digits for the silence duration value
    • option to remove blank lines after each annotation (block)
    • option to specify the width (number of characters) of the tier label
    • option to merge multiple annotations by the same speaker with only short silences between them on the same line of the transcript
    • wrap lines setting preference is stored
  • it is now possible to drag a time-alignable annotation (with dependent annotations) vertically from one tier to another (top level) tier
  • some currently predefined properties can be customized via the edit preferences dialog:
    • the color (current default : yellow/orange) used for identifying the non time alignable annotations
    • the information that is copied to the clipboard when copying an annotation (annotation text only or annotation+begintime+endtime)
  • new horizontal divider for resizing the media area in the annotation mode (replacing the icon with arrows on the right hand side)
  • in segmentation mode it is now possible to create continuous segments by having the new begin time coincide with the previous end time (when the media is not playing)
  • improvements to the clip media (with a script) functionality:
    • an option to be prompted for a file name
    • an option to only clip the first (master) media file
    • more error reporting via message windows
    • on Mac OS X a few Applescript scripts are included, allowing to call the QuickTime Player for clipping (esp. with .mp4 and .mov)
  • option to clear the selection with a single mouse click (in the signal viewer and timeseries viewer). It is selected by default
  • the JMMFPlayer (on Windows Vista and Windows 7) now also has a zoom menu that allows to zoom in on the video. The visible section of the video image can be changed by dragging with the mouse.
  • the non-adjusted media time (i.e. ignoring the offset) can be copied via the context menu in the the media players and the signal viewer

The main bug fixes are:

  • fix in the text edit box in Annotation mode: there were problems with typing in letters with accents, or entering characters that require multiple keystrokes to produce one character
  • Windows 7: a problem with loading an eaf in which offset values for mp4 files have been specified is fixed
  • buttons in the delete multiple tiers warning dialog go off screen when the list of selected tiers is very long
  • some annotations were not exported in the ‘Tab-delimited text export’ when the option “Separate column per tier” was selected
  • step and repeat mode in segmentation mode is working again
  • shortcuts were not working after pasting an annotation
  • when creating a new annotation it is made the active annotation. The option to clear the selection after creating an annotation is now taken into account.
  • a problem with changing the duration of overlapping annotations in shift mode is fixed
  • when locating the media via the locate media dialog, the file extension of the media is added again
  • font size preferences of different viewers are now restored correctly while/after switching between different modes.
  • in export dialogs, sometimes duplicate tier names were in the tier list or a few tier names were missing. This is now fixed.

June 6, 2012: Release 4.3.3

This release fixes some recently introduced bugs.

It concerns the following problems with media players:

  • using a QuickTime based player on Windows would fail in many cases (depending on the QT version)
  • erratic behavior of the media player when playing .mov and .qt files on Mac OS
  • double audio playback on Windows 7 when the first media file (the master media) is a ‘.wav’ file followed by one or more ‘.mp4’ files

May 7, 2012: Release 4.3.2

This release is mainly intended to fix some bugs.

The main fixes are:

  • the delete/remove button in the multiple file domain dialog is functional again.
  • linking more than 4 videos could cause errors. On the Mac it could lead to a crash.
  • changing the font size in the Interlinear Viewer is applied correctly.
  • changing the file filter in the save as dialog now correctly updates the extension of the default filename(if suggested).

April 12, 2012: Release 4.3.1

This release fixes a bug introduced in version 4.3.0. When trying to solve a few problems occurring when using keyboard shortcuts while editing in the Inline Edit box, some standard functions in the edit box (like select all, copy, paste etc.) disappeared as well. These keyboard actions are now functional again.

April 10, 2012: Release 4.3.0

This release contains the following new features:

  • multiple file export to Toolbox and Praat
  • on Windows 7 a new media player has been implemented based on Microsoft Media Foundation. It adds support for .mp4 files (and a few other file types). On Vista this player supports .wmv files. The use of this player is optional, it can be switched off via the preferences (the Platform tab in Edit->Preferences). There are still a few issues with this new player:
    • when not playing and jumping to a new position (e.g. frame forward) updating of the video image is fairly slow
    • when dragging a selection the video is not updated very smoothly
    • when pausing or at the end of playing a selection there is a little overshoot
    • sometimes the video image disappears (e.g. when switching between windows or applications); going a frame forward or backward or resizing the area for the media players can force the image to be updated
    • built-in behavior is that the player rewinds (jumps to the beginning) once the cursor reaches the end of the file. ELAN tries to stop the player just before the end is reached
  • export subtitles in LRC and in TTML (Timed Text) format
  • Russian is added as a language for the user interface
  • the distribution for Mac OS contains new properties that make .eaf files searchable with Spotlight. The eaf files have to be imported first by running the command “mdimport path/to/*.eaf” in a Terminal window. Contributed by Tom Honeyman
  • the tooltips for the play/pause buttons etc. now also display the keyboard shortcuts for those actions

Other changes and bug fixes are:

  • the “all files” was missing from some file open dialogs
  • the Create Regular Annotation feature is functional again
  • an uppercase/lowercase problem with file extensions in some file dialogs is solved
  • selecting a folder with .eaf files for annotations from overlaps and subtraction did not correctly add the contained files
  • selecting the “sendpraat” executable when opening a file in Praat had no effect
  • when copying a tier with depending tiers the hierarchical relations were not always correctly maintained
  • multiple fixes and improvements in Transcription mode:
    • more media player options like go to next/previous frame, pixel & second are added as shortcuts actions
    • transcribing a tier with a cv attached is improved
    • working in loop mode is improved
  • the participant marker in .mkr marker files with Toolbox import was ignored
  • tried to fix a problem with the JMF SoundRenderer that caused errors
  • use of regular expressions in single file search on tiers connected to a controlled vocabulary did not work
  • some improvements to the timeseries viewer:
    • restoring of tracks in the viewer after re-opening a file
    • extracting numerical data from non-continuous tracks is more accurate
    • when extracting data from non-continuous tracks, undefined characters appeared for intervals that contain no data
  • the search function in the help system that is distributed with ELAN has improved; highlighting of hits is now more accurate
  • some problems with keyboard shortcuts in the Inline Edit box are solved
  • the Clip media with script function now first looks in the ELAN “home” folder for the “clip-media.txt” file and then in the ELAN install folder. The “home” folder is <user_home>/.elan_data on Windows and Linux and <user_home>/Library/Preferences/ELAN on Mac OS

February 03, 2012: Release 4.2.0

This release contains a number of new features and bug fixes.
This is the first version of ELAN that requires Java 1.6.

The main new features are:

  • annotations from subtraction has been added, to complement the annotations from overlaps and annotations by merging functions.
  • multiple file export of annotation overlaps information. Annotations of a reference tier are compared with annotations of one or more other tiers; overlap information and information on the first annotation after the reference annotation are stored in a tab-delimited text file.
  • on Mac OS the default File dialog is now the native Mac dialog, with a Sidebar, a search field, recent places etc. Via a Preference it is possible to switch back to the old (Java dialog). On Windows a similar File dialog is available after switching to the Windows look and feel via Preferences.
  • creation of multiple transcription files in one action for all media files in a selected folder and sub-folders. Videos can be combined in one transcription based on suffix or prefix.
  • an option has been added to show the number of annotations with the tier names in many drop down boxes.
  • in the Grid Viewer it is now also possible to show (concatenated) annotations of symbolic subdivision depending tiers along with the selected parent tier.
  • in the search results of the single file search columns have been added showing the parent annotation and (concatenated) child annotations for each hit.
  • in the single file search it is now possible to not only query all tiers or one particular tier, but to create a custom selection of tiers to search.
  • in most export functions an extended tier selection tab pane has been added. Tiers can now be selected by tier, type, participant or by annotator.
  • keyboard shortcuts can now be viewed and edited per working mode. For each mode a custom set of shortcuts can be specified.
  • in Segmentation mode splitting and merging of annotations has been added.
  • in Transcription mode play around selection has been added.
  • in the Metadata Viewer long lines are now wrapped in the table and tree cells. An option has been added to hide empty fields.
  • in the Timeline Viewer an option was added to collapse all toplevel tiers (in show by hierarchy mode).
  • to the Toolbox import an option has been added to “scrub” annotations (replace a sequence of multiple white spaces by one).
  • when exporting to Toolbox it is now possible to export a media marker that enables Toolbox to play the sound for each record.
  • in tab-delimited text export an option has been added to export the description of controlled vocabulary entries along with the annotations.
  • a preference to always center the active annotation, a menu for copying the current time to the pasteboard and a shortcut for starting the inline editor have been added.
  • a menu for checking for updates of ELAN was added.
  • added the sec.ms time format to the Grid viewer and the player time label.
  • play selection in loop mode has been improved.
  • crashes due to an error in the Text Viewer are prevented (by not making a new annotation the active annotation).

The main fixes are:

  • the Add Participant function now preserves tier hierarchies.
  • freezing (or very slow editing) in Transcription mode is solved, as well as a problem occurring when changing the column types.
  • several problems with storing/restoring of preferences (visible tiers, sorting mode, visibility of the signal viewer) have been fixed.
  • common mistakes in “Open” and “New” are now dealt with more gracefully.
  • clipping with Praat now takes the media offset into account.
  • fixed the “scrambled” log file, View Log only shows the logging of the current session.
  • the name of the root tier in front of the SD (silence duration) label in the interlinear text export has been removed.
  • when switching modes the changes in an open edit box are saved.

August 15, 2011: Release 4.1.2

This release is mainly intended to fix some bugs.

The main fixes are:

  • with the “snap annotations” option selected, creating an annotation or modifying an existing one would sometimes position the annotation at the wrong (time) location. Existing annotations would be changed without the possibility to undo.
  • on Mac OS X exiting the application by using Command+Q or by using the Quit option in the ELAN 4.1.x menu would correctly show a message in case of changed transcription. But when clicking the Cancel button in this “save changes” window, the application would quit anyway without saving the changes. This is now fixed.
  • on Mac OS X all mnemonics (open a menu with an alt+letter combination) are now disabled. Until now only some were disabled.
  • fixes in Transcription mode:
    • the table would sometimes jump to unpredictable positions when clicking a cell where there is no annotation yet
    • with a tier linked to a controlled vocabulary, activating a cell would always select the first entry, even after hitting the escape key or the enter key with no entry selected.
    • wrapping of text in a cell and updating the row height accordingly has been improved
  • after zooming in the waveform was not always positioned correctly relative to the time ruler. This became especially noticeable with the introduction of the more fine grained zooming levels.
  • in export statistics for multiple files, files in sub-directories were not included. A bug in the calculation of the median value in the participant tab has been fixed as well.
  • in export as Toolbox file the correct alignment of words sometimes appeared to be lost when opening the file in Toolbox. A correction is now made specifically for multibyte characters.
  • importing a CSV file would fail in some cases if there is no header line (with tier names). This is now fixed.
  • in single file statistics empty annotations were not present in the statistics, fixed.
  • colors and keystrokes for Controlled Vocabulary entries in templates were not correctly applied to transcriptions based on the template
  • the alignment in interlinear text export was sometimes incorrect in case of non-spacing modifying characters in the “tokens”

The main new features are:

  • in Transcription mode it is now possible to hide a group of tiers. This option is available in the settings window as well as via the right mouse button popup menu of the table. Via the popup menu one can now also make a single tier non-editable.
  • an option has been added to create empty depending annotations for the active annotation
  • in the tooltip of annotations based on a CV entry the description of that entry is now also visible
  • in the export to HTML it is now possible to include an audio or video tag with a hyperlink per block of text that allows to play the
    corresponding segment. This function requires a browser supporting HTML 5.
  • in the interlinear text export an option has been added to use tabs instead of whitespaces to separate words/tokens
  • the order of the tiers in the timeline viewer is now also the default order in export options. For all export windows the custom tier order is now stored as a preference.
  • the option to Show/Hide more tiers in the tier label part of the timeline view has now been moved to the top level of the
    popup menu
  • a preliminary implementation of an Activity Monitor has been added
  • Simplified Chinese has been added as a language for the user interface

June 01, 2011: Release 4.1.1

This is primarily a bug-fix release but it also contains a few new features.

The main bug fixes are:

  • with the “snap annotations” option selected it was no longer possible to create an annotation before the first annotation of a tier: the new annotation would always appear after the first annotation
  • the order of the tiers in in the timeline viewer would not always correctly be restored when reopening a file
  • the export of a list of annotations from multiple files failed with a “Null” error
  • Transcription mode:
    • wrapping of text was not performed at word boundaries
    • the editing cell was not always visible, the table will now scroll to ensure it is visible

The main new features are:

  • Transcription mode:
    • the cell that is being edited now resizes while entering text
    • the maximal number of columns is no longer limited to three. A new user interface allows to change the number of columns in the table.
  • multiple file search, in the multiple layers tab:
    • dropdown boxes that allow to define constraints between the tiers in the layers have been added
    • filled in information in the grid is no longer lost when changing the size of the grid. By default the grid has now two columns
    • handling of columns in hit export has been improved
  • some menus containing a list of tiers are now “scrollable” when there are many tiers
  • in annotations from overlaps the selected tiers can now be sorted in case concatenation of values is selected
  • on Windows 7 a new approach has been implemented to attempt to use the Elecard MPEG-2 plugin, when installed, for MPEG-2 files
  • the scrollwheel can now also be used when the mouse is over the area where the tier labels are

April 28, 2011: Release 4.1.0

This release contains some major new features and fixes a few bugs.
This version introduces a new working mode, the “Transcription Mode”, designed for easy text entry into existing segments (annotations). It is a common workflow to first create the segmentation of the recording and to enter text in the next step. The Transcription mode is optimized for this second step; text can be typed into the (empty) segments in a spreadsheet-like table that occupies most of the screen. The mode can be activated via the Options menu. The Annotation and the Synchronization mode were already located here, now the Transcription and Segmentation mode have been added as well.

The main new features are:

  • Transcription mode, rapid text entry on all the tiers of user definable tier types, e.g. all speech tiers and all translation tiers. Color coding can be applied to visually distinguish the tiers. Navigation from one cell to the next can be done by keyboard shortcuts. On activation of a cell the media player starts playing that segment.
  • on-the-fly Segmentation moved from the Tier menu to the Options menu and is now a mode in the main window (instead of in a separate window). It is now easier to switch between tiers while the media is playing. Segments are created by keyboard strokes and can be modified by dragging with the mouse. This mode introduces a preliminary step-and-repeat (a.k.a “walker”) playback mode.
  • the Synchronization mode now adds a waveform view to audio only (wave) files. It is now also possible to create and play a selection in the mode.
  • the annotations from overlaps function has been extended with new configuration options and can now also be performed on multiple files
  • annotation statistics can now be produced for multiple files as well, the tiers to process can be selected beforehand
  • an option has been added to create a group of tiers for a new participant based on the tiers of an existing participant
  • tier attributes (color, font etc.) can now be applied to multiple tiers in one action
  • the order of the tiers in the timeline area is now reflected in the tier list of other viewers (Grid, Text etc.)
  • multiple tiers can now be deleted in one action
  • annotations can now also be split if there are only symbolically associated depending annotations
  • annotation boundaries can now be changed by using keyboard shortcuts
  • improvements to the structured multiple file search
    • the query matrix in the multiple layer search has become more flexible, rows and columns can be added and removed
    • the annotator attribute of tiers can now be used in the tier selection boxes, is visible in the info balloons and in the hit export
    • an alternative view on hits frequency has been added
    • general user interface improvements
  • new preferences have been added:
    • create annotations aligned with video frames (rounds time values for new annotations to video frame start/end)
    • snap annotations to existing annotation to avoid small gaps between annotations
    • new matching options for the suggestions panel when working with controlled vocabularies
    • switch to a more native look-and-feel on Windows

The main bug fixes are:

  • import of Transcriber files is functional again
  • in Praat TextGrid export media offsets can be taken into account
  • merging of tiers would fail
  • shifting of annotations was in some cases not reliable (negative values in hh:mm:ss.ms format)
  • tier labels in the interlinear text export would sometimes disappear
  • a bug in the Toolbox export would in some case lead to a “null” error message
  • handling of special characters in file names in the Praat scripts has been improved

December 23, 2010: Release 4.0.0

This release contains a number new features and a few bug fixes.
There is now support for external (closed) controlled vocabularies and preliminary support for interacting with the web-based lexicon tool LEXUS has been added.
A number of new tier-based operations have been implemented as well as a new player solution for Windows.

The main new features and other changes are:

  • new option to delete all annotations or annotation-values on multiple tiers. It is also possible to apply this only to annotations with a specific value
  • Empty annotations or annotations with the value of their parent annotation can be created on selected dependent tiers, in one action with only a few clicks
  • new preference setting to automatically create annotations on all dependent tiers when creating an annotation on the parent
  • Controlled Vocabulary entries can now be sorted alphabetically
  • Controlled Vocabularies files can now also be imported from .txt and .csv files
  • for Windows a new player based on DirectShow has been developed, JDS player (Java – Direct Show). The goal is to overcome a number of problems especially with playback of mpeg-a and mpeg-2 files
  • new export options for SMIL. It is now possible to export for QuickTime in addition to Real Player. Some display settings (color, font, alignment) can be specified
  • option to create a clip of a video file, based on the current selection, by using a script in which the user can specify the application to use
  • options have been added to the merge transcription function to append the annotations of the second eaf to those of the first. Either based on the media duration of the first file, on the last annotation or based on a
    user definable offset
  • preliminary support for interaction with LEXUS. A new lexicon viewer has been added allowing to look up values in an online lexicon and applying it to the selected annotation
  • support for externally defined (e.g. on the web) closed controlled vocabularies has been added. The format of these type of files, .ecv, is close to eaf
  • in the inline edit box the user can activate a suggestion list that is updated while typing. Currently only for tiers connected to a controlled vocabulary
  • Shift All annotations now supports all type of time formats as input
  • pause / silence information can also now be added to the interlinear text export file and the number of decimal digits for the pause information can be selected
  • tiers for the recognizers can be exported in AVATech tier format
  • preference setting for placing the videos in the (horizontal) center of the window
  • preference setting for the order of the viewers in the tabpane(s)
  • new menu option to show/hide (create/destroy) viewers.
  • further implementation of the Audio and Video recognizer interface. Support for both AVATecH Tier and Timeseries format, for export and import
  • in the Timeseries viewer a track can be exported as a AVATecH Timeseries file
  • showing and hiding of tiers based on their linguistic types, particpant or annotator attribute
  • time aligned annotations can be split into two annotations with the same value
  • similar to merge with next annotation, annotations can now be merged with the previous annotation
  • when saving a selection as EAF, there is now an option to clip the media of the selection and link it with the new .eaf (clipping based on the “clip script”)
  • CTRL+ALT+/ allows to move the crosshair to the center of the selection
  • CTRL+ALT+G copies the current time to the clipboard. The time format for this purpose can be set in a preference
  • option to exclude the tier names from the export as tab-delimited text file

The main bug fixes are:

  • import of csv/tab-delimited text with empty cells in the first rows would fail. This has been fixed
  • the first annotation was missing when exporting as Praat TextGrid file
  • the value of the first annotation in a file was missing in the results of a multiple file search

August 18, 2010: Release 3.9.1

This is a bug fix release intended to solve a number of problems some of which could lead to corruption of data.

The main bug fixes are:

  • export as tab-delimited text would in some cases export everything on one single line
  • import tiers could lead to corrupted files if the imported files contained “Included In” tiers
  • using “None” for the default language of a tier would make it impossible to save the file
  • processing of multiple ELAN files in a single action:
    • when deleting a tier now all descendants will be removed from the table (not only the direct children)
    • sometimes some selected tiers would not be deleted while some unselected tiers would be
    • added “None” to the list of Stereotypes for new linguistic types
  • exporting a selection of tiers to new files in a single folder failed on the Mac
  • Timeseries viewer:
    • some problems with storing and reading preferences and track configuration information have been solved
    • the time offset specified for a CSV file is now applied to all new tracks extracted from that file
  • wav clipping using Praat sometimes failed on Windows because the wrong script file was used
  • “globally” selected IMDI metadata keys are now not only applied the first time a metadata file is selected but also any next time the eaf file is loaded
  • in export as Subtitles sometimes only the first annotation was exported and if the format is .stl sometimes “|” characters were missing
  • in the structured multiple file search empty annotations as hits would not be dealt with correctly in the representation and in the hit statistics

Main other changes:

  • disabled Ctrl+Z and Ctrl+Y in the annotation edit box because these combinations trigger the main undo/redo function (instead of undoing the last text edit action)
  • in the Timeseries viewer an option has been added to display the value of the tracks at the time of the crosshair (when the media player is not playing)
  • in the merge transcriptions window buttons have been added to select or deselect all tiers
  • in the export as subtitles window the same buttons have been added as well as an option to sort the tiers
  • ELAN is a tool for annotating time based media but some users found workarounds to annotate still images as well. Annotating images has now been made a bit more convenient; JPEG and PNG files can now be selected as media files whereupon a specialized media player will be created. The context menu lets the user specify a virtual duration.
  • window dimensions read from a preference file are now checked against the (current) screen size
  • in the Toolbox export the number of prepended zeros in the record numbering now depends on the number of records in the export, with a minimum of two
  • CHAT import:
    • the media markers plus the segment times are now retrieved regardless of the presence of the “magic bullet”
    • in the case of multiple lines (annotations) within one time segment, interpolated time values are calculated instead of creating unaligned annotations on a top-level tier

May 07, 2010: Release 3.9.0

This release contains a number of bug fixes and adds new features as user preferences enhancements, color coding in combination with controlled vocabularies, a new version of the Recognizer API and new multiple file operations (some of which have been contributed by the Sign Language research group at the Radboud University in Nijmegen).

The main new features and other changes are:

  • new options for processing multiple ELAN files in a single action:
    • add or change tiers and linguistic types and edit their attributes
    • export a selection of tiers to new eaf files
    • remove superfluous tab, new line and whitespace characters (scrub)
  • a controlled vocabulary entry can now have a color and a keyboard key assigned to it. (These assignments are stored in the preferences files!)
  • in the Segmentation mode annotations with content can be created using the keystrokes of the controlled vocabulary entries
  • most viewers now use the colors of controlled vocabulary entries in their visualization of annotations. In the Timeline viewer annotations on a tier with a single symbolically associated child tier, will “inherit” the color of their child annotations.
  • the Timeline viewer has a new option to display tiers with reduced tier height. This way more tiers fit in the display area.
  • in the Timeline viewer a little dot marker in the right lower corner now indicates that an annotation value has been cut off
  • when saving a template a preferences file is created along with it. New documents based on the template will be have the preferences settings applied to them.
  • the color chooser used for tier and cv entry colors has been extended with a panel for (persistent) favorite colors
  • the width of the tier label panel left of the timeline viewer can now be changed by dragging a separator with the mouse
  • new preferences for suppressing “save messages” when only the media file paths have changed and one for setting the number of backup files to use. This has always been a single file and can now be set from one to five files.
  • the export as subtitle text now supports the .stl file format. The time format can be PAL, NTSC drop frame or NTSC non drop frame.
  • in the export as tab delimited text the columns for tier name and participants can now be omitted
  • the API for pattern recognition components has been extended. Video recognizers are supported and complete tiers can be passed as input for the recognizer.
    There is preliminary support for standalone components as extensions, based on a component metadata description.
  • shift-click in the Timeline viewer when there is no selection, creates a selection from the current media time to the clicked time.
  • annotations from gaps can now operate on multiple tiers
  • Select All and Select None buttons have been added to most export options
  • “None” has been added as an option (the default) for the language attribute of a tier. This attribute is used for input methods.
  • more warning messages are generated when creating an annotation fails (e.g. on a symbolic association tier when there is already an annotation, on time subdivision and included in if the parent is not fully aligned)
  • the width of the controlled vocabulary popup now scales with the length of the longest entry

Main bug fixes are:

  • Toolbox export: the last annotation would sometimes disappear, this is now fixed
  • multiple files search: in relevant overlap constraints time comparisons are now done excluding the end or begin time
    (i.e. overlaps of 1 ms, endtime == begintime on the other tier, are no longer a hits)
  • Timeseries viewer: when there is no waveform viewer the tracks defined would be lost; fixed. The track label is now painted left of track area.
  • import of Praat files: ISO control characters are filtered out of the annotation texts and replaced by whitespaces
  • in multiple file search-and-replace an error no longer occurs when one of the tiers is not in a one of the files
  • creating an audio clip with Praat on Windows sometimes didn’t work because of a call to the wrong script

November 18, 2009: Release 3.8.1

This release is intended to fix a few bugs that can lead to data loss. At the same time, some new functionality that has been implemented since the previous release is included as well.

The main bug fixes are:

  • shifting annotations on the active tier or on all tiers, left or right of the crosshair, could lead to the loss of time information of annotations on depending tiers. As a result such file could no longer be opened after the changes had been saved.
  • editing adjacent annotations in a tier hierarchy containing at least one time-subdivision tier with unaligned annotations, could lead to disappearing neighbouring annotations. In some cases this would occur after an “undo”.
  • merging annotations with unaligned time-subdivision depending annotations could lead to an incorrect order of these unaligned annotations.
  • the newly introduced function of storing named multiple file domains, appeared to store only the last used domain. This is now fixed.
  • the import of FLEx files did not handle irregularities in type and order of the different FLEx elements correctly, leading to overlapping annotations. This import has been improved and is now more robust.
  • fixed a problem with long annotations starting with a whitespace character in the timeline view. Activating such annotation could lead to hang the aplication.
  • export as QT Text now ensures a gap of 10 ms between the subtitles in order to make them pop up correctly when played by QuickTime. Square brackets are now replaced by parentheses.
  • fixed the Regular Annotations dialog; the tiers are now in a scrollable list.
  • printing of shortcuts is re-activated again.

The few new features and other changes are:

  • an option to create annotations based on the gaps between existing annotations. The new annotations can be inserted on the same tier or created on second tier.
  • pasting a group of annotation (annotation plus depending annotations) on an other tier than the ones they were copied from, now tries to paste the depending annotations as well. This will only work if the target group of tiers has a similar structure and naming convention (e.g. from ref@AA and tx@AA to ref@BB and tx@BB).
  • an option to merge two tier groups onto a new target tier group. The mapping of original tier names to target tiernames is based on the pattern in the tier names (see previous item). In contrast to the existing “Merge Tiers” function, here the annotations themselves are not merged
  • it is now possible to change the case (uppercase/lowercase) of all annotations of selected tiers.
  • the aspect ratio of the video image can now be set by the user (in case the media framework interpretes the encoded aspect ratio incorrectly).
  • the author field of the document can be set once again.
  • a preliminary function to compare the segmentation on two tiers, e.g. created by different annotators, has been added.
    The comparison is based on the amount of overlap of annotations on the two tiers.
  • the multiple file chooser now supports selection (and deletion and moving) of multiple rows in the list on the right (the selected files).
  • the (unstructured) multiple file search now supports loading files from a named domain as well
  • the limit of 5000 hits in the structured multiple file search has been raised to 100000.

August 26, 2009: Release 3.8.0

This release adds new customization options, like user definable keyboard shortcuts, improves and extends import and export options and takes a first step in combining metadata and content search.
The Cocoa-QuickTime player (Mac OS X) is now 64-bits compatible, meaning that it is ready for Java 1.6 and Snow
Leopard.

The main new features are:

  • the keyboard shortcuts are now user definable (Edit->Preferences->Shortcuts), the existing doubly
    defined shortcuts have been removed
  • added options to delete all annotations, on all tiers, left or right of the crosshair
  • randomly select multiple annotations in the Timeline viewer (by alt-clicking) and delete them in a single action
  • a tree representation has been added to the IMDI metadata viewer
  • load the exported result of an IMDI metadata search (on the local file system) and execute content search queries in all eaf files listed in the IMDI sessions
  • it is now possible to save and load domains of multiple files, i.e. sets of files (and folders) for multiple file search or export
  • menu’s have been added to jump to ELAN’s webpages and to the mailing list subscription page
  • a new Annotations tab has been added to the annotation statistics panel, one which is more consistent with the layout of the other tabs
  • new and improved import and export functions:
    • an import function for FieldWorks’ FLEx format has been added
    • an export option has been added to the multiple file search allowing to save per file hit statistics
    • it is now possible to customize the output when saving (concordance multiple file) search results; columns can be selected and their order can be specified
    • the wordlist export window now allows selection of tiers based on their linguistic type
    • Praat TextGrid files can now be imported without having an annotation document opened; a new document is created in that case
  • new options in the Edit Preferences window:
    • switch between available media player frameworks, for Mac OS and Windows
    • set the number of subtitle viewers to appear in the Subtitles tab
  • video frame images can now also be saved in .bmp format (in addition to jpg and png)

Main bug fixes and other changes:

  • the DCR connection has been restored, it now connects to the ISOcat data category registry (www.isocat.org)
  • new versions of M2-edit are now also supported for clipping media (mpeg) files (uses the “M2-xcode-cl” command)
  • Toolbox import: fixed a problem with multiple mediadescriptors
  • the undo function of “merge annotations” was depending on the time change propagation mode, this is fixed
  • some fixes in buldozer mode when annotations are moved to the left:
    • when an annotation is moved beyond the begin time of the previous annotation, resulting annotations would overlap
    • when an annotation is shifted completely beyond the zero point, undo wouldn’t recreate it
  • synchronisation of media and synchronized playback using the Cocoa player (Mac OS) has been improved
  • deletion of annotations on the active tier and within the selection would sometimes fail for unaligned annotations on time alignable tiers
  • some fixes in the structured multiple file search: vertical constraints with time values (e.g. begin 1 – begin 2 = x ms) would not be interpreted correctly
  • the menubar now contains all items from the start (the empty window); functionalities that are independent of a particular annotation document, are now all available before loading or creating a document
  • color shifts in the scaled images of a filmstrip export on the Mac OS have been fixed
  • saving a video frame as a .jpg on Mac OS no longer results in a black image
  • the “play around selection” toggle value and the “playback rate and volume” toggle values are now application wide preference settings
  • removed writing and reading of empty EXTERNAL_REF elements

June 18, 2009: Release 3.7.2-1 for Mac OS X

This Mac-only release attempts to solve a number of problems, mainly in the area of media playback. Depending on the combination of Mac OS version, QuickTime version and the hardware (PowerPC or Intel) the following errors could occur:

  • 1) crash of ELAN when opening a (transcription with a) media file
  • 2) crash of ELAN when closing a window containing a media file
  • 3) synchronisation of multiple video’s not working correctly
  • 4) crash when detaching/attaching a video from/to the main window
  • 5) crash or freeze when changing the linked media files
  • 6) freeze of ELAN when saving a video frame as an image
  • 7) freeze of the video image once the playhead has crossed the (roughly) “60% of media duration” boundary
  • 8) slow updating of the screen once the end-of-media is in the visible area of the timeline viewers

The problems 1 through 3 are related to our Cocoa-QuickTime based media player solution. We have tried to fix these problems and have tested the new player on several combinations of Mac OS version and processor. But, we don’t have all possible combinations around, so please report if you still experience crashes.
Problem 4 and 5 could occur both with the Cocoa-QuickTime and the QuickTime for Java player, 4 can still happen with QT for Java players. Problems 6 and 7 are specific to the QT for Java player, where 7 only seems to occur with QuickTime version 7.6. or higher. The last problem was independent of the media player in place.

The problems in the QT for Java player have not been solved, these problems seem out of our control. Therefore the default player is again the Cocoa-QT player. If needed, one can still switch to the QT for Java player as described here:
http://www.lat-mpi.eu/tools/elan/elanforum/simpleforumpost.2009-04-07.5053039867

April 01, 2009: Release 3.7.2

This is an intermediate release that primarily targets at overcoming a number of media playback problems that have been reported by Mac users. Most of these problems seem to be caused by QuickTime for Java.
This is the first release that includes our own Cocoa-QuickTime based media solution.
Apart from that some new features have been added.

The main new features are:

  • a Cocoa-QuickTime based media player that solves a number of problems:
    • attaching and detaching a player from the main window should be save again
    • changing the linked media files should be save again as well
    • extracting a video frame image should no longer freeze the application
    • the 7.6 update of QuickTime caused freezing of the video image somewhere in the second half of the video on some Macs. This player should not suffer from this.
    • ready for display of 2D graphics on top of the video

    The QuickTime for Java based player is still available as an alternative.

  • on the Mac it is now also possible to associate ELAN with the .eaf file type (via the Get Info option in the Finder). An eaf file can also be opened by dropping it on the ELAN icon in the Dock
  • also for Mac users: in the Preferences window it is now possible to specify to use the main menu bar instead of having the menubar in the ELAN window
  • other new preferences: specify a default folder for ELAN preferences files, change the number of items in the Recent Files list, disable tooltips and display all video’s in one row and with equal sizes
  • new tabs have been added to the Annotation Statistics window: overviews are now available per linguistic type, participant and annotator. Besides that, a few options have been added to the Annotations tab to harmonize its behaviour with the other tabs
  • a Filmstrip export has been added: save snapshots of video frames alongside the corresponding waveform segment as an image file
  • new options for tier selection in the Annotation Density viewer: select individual tiers or groups of tiers based on linguistic type, participant or annotator
  • added support for Praat PitchTier and IntensityTier file types, for visualization by the Timeseries viewer. The files can be added in the Linked Secondary files tab of the Linked Files window.
  • delete multiple annotations, on the active tier, in a single action. Either the annotations within the selected time interval or all annotations left or right of the crosshair.
  • a highlight color for a tier can be specified in the More Options window. The color is used as background color in the Timeline view.
  • added milliseconds as format to the time label below the video. Pal and NTSC time formats have been added to the Grid viewer.
  • a keyboard shortcut has been added to position the time selection in the center of the Timeline viewer: Shift + A

Main bug fixes and other changes:

  • Toolbox import: the name of the marker would appear as the value of the annotation in case of an empty marker line. This is now fixed.
  • Toolbox import with a marker file: if a participant marker was defined as a subdivision type, it wouldn’t be recognized as a participant marker.
  • fixed wrong annotation order after merging two annotations with symbolic subdivision depending tiers that are not a direct child of the root tier
  • fixed a problem with Paste Annotation Group, where Controlled Vocabularies are involved in combination with the “auto
    commit” setting. A wrong value would sometimes be pasted.
  • the constraint types Within and Surrounding in the multiple layer tab of the multiple file search appeared to be non-selectable. This is now fixed.

February 09, 2009: Release 3.7.1

This is a bug fix release.

This release fixes a bug in the Toolbox import which has been introduced in version 3.7.0, alongside the existing Shoebox import. When using an ELAN marker file (.mkr) to specify the tier setup, the option for specifying which marker is the participant marker (a marker containing a participant’s name or code) was not functional. As a result only tiers with suffix “@unknown” were created.
Apart from that, the last used marker file is now automatically loaded for the next import and the last media file type selection (in “New..” and “Linked Files”) is also stored as a user preference.

January 29, 2009: Release 3.7.0

This release adds a number of new features and fixes a few bugs.

The main new features are:

  • a viewer for metadata that are relevant to the annotation document has been added to the tabpane. The metadata are extracted from an external source, which can be selected by the user. There is built in support for Imdi session files. A configuration panel allows selection of metadata keys to be displayed in the viewer.
  • find-and-replace in multiple eaf files. Allows selection of files and folders and the query can be executed in all tiers or in a selection of tiers. The search query can be case sensitive and regular expression may be used.
    Each file changed by this find-and-replace will be overwritten: there is no undo! It is thereforeimportant to create backup copies before using this feature.
  • shifting annotations: in addition to the existing “Shift All Annotations”, it is now also possible to shift:
    • a single annotation, with depending annotations
    • annotations on the active tier or on all tiers, left or right of the crosshair
    • annotations on the active tier in the selected time interval

    The number of milleseconds, positive or negative, can be specified. Existing annotations can limit the amount of ms: it is not possible to shift annotations on top of or beyond other annotations.

  • options have been added to the column-per-tier variant of the tab-delimited text export, to reduce the number of empty cells. Values of annotations that span other annotations (e.g. an annotation with depending subdivision annotations) can now be repeated or replicated in the rows containing the spanned annotations. The row of the spanning annotation is then removed.
    This repetition of values can be restricted to annotations in the same hierarchy or be applied to annotations on all tiers. One should be aware that repeating values effects statistics.
  • tab-delimited text export from multiple files now also features the column-per-tier option.
  • the “annotations from overlaps” and “merge tiers” features now have the option to only process annotations with the same value, or a specific value
  • the Timeseries viewer can now be attached to the main window, sharing the space with the waveform viewer
  • the time ruler in any timescale based viewer can be hidden
  • the selected text in the Text viewer can now be copied to the clipboard
  • multiple file search: the font and font size used for the search results (hits) can now be set by the user
  • a frequency count has been added as an option to the export of words/annotations
  • new global preferences have been added:
    • specify a default location to find media files
    • an option to clear the selection after creating an annotation
    • specify a default set of Imdi metadata keys
  • a new Toolbox import function has been added (as an alternative to the Shoebox import). This assumes unicode encoding and tries to prevent cutting off words in case of misalignment. Non-spacing characters are taken into account. This import offers more flexibility in terms of tier dependencies.
  • Praat import: an option has been added to skip empty intervals
  • Japanese has been added as a language for the user interface

Main bug fixes and other changes:

  • after playing in selection mode and stopping, the selection is always updated to match the crosshair
  • Praat TextGrid export: a last empty interval is added in case the last annotation ends before the media end
  • Praat TextGrid export: escaping of quotation marks in annotation text is added
  • Praat TextGrid import: selecting “Default” as the file encoding did not result in the default for Mac and Windows
  • the edit box in the Timeline viewer is now always visible, even if part of the selection is outside the view area
  • SRT export: prevent export of subtitles with the same begin and end time
  • fixed improper order of symbolic subdivision annotations when merging two annotations
  • Alt+C shortcut a an anternative to Alt+Shift+C
  • shortcut for clearing the selection and selection mode at once: Ctrl+Shift+Z
  • shortcuts for jumping to next/previous annotation and start editing at once: Ctrl+Alt+Right (Left)
  • ELAN now requires Java 1.5 or higher, therefore Mac OS 10.3 is no longer supported

August 14, 2008: Release 3.6.0

This is a release that adds some interesting new features and fixes a few bugs.

The main new features are:

  • introduction of an extensible Audio Recognizer framework. This is an integrated, pattern recognition based, semi-automatic segmentation and annotation facility. The system offers the means to select a recognizer, interact with it and subsequently convert the segmentations produced by the recognizer to tiers and annotations.
    An Audio Recognizer tab has been added that contains the list of recognizers and that provides the controls for the interaction with the selected recognizer. The list initially contains only one recognizer,the Silence Recognizer. It creates segments representing either silence or non-silence. It operates on a .wav file and it requires that the user provides it with one or more segments (selections) containing “silence”. A minimal duration can be specified for both silence and non-silence fragments. It is fairly fast and especially effective when operating on relatively “clean” audio files.
    The Waveform viewer visualizes the segments and their labels produced by the recognizer. The conversion of segmentations to tiers is customizable; new segment labels can be specified and there is an option to number the segments.
    An API has been designed so that developers can implement their own recognizer and add it as an extension to ELAN. More information can be found on the sources release notes page. The intention is to add a similar framework for video recognizers in the future.
  • merge an annotation with the next annotation on the same tier; when there are depending annotations these will be merged as well
  • merge two tiers into a third tier; overlapping annotations are merged, with a few options for handling the content
  • a ticker mode has been added to the timeline viewer; when the crosshair reaches the center of the viewer, the viewer scrolls to the left while the crosshair stays in the center
  • a zoom to selection option has been added to the Timeline and Waveform viewers
  • zooming in and out by using the mouse scrollwheel in combination with the Control (Ctrl) key
  • added an extra column to the tiers statistics table containing the total annotation duration as a percentage of the (master) media duration

Main bug fixes and other changes:

  • export to SubRip .srt: the index of the first subtitle is now 1 instead of 0
  • import of csv/tab-delimited text: if a separator was detected it was shown in the ui but not applied to the actual import function, unless the user explicitly confirmed that separator. The detected separator is now used by default.
  • logging on Mac OS: logging is now done to the new preferences dir
  • on Windows with audio as the “master” media, the media playhead would jump to the end of file after playing a selection. Fixed.
  • adding an annotation on a time subdivision of a time subdivision tier where the target parent annotation is unaligned would sometimes create an annotation under the wrong parent. This is also fixed.
  • creating a time aligned annotation as a child of an annotation that is (partially) unaligned is now prevented as much as possible
  • using the Label and Number function without the number part now works correctly (instead of producing a NullPointerException and application crash)
  • Toolbox import: fixed the bug introduced in 3.5 that led to duplication of annotation values on symbolic association markers/tiers directly under the record marker/tier
  • text elements in the ELAN user interface are now anti-aliased by default

May 19, 2008: Release 3.5.0

This is an intermediate release that adds preliminary client support for ISO Data Categories and that also fixes a few bugs.

The ISO Data Category Registry (DCR) is one of the ISO initiatives towards the establishment of standards for Language Resources and towards improvement of interoperability in Language Resources. The DCR specifies the names and definitions for data categories relevant to the language resource domain. The categories are elementary descriptors of linguistic concepts (e.g. “part of speech”, “grammatical number”). The registry is accessible online (at www.isocat.org) and provides web services for tools to interact with it. It is still in development.

The main new features are:

  • support for Data Categories, references to concepts defined in the ISO DCR can be made from:
    • individual annotations
    • entries in a Controlled Vocabulary; when the value of an entry is applied to an annotation, its concept reference is also applied to the annotation
    • Linguistic Types; thus a tier can refer to a data category (e.g. partOfSpeech) through the linguistic type it is associated with
  • a new way to conveniently create translations for the user interface. Translators can now see their translations in the user interface without having to wait until these are included in a new version of ELAN. Also, research groups that involve native speakers in the annotation process and only need to translate a limited part of the interface, don’t have to provide a full translation anymore. There are now 2 files in a folder ‘locale’ in the ELAN install directory, containing key-value pairs for the interface elements. When copied to ELAN’s “data” folder these files are used when Options -> Language -> Custom is selected in ELAN.
  • Chat export: the language of a participant can now be included in the output as a “%lan:” line
  • export of a media clip with m2-edit now creates a clip of all associated video files

Main bug fixes and other changes:

  • the location of ELAN’s “data” folder on the Mac has been changed to <user_home>/Library/Preferences/ELAN (on Windows and Linux this is still <user_home>/.elan_data)
  • added the shortcut key combination Ctrl+Shift+K as an alternative to CTRL+/ for changing the active boundary of a selection
  • copying/reparenting tiers: when the type of the tier that is copied changes from a subdivision type to symbolic association, the values of sibling annotations are now combined into one annotation. In the same situation the new parent of any annotation is found based on its current (temporal) position.
  • single file search: when a query is loaded from file and consists of more than one constraint panel, there are always zero hits. This is fixed.
  • a problem with some uppercase escaped characters in (multiple file) regular expression search has been solved
  • Chat export: the languages entered in the main tiers’ Language input boxes are now added to the @Languages line
  • export of media clips with m2-edit now correctly handles files on network shares
  • Toolbox import with only symbolic association tiers is now also handled correctly for depending tiers with ‘level’ > 1
  • Praat import: improved handling of time segments with a duration of less than 1 ms, handling of empty lines and duplicate tier names
  • SMIL export: fixed the “FileNotFoundException” that occurs on Windows systems
  • a problem in the Waveform viewer with some files containing cue point markers has been solved

March 04, 2008: Release 3.4.0

This is a minor release that fixes a few bugs and adds a number of new features and new customization options.

The main new features are:

  • edit preferences; initial version of a user interface allowing to change some (application wide) user settings. It is now possible to let changes made in the text edit box be committed when the edit box is (unintentionally) deselected. It is now also an option to commit changes in the edit box with only the Enter key (without the Ctrl key). There is also an option to change the frame forward/backward behaviour to always jump to the begin of the frame.
  • timeseries data stored in csv or tab-delimited text files can now be handled by the Timeseries Viewer. After adding the file as a secondary linked file (Edit -> Linked Files) the user can configure tracks in a configuration dialog. For each track the column has to be specified from which to take the data.
  • the Segmentation viewer or mode has been extended with new options. Annotations with a, definable, fixed duration can be created with a single key stroke. whereby the stroke can identify either the begin or the end of the annotation. The media navigation keyboard shortcuts are now also active in segmentation mode.
  • the Annotation Density viewer has now a right mouse button popup menu to select a single tier of which the distribution of annotations should be shown. By default the annotations of all tiers are shown.
  • the Statistics window has been extended with a panel showing some information per tier, such as number of annotations,
    min., max. and average duration etc.
  • a left mouse button click on a video image copies coordinates to the clipboard in one of the following formats:
    • no modifier key: x,y [orig.width,orig.height] in pixels
      where the width and height are the sizes returned by the media framework
    • with Alt key: x,y
      where x and y are values between 0 and 1, identifying a relative location within the image dimensions
    • with Shift key: x,y
      where x and y are pixel values in the original coordinate system

    In all 3 cases, especially the first and last, the result depends on dimension and aspect ratio returned by the media framework

  • support for page numbers has been added to interlinear printing
  • all videos associated with a document are now listed in a menu under View -> Media Player. They can individually be
    activated/deactivated through this menu, without having to (permanently) remove them entirely from the document via the Linked Files menu. Playback performance can be a reason to temporarily remove a video.
  • the manual is now included in the release and accessible through Help -> Help Contents in ELAN. This in addition to the pdf and online html version.

Main bug fixes and other changes:

  • a bug in the import of csv/tab-delimited text that prevented import of files without delimiter (1 value per row) has been fixed
  • the csv/tab-delimited text importer now treats files as utf-8
  • the remove all tracks option from a trackpanel in the Timeseries viewer now removes all tracks, instead of half of the tracks
  • the text edit box in most viewers now uses the font and font size as specified for the tier it concerns
  • the media player rate is no longer changed while dragging the slider; only when the slider is released the rate is changed

December 13, 2007: Release 3.3.0

This release fixes a few bugs in the structured search in multiple files and adds a number of new features, mainly in the area of import and export.

The main new features are:

  • export multiple .eaf files to one output file, as:
    • tab-delimited text. A column is added for the file name(s).
    • a word list. The output options are the same as in the single file export.
    • a list of annotations. All unique, unparsed annotation values are exported in alphabetical order.

    In all 3 cases it is possible to select a set of tiers to include in the export.

  • export as Toolbox file (utf-8). This is a new, Unicode only implementation of the export to Toolbox format. This export adds new customization options such as line wrapping, insertion of empty lines, use of a custom record marker and the possibility of specifying whether the label of markers without content should be in each record or not.
    The "old" export to Shoebox remains available.
  • Subtitle text export in SubRip .srt format. With options to select tiers and to specify the (preferred) minimal duration of a subtitle.
  • import of annotations from a .csv (comma separated values) or tab-delimited text file. Rather than limiting this import to a fixed structure of the source file, the import window allows the user to specify which columns to use for the import and what kind of values each column holds: e.g. begin time, end time, annotations or tier names. At least one column should hold annotation values. If there is no timing information, a default duration can be specified.
  • in addition to the absolute path of media files, a relative path (relative to the annotation file) is now also stored.
  • in the tier label area of the Timeline and Interlinear viewer, the number of annotations on each tier can be displayed.

Main bug fixes and other changes:

  • the bugs in the structured multiple file search that are described here have been fixed.
  • the Praat import and export now also support files in utf-8 and utf-16 encoding.
  • the export to QuickTime text files has now better support for Unicode. New output options like merging of tiers into one file and specifying a minimal duration per subtitle, have been added.
  • single file search: regular expression search in tiers that are associated with a Controlled Vocabulary is now also possible (in addition to search for one of the entries of the CV).
  • when editing an existing annotation, leading and trailing spaces, tabs and new line characters are no longer removed from it’s value the moment it is placed in the edit box.

October 04, 2007: Release 3.2.0

The main changes in this release are a considerable extension of the structured search in multiple files and the restructuring of the user preferences mechanism.
The multiple files search system is essentially the same as the web-based search in ANNEX. In the standalone environment of ELAN there can be performance issues when executing complicated queries. Although thoroughly tested we advise to be alert when interpreting results; feedback on strange or unexpected results is much appreciated.
It is advised to read the manual section about the new searching options.
User preferences are now stored in an XML file and a window configuration can be exported and imported into (applied to)another document window. The new structure of the mechanism allows for future extended customization options.

The main new features are:

  • the structured multiple file search window is now a window with tabs, each tab representing a different level of complexity. In all tabs the search results can be shown in concordance or frequency view. The two new tabs that have been added are:
    • Single layer search: a search based on patterns or structure within a tier (n-gram), with options for case
      sensitivity and type of matching (substring, exact, regular expression). It maintains a query history.
    • Multi layer search: powerful (and rather complicated) search based on structure within and between tiers. With options for case sensitivity, type of matching and timing constraints (begin time, end time, duration). It has also a query history.
  • the user preferences are stored in an xml file (.pfsx). A configuration can be saved and loaded. There are two new user preferences on the tier level:
    • define a preferred color for the tier label (lower left area of the window) instead of a random color
    • select a preferred font. Note that the contents of annotations remains font independent (Unicode), and that the font name is not included in the .eaf, only in the preferences. A Font Browser utility has been added to assist in finding a suitable font for a given Unicode block.
  • the size of the area available for the video(s) (and the tabbed viewers) can be changed by dragging a resize icon on the right of the window
  • annotations on the active tier can now be created with two keystrokes (independent of a selection): place the playhead at the begintime of a new annotation and press Shift+Enter, move the playhead to the end of the new annotation and press Shift+Enter again
  • label and index: an option has been added to (re)create the contents of annotations of one or more tiers with a label and a (increasing) number (e.g. a Toolbox style sequence like ref 001, ref 002 etc.)
  • export as tab-delimited text: an option has been added to export with a separate column per tier (tier name is the column name); annotations with the same begin and end time appear in the same row
  • export as traditional transcript: an option has been added to include a pause or silence indication between two successive annotations. Formatted as “(ss.ms)”.
  • import Praat TextGrid now also supports PointTiers; the duration of the new annotations can be specified
  • export to Praat TextGrid format is now also possible. The tiers to export can be selected.
  • export an alphabetically ordered list of (unique) words, present in a selected set of tiers

A number of features have been contributed by Ouriel Grynzspan, for which we thank him very much:

  • remove the annotation value of the active annotation
  • option to create regular annotations: create adjacent annotations with a specified duration on a selection of tiers
  • a shortcut to copy the value of an annotation to the next adjacent annotation (if the next annotation has the same duration and is empty)
  • annotation spreadsheet: a table with all annotations of the top-level tiers
  • annotation statistics: a table with a per tier overview of annotation values and their frequency, average duration,
    latency etc. Consult the manual for an explanation and the rules applied

Bug fixes and other changes:

  • switching selection mode on doesn’t change the selection anymore (in some situations the selection would be changed to the interval 0 – current media time)
  • export as Quick Time text: if more than one tier is selected multiple text files are created, one for each tier
  • edit tier dialog: after switching to the Add tab (from the Change or Delete tab) the list of potential parent tiers would not be complete, this is now fixed
  • an error that prevented the use of .wmv files on Windows has been solved
  • annotation ID’s are now maintained over successive save actions. Once an annotation has an id it will always keep that id and an id, once used, will never be reused for an other annotation
  • the detached edit window can now also be closed with Ctrl+W
  • closing the last empty window (e.g. with the windows close button or File -> Close) now quits the application
  • on MacOS it should now be possible to open or clip a .wav file with space characters in it’s name in Praat
  • on Windows with QuickTime 7.2: the installer of ELAN will attempt to create a policy file that grants sufficient permissions for QuickTime to be accessed from ELAN

May 07, 2007: Release 3.1.0

This release is primarily intended to fix some bugs and to overcome some annoying phenomena on MacOS and Windows XP. On MacOS video images initially appeared outside of the ELAN window, in the topleft corner of the screen, leaving a polluted main menu bar. On Windows XP refreshing of the screen in the TimeLine viewer area could be very unpredictable, to the extent that all annotations could seem to have disappeared.
But a few new features have been added too.

The main new features are:

  • import of Praat .TextGrid files. Only IntervalTiers are supported in this version. In contrast with the other import functions, the tiers from the Praat file are added to the current open document. Praat’s special symbols are converted to their Unicode equivalents.
  • a horizontal scrollbar has been added to the TimeLine viewer. As an option; it can be removed from the view. Scrolling by means of alt-dragging the ruler is still functional.
  • the tiers in the TimeLine viewer can now be sorted by their Annotator value
  • automatic backup interval of 1 minute has been added
  • most dialogs can now be closed by pressing the Escape key or the Ctrl-W combination

The main bug fixes (apart from the ones mentioned above) are:

  • after playing a selection (to the end of the selection) in selection mode, changing the selection through Ctrl-Right etc. did not work.
  • in the edit tier dialog it was impossible to change the tier’s default language to the system’s default language.
  • the order of tiers in a template file was lost when importing tiers in the import tiers dialog. The order is now maintained as much as possible.
  • loading of .eaf files with accented letters in the file path would fail (result of the transition to a newer version of the parser library).
  • there has been a problem with the user interface of the structured multiple file search window after changing the search domain. This is now fixed. The tab-delimited export of the search results has been improved.
  • Shoebox import: wrapped lines are now all parsed with the same encoding (the encoding of the first line, the one immediately after the "\markername")
  • Shoebox files with only one marker as part of the interlinearisation were not processed correctly; the words of this one marker would be added to a single annotation, padded with extra spaces. Now separate subdivision annotations are created.
  • on Windows ELAN uses QuickTime by default for .mov and .qt files when a new transcription is created. But when an existing eaf file with such media was opened it failed to do so.
  • the Windows installer should now be functional on Vista too and the Linux version should now install on Ubuntu without problems

February 07, 2007: Release 3.0

This release of ELAN brings a number of important changes.

New logo, new webpage
First of all, a new logo has been introduced. ELAN is part of a set of tools that have now been brought together on MPI’s new LAT (Language Archiving Technology) website. This will be the place for ELAN news, downloads, documentation and a user forum.

New menu layout
The structure of the menu bar has been changed significantly. Items from the old Edit menu have been promoted to the main menu. Operations concerning Annotations, Tiers and Linguistic Types can now be found under their own item in the menu bar.

New close and exit behavior
Until now the File -> Exit menu only closed the active window. From this version on Exit closes all open windows and quits the application. A new File -> Close menu item only closes the active window. In relation to these changes a new Window menu has been added to the menu bar, listing all open windows. A window can be activated from the menu and there are keyboard shortcuts to activate the next or previous window in the list (Shift+DOWN, Shift+UP).

The other main new features are

  • a new attribute, ‘Annotator’, can be added to Tiers
  • options to copy and paste an annotation or an annotation group (annotation + depending annotations), through the menu or keyboard shortcuts. There is also an option to create a duplicate of an annotation on another tier
  • TimeLine viewer
    • visual demarcation of tiers by means of a slightly different background color
    • the selection is now painted over the whole height of the viewer, regardless of the number of tiers
  • Waveform viewer
    • vertical zooming (or visual amplification) has been added
    • a script has been integrated to let Praat create a clipped wave file based on the current selection
      in ELAN. The new file is saved in the same directory as the original wave file and the selection begin and end time are part of the file name. The new file is then opened in Praat. Accessible through the popup menu.
  • Single file search
    • the query construction panel has been changed such that the query will be human readable as much as possible. Also the structure of the constraint panels is now visualized as a tree.
    • submatches (annotations on other involved tiers than the first mentioned tier) are now part of the results. They can be seen in the ‘tooltip’ when the mouse pointer is moved over the result table. Also, when exporting through Query -> Export, the option to save as HTML has been added, either in a table or as a tree view. This last option includes the submatches in the output.
    • to the result table popup menu an item has been added to export the results, with submatches and some context, to an .eaf file. Context here meaning the group of annotations the hit is part of (if any).
  • a new multiple file search option is now available: structured search in multiple eaf files. This is a preliminary implementation of structured search (as opposed to the existing ‘simple text’ multiple file search). Currently the search is limited to structure in annotations within a tier. This feature will be extended in coming releases in the query construction part as well as in the results representation part.
  • export a transcription as HTML, interlinear style. For each tier a separate CSS style block is added, making it relatively easy to customize the looks of the output.
  • export as Tiger xml; the result can be opened with the syntax editor and viewer Synpathy
  • support for rtsp streaming media (QuickTime based) has been added (contributed by the SIDGrid team)
  • a ‘Recent Files’ list has been added to the File menu
  • a ‘wizard’ has been implemented to create annotations based on the overlaps of annotations on two different tiers. The duration of the overlap can be added ad the value of the new annotation. Useful for extraction of co-occurring phenomena.
  • a lot of new shortcuts have been added. Also the keyboard access to the menu’s has been re-implemented.
  • TimeSeries viewer: support for non-continuous tracks is added
  • Printing: the show empty slots option has been implemented
  • Interlinear text export: an option has been added to insert an additional tab between annotations

The main other changes are:

  • the shortcuts table has been re-grouped and can now be printed
  • some shortcuts have been changed. The 3 remaining alt+character combinations now have an alternative ctrl+alt+character with the same function (to avoid the ‘input window’ on Mac OS 10.4 systems)
  • loop mode isn’t switched off anymore when play selection is stopped. It has to be switched off explicitely.
  • Shoebox/Toolbox import and export: when using a .typ file an option has been added to specify that all markers are in Unicode
  • Shoebox/Toolbox export: auto generated marker files are now stored in the same directory as the exported .txt file
  • a new transcription created without a template will now start with one default linguistic type and one default tier
  • the import of Controlled Vocabularies is functional again, and an option to merge two CV’s with the same name has been added. New entries can be added by hitting the return or enter key.
  • the first item in the languages list is always the system’s default language, indicated by the additional label "System default", English has been added to the list of languages
  • the Linux distribution has been extended with a library to add support for e.g. mp4 files

The main bug-fixes are:

  • the automatic backup with a new file based on a template will now popup a warning message to save the file first
  • on Mac OS (read: QuickTime based players) the playhead sometimes couldn’t be moved beyond the boundaries of the selection. This is fixed.
  • Shoebox/Toolbox export: on export with a marker file in combination with a tiergroup per participant not all markers would receive the right encoding. This is fixed now.

Known issues:

  • on Mac OS 10.3.x systems the application sometimes crashes when a window is being closed.
    It is advised to upgrade to 10.4, also considering the fact that gradually the move to Java 1.5 (not available on Mac OS 10.3) will be made in the future.
  • on any Mac: grabbing a frame from a video can cause the application to crash or freeze
  • the installer for Mac OS 10.3.x should be functional again

August 07, 2006: Release 2.6.3

This is a release primarily intended to fix a number of bugs that could lead to inconsistencies in the structure of (depending) annotations.

The main bug fixes and improvements:

  • several changes have been made in the way existing annotations are updated after adding a new annotation or modifying an annotation’s begin and/or end time. The problem mainly occurred on depending Time Subdivision tiers: in some situations changes were not properly propagated to depending tiers.
  • Shoebox/Toolbox import:
    • the last record/annotation in a Shoebox/Toolbox file was not properly aligned in the case of files containing no time information at all. The last annotation would overlap other annotations in such cases.
    • files starting with a Byte Order Mark will now be processed like other files. The BOM will be ignored.
    • the parsing process has been made a bit more flexible; lines not starting with a "\" immediately followed by a marker name will now be appended to the previous line.
  • Shoebox/Toolbox export:
    • in the Shoebox/Toolbox export dialogue an additional textfield has been added to specify the "database type name". This name will be used in the case of export with "field markers" or with "ELAN tier names".
  • some changes have been made in the "play selection" mechanism on the Mac (i.e. QuickTime based player). This solves a number of problems that users have experienced in selection playback.

Known issues:

  • Adding an annotation or changing an annotation’s boundaries can still lead to inconsistencies in the structure of depending annotations, in the case of a mixture of aligned and unaligned annotations on Time Subdivision tiers. Especially when such mixture is used on two or more depending tiers.
    Also these actions may take considerable time in large transcriptions.

June 15, 2006: Release 2.6.2

This is a release to fix a bug introduced in the previous release.
In ELAN 2.6.1 the “add”, “change” and “delete” linguistic type windows were integrated as tabs in a new “edit type” window. As an unintentional side effect of this integration process it became possible to create “invalid” or inconsistent linguistic types. This can lead to unpredictable behavior of the application and even corrupted files.

Other bug fix:

  • in the Search window with more than 3 constraint panels, the “in distance of” .. “tier-annotations” dropdown list would not update correctly after changing the selected “from matches found in” constraint panel.

May 31, 2006: Release 2.6.1

This minor release introduces a few new options, mainly concerning import and export. It also includes some bug fixes and performance improvements.

The main new features are:

  • The user interface of “add”, “change” and “delete” linguistic type has been changed and is now consistent with the “edit tier” user interface. The options are now available as tabs and a table is showing information on the existing types.
    Also an option to import linguistic types from other .eaf or template (.etf) files has been added. Controlled Vocabularies associated with imported types will also be imported.
  • export to CHAT, Shoebox or tab-delimited text:
    an option has been added to add the main media’s time offset to the begin and end time of the annotations.
    This way annotations can be in sync with the media in other applications
  • CHAT export:

    the media file name and the annotation’s begin and end time can now either be appended to the main annotation (on the same line) or it can be written on its own line. This last option was the default situation in previous versions.
  • Shoebox/Toolbox import:
    an option has been added to extract the begin time of annotations from the “record marker” (e.g. \ref 0008.218). Like in files produced by Econv.
  • Transciber import:
    an option has been added to create just one (speech) tier in ELAN for all speakers in the Transciber file, instead of a separate tier for each speaker (Turn oriented versus Speaker oriented import).

The main bug fixes and improvements are:

  • faster loading and saving of .eaf files, noticeable especially with larger files
  • improved handling of missing secondary linked files
  • Shoebox/Toolbox import:
    • the order of the markers is maintained as much as possible on import and export
    • improved handling of records or units with incomplete interlinearization
    • improved handling of a mixture of Unicode and ISO Latin markers

April 05, 2006: Release 2.6.0

In this release a number of new features have been added.
One important change is the introduction of a new linguistic “Stereotype”, allowing the creation of new relationships betwen tiers (see below). For this and some other features, changes had to be made to the .eaf structure. As a consequence files created with ELAN 2.6 that use these features may or may not open in previous versions of ELAN. In any way, they should not be edited in previous ELAN versions, so there is an effect on backward compatibility when using these features.

The main new features are:

  • New linguistic stereotype, “Included In”.
    This type can be used for time-alignable, dependent tiers. Annotations on such tier must be within the time interval of an annotation on the parent tier. The main difference with “Time Subdivision” is that gaps are allowed between annotation on a “Included In” tier.
    This feature is not compatible with previous versions of ELAN.
  • Introduction of a “Timeseries Viewer”.
    The Timeseries Viewer can display time series data as line graphs. Like the Timeline and Waveform viewer, it has a horizontal timescale bar and indicators for the media time and selected time interval. The viewer can contain multiple “trackpanels” and each trackpanel can display multiple timeseries “tracks”. The viewer has a facility to transfer data from
    a track to annotation values. All features are accessible from the right mouse button popup menu.
    It will be created after at least one supported timeseries data file has been associated with the transcription via menu Edit -> Linked Files and then the tab “Linked Secondary Files”. See the source release notes for more details on supported file formats.
  • Option to merge two transcriptions.
    A process in a “wizard” style; the first source can be the “current” transcription (open in ELAN) or an .eaf file, the second source is an .eaf file and the result is written to a third .eaf file (the source files stay untouched). The user can select which tiers from the second file should be added to the first file. If a tier exists in both documents the user can decide whether or not annotations in the first source may be overwritten by annotations in the second source.
  • The user interface of “add”, “change” and “delete” tier has been changed. These options are now available as tabs in the same window. A table has been added showing the tiers that are already defined in the transcription.
    Furthermore an option has been added to import tiers, without its annotations, from another .eaf or template file. The user interface of “add”, “change” and “delete” linguistic type will be adapted in the same way later.
  • Keyboard shortcuts to switch the active tier (in the Timeline viewer).
    Ctrl-up will activate the tier above the current active tier, Ctrl-down activates the tier below the current active tier.
  • A new, extended IPA input method has been added to the existing SAMPA and ipa-rtr methods. Like ipa-SAMPA it is based on the GATE Unicode Kit and aims to be as complete as possible (all characters in the “IPA Extensions” and “Combining Diacritical Marks” code blocks are included). A combination of two or three keystrokes results in a single ipa character. An overview of the keyboard mappings can be seen on this page.
  • Font sizes 24 and 36 have been added to most viewers.

  • Support for the hh:mm:ss.ms (00:00:00.000) time format has been added to the Shoebox/Toolbox import and export functionality.
  • Support for the SMPTE timecode format, hh:mm:ss:ff, has been added to the export as tab-delimited text option. The media time label, right above the play/pause button, now has a popup menu (right mouse button) to change the format to timecode as well. The choice for either PAL or NTSC is independent of the actual variant of the associated video(s).
  • French has been added to the list of languages for the user interface. We’re very grateful for all the efforts made by the translators.

The main bug fixes are:

  • A workaround has been found for the attaching and detaching of Quick Time based media players. Therefore this feature should now work on MacOS systems as well.
  • The Quick Time related problem of sound muting after 2 seconds of playback has been fixed.
  • Tabs and linebreaks within annotations are now maintained when saving to .eaf.
  • Changing the offset of a media file in “Synchronization Mode” is now correctly registered as a change to the document (Ctrl-S will now save the changes).
  • The problem on Time Subdivision tiers where the creation of a new annotation over only the first annotation was not correctly propagated, has been fixed.
  • CHAT import has been improved with the support for files where mediafile and time information is on the same line as the participants speech tier. Export to CHAT has not yet been extended with support for that format.
  • Import of Shoebox files with bundles of aligning lines would sometimes silently fail, resulting in missing annotations in ELAN. This problem seems to be fixed.
  • Some “export” windows (to SMIL, to QuickTime text) wouldn’t fit on the screen for transcriptions with a lot of tiers. This is fixed now.
  • Improved handling of page up and page down in the input method’s lookup list. Pressing the escape key now cancels the edit.
  • Faster loading of .eaf files, especially with very large files.

Known issues:

  • The Interlinear Viewer does not handle annotations on the new “Included In” tiers correctly. This will be fixed later.
  • Media playback on several platforms can be buggy with MPEG-2 codecs installed. On Windows 2000 and Windows XP problems with media playback can also be caused by an installation of iTunes.
  • On MacOS 10.3.x the installer of ELAN will work but the installed application won’t launch (the icon bounces in the dock and then just disappears). Use the Webstart version instead or the startup shell script (ElanMac_2-6-0.sh) in the installation folder. Or upgrade to MacOS 10.4.

November 15, 2005: Release 2.5.1

This is an important bug fix release.

In the previous release a few important components are missing, due to the changes in the structure of the source tree. As a result the search features are not functioning properly. This is a serious bug, so we recommend every ELAN 2.5 user to update to 2.5.1.

November 02, 2005: Release 2.5

In this new release we again introduce a few new features and bug fixes.

Apart from that (and transparent to the user), we have been screening the ELAN source code after the previous release, removing old, obsolete parts and restructuring the source tree.
A new source distribution, including a build script and libraries, has been made available and it will be updated with every new, main release of ELAN. (Read the release notes accompanying the source distribution.)
We hope to be able to update the manual this month, finally, and bring it in sync with the current state of ELAN. In the mean time a concise "getting started" manual for ELAN has been published by Albert Bickford (for which we thank him very much).
A link to this document can be found on the ELAN web page.

The main new features are:

  • Option to change the parent of a tier.

    In older versions this had been an option in the "Change Tier Attributes" dialog. But since this action potentially could corrupt the data structures in the document, it was disabled in the past few releases. Now, a separate menu item (Edit -> Change Parent Of Tier…) has been added that starts a multi step process (a.k.a. a wizard), that guides the user through the selection of a tier, the new parent tier (can be none) and optionally a new Linguistic Type for the tier. Because this can be a very complex operation (depending tiers have to be moved/changed as well and the complete process should be undoable), it is implemented such that a copy of the tier is created under the new parent, leaving the original tier untouched, thus ensuring data integrity. The user can review the results and delete the original tier if all went well. Since such operation changes the structure of the document, it would be a good idea to create a backup copy of the file beforehand.
  • Option to copy a tier.
    This process is similar to that described above. The differences are that here the inclusion of depending tiers is optional and the parent for the copy of the tier can be any tier in the transcription, including the tier itself.
  • Save annotations within a selected time interval as a new .eaf.
    This creates a copy of the transcription with only the annotations within the selected interval or overlapping this interval. All remaining annotations will be forced into the selected time interval and will be shifted to the left with a value of the selection begin time. (This means that the selection begin time corresponds to the new document’s begin time.)
  • To most text export options the choice has been added to use utf-16 encoding for the file. By default utf-8 is used, but some MacOS applications, like TextEdit, don’t interprete such files correctly. Noticeable e.g. in phonetic tiers.
  • Save a video frame as image (.jpg or .png)
    The availability of this option depends on the media framework and success can depend on hardware settings. The option can be found in the video’s right-mouse-button popup menu. (See also the known issues.)
  • The Waveform Viewer now can handle 24-bits .wav files in addition to 8-bits and 16-bits files.
  • Shoebox import now can handle Shoebox/Toolbox files with a mixture of aligned and unaligned records (records with or without time markers). In the calculation of the begin and end time of unaligned records, the "preferred block duration" is taken into account.
  • Portuguese has been added to the list of languages for the ELAN user interface.
    Again, many thanks to all volunteers providing the translations.

The main bug fixes are:

  • The bug, known on the mailing list as the "data eating bug", has been solved. This bug could corrupt the document’s structure when creating an annotation on a time-subdivision tier, while the time selection’s begin and end time coincided with the parent annotation’s begin and end time.
  • The export of search results in the multiple file search, now includes the hits in all browsable result "pages".
  • On MacOS it is now also possible to use platform specific input methods / keyboards. On tiers with the system’s language as the default language, the Mac Input Menu can be used to select an alternative input method.
  • The ruler in the Waveform Viewer and Timeline Viewer now attempt to display rounded time value as much as possible.
  • On Windows QuickTime is now used by default for .mov files.
  • On Windows opening a selection of a .wav file in Praat fails when there are space characters in the file path. ELAN now tries to convert folder names containing space characters to a format usable for sendpraat.exe. It will still fail when the filename itself contains space characters.

Known issues:

  • As reported on the ELAN mailing list there are some problems with QuickTime 7.0.3 on Windows ("QT for Java expired").
    Also, the installation location of necessary QuickTime files has been changed, causing them not being found by ELAN. We try to circumvene these problems as much as possible, but success is not guaranteed.
  • Caution
    On MacOS grabbing a video frame can sometimes cause the application to crash or hang. It is not yet known which combinations of OS, Java and QuickTime version are involved. Therefore caution is advised with this option.

July 19, 2005: Release 2.4.2

This minor release provides a few new features and bug fixes.

The main new features are:

  • Export as Interlinear Text:
    this generates a (UTF-8) text file of interlinearized annotations on a selection of tiers. The output is highly customizable and the user interface for preview and output options is similar to that of the printing feature (see Printing).
    The interlinearization is based on the number of characters in related annotations. Therefore, the resulting text file can best be viewed in a text editor using a “monospaced” or “fixed-width” font.
  • Printing:
    A few enhancements have been realized here:

    • Line wrapping is implemented. This only has an effect in combination with one of the block wrapping styles.
    • The option to only print annotations within the selected time interval has been activated.
    • In addition to the existing tier sorting options, it is now possible to customize the order of the tiers in the output by moving them up or down in the table of tiers.
    • The preview window is now resizable.
    • Improved preview performance, noticeable especially with larger files.
  • The maximum number of constraints in the search dialog has been raised to 10. The search window will now always fit on the screen, irrespective of the number of constraints that have been added to the query panel.
  • The search results in the multiple file search are grouped in browsable ‘pages’ of 50 hits each.
  • On MacOS the menu items are not always updated correctly and they sometimes get mixed up. To circumvent this problem an option has been added to the empty, startup window to change the appearance of the user interface to the native MacOS X look and feel.
  • Export as Tab-delimited Text:
    The export window will now always fit on the screen, irrespective of the number of tiers in the transcription. The order of the selected tiers can be changed and the user can further customize the output by selecting which time columns and which
    time formats should be included in the export. Plain millisecond values have been added as possible time format.

The main bug fixes are:

  • Changing the media playback rate on Windows only changed the rate of the first media player. This has been fixed: all media players are now updated.
  • The Gridviewer in multi tier mode would display incorrect column headers after changing the name of one of the dependent tiers in the table. Is fixed.
  • The CHAT export window will now always fit on the screen, irrespective of the number of tiers in the transcription.
  • The search result table in the multiple file search window sometimes showed incorrect (negative) time values for certain annotations. This prevented the jump to the right position in the file when clicking such a hit. Is fixed.
  • Shoebox import:
    Blockwise wrapped lines were not always converted correctly to annotations in ELAN and import with a Shoebox .typ file had become unreliable with a previous fix. Both problems seem to be fixed now. When a .typ file is used, the encoding of a tier is not always detected.
  • Shoebox export:
    ELAN now tries to detect whether or not a new MarkerRecord (named ‘block’) should be written to the Shoebox file.

Import from and export to Shoebox remains a bit problematic. The current implementation is very likely not the ultimate solution and it is subject to change in future versions.

May 04, 2005: Release 2.4.1

This is a minor intermediate release. The main changes are:

  • Import and export.
    • The “database type” reference in the header of Shoebox/Toolbox files is now used with import and export.
    • An additional dialog has been added to the Shoebox export that allows:
      • to specify a subset of tiers for export
      • to specify the order of the tiers for export
      • to specify the number of characters for block wrapping
      • to specify the source of marker information for export, either a .typ file or a marker file
      • to automatically create a marker file based on ELAN tier names
    • When importing a Shoebox file without time information, a custom duration per block can now be specified.
    • The import and export of CHAT files has been improved.
  • The selected entry in the list of a controlled vocabulary can now be committed by the Enter key (as well as the traditional Ctrl+Enter combination).
  • Activating an annotation on a symbolic subdivision or symbolic association tier now changes the selection to the interval of the first time aligned “ancestor” annotation.

Main bug fixes:

  • Multiple file search:
    • when some files in the search domain cannot be read or parsed, they will be skipped without aborting the search operation altogether.
    • the media times in the result table sometimes weren’t correct. This is fixed now (See known issues).
    • sometimes a file would be opened twice when clicking annotations in the result table. This is fixed now.
    • when clicking an annotation in the search result table, ELAN will try to position the annotation as well as possible within the main view. (This sometimes fails with the first annotation, the one that opens the file.)
  • the filter input field in the “Filter Tier” dialog sometimes disappeared. Is fixed.

Known issues:

  • In the multiple file search result table, unaligned annotations on time-alignable tiers and their depending annotations, can have a begin- or endtime of -1 or an incorrect duration of 1 ms. Clicking on such a result annotation will open the file, but the view will not jump to the right position.
  • Annotations on subdivision tiers that have been exported to Shoebox and imported again will sometimes be concatenated.

March 17, 2005: Release 2.4.0

In this release we were able to add a substantial number of new features:

  • The maximum number of coupled video players has increased from 2 to 4. Double click on a thumbnail makes this video the largest one.
  • Other media formats than MPEG and wav are now also supported. Choose ‘All files’ in the file open dialog. ELAN will then check if the used media framework can handle the selected file.
  • The popup menu of a video panel contains a new menu item giving information about the video.
  • Linked media files of an ELAN document can now be added, removed or changed at all times.
  • It is now possible to switch between Annotation Mode and Media Synchronization Mode even when annotations already exist. Offsets for audio players can now also be specified.
  • Additionally, it is possible to shift all annotations in a document in time.
  • The user can specify preferred values for playback rate and volume. Keyboard shortcuts are avaible to toggle between the default playback rates and volumes, and the user defined ones.
  • The popup menu in the time line viewer allows the user to choose whether the active annotation is shown with a bold blue line or with a normal one.
  • Annotation begin and end times can be changed by dragging the ends of the annotation segments in the time line viewer (move mouse with Alt pressed)
  • “Open” vocabularies are supported. The user can enter annotation values that are not in a tier’s controlled vocabulary by pressing Shift when double clicking an annotation (or, in the time line viewer, by right clicking on the active annotation and choosing Modify Annotation).
  • Segmentation mode: it is possible to make a rough initial time segmentation on a given tier, just by pressing Enter at relevant times. When clicking ‘apply’ this segmentation is stored in the tier as a number of empty annotations. Adding a complete segmentation to a tier is undoable in one operation.
  • Filter tier: annotations on a source tier can be copied to a (symbolically associated) target tier. Filter value strings can be specified. These strings are filtered out of the annotation values. This feature allows you for example to filter out embedded encoding or make a copy of a tier’s annotations.
  • Localization:
    ELAN’s user interface can now be switched to German as well.
  • We made ELAN available for Linux as well. Note that media playback behaviour is not optimal. Use ELAN on Linux only when exact media alignment is of lesser importance.
  • Import and export
    • In the popup of the Grid Viewer we added an ‘Export Table As Tab-delimited Text’. The Grid Viewer’s visible columns are exported to tab delimited text. This is also the case when the Grid Viewer is in multi tier mode.
      The option is also available in the ‘search results’ table.
    • Export as Traditional transcript: generates a (UTF-8) text export of annotation values on selected tiers. Annotations are sorted by begin time. Several options are available to customize the export.
    • Any document opened in ELAN can now be exported to (non-XML) CHAT format. The user has to specify mapping of ELAN tiers to CHAT’s 3 character labels manually. Annotation values on subdivision tiers in ELAN are concatenated per root annotation.
    • Transcriber import is added.
    • WAC (Word Annotation Converter) import was removed, since it was buggy. We will only consider debugging and adding it again when it is explicitly requested.
    • Output of any document in ELAN to Shoebox/Toolbox format is added. ELAN adds additional information about time alignment, participants and linked media files to the exported Shoebox file. It is up to the user to define a database type, interlinear ‘processes’ and language definitions in Toolbox to match the exported data. In a future release we will streamline going back and forth between Shoebox and ELAN a bit more by automatically generating marker files to go along with the exported Shoebox files.

January 12, 2005: Release 2.3.1

This is a minor intermediate release. The main new functionality is:

  • The Search module is extended with a Replace function.
  • Printing is added. We chose a representation on paper that shows all annotation text, maintains annotation structure and time ordering and can give an indication about time alignment. Using Print Preview the user can set a number of parameters to control the exact layout.
  • Several locales are added. The ELAN user interface can now be shown in English, Dutch, Spanish, Catalan and Swedish. We express our thanks to the volunteers who did the translations for us.

Main bug fixes:

  • Clicking on a result in the result list of multiple file eaf search now correctly opens the relevant document at the right place.
  • Search queries can now not only be stored, but also be read in again.
  • In case of a movie without sound ELAN sometimes froze on Exit. This is now fixed.
  • Playing selections in very big mpg files was not smooth. Is solved, size does not matter anymore.
  • When importing a Shoebox/Toolbox block with no dependent annotations for the first word alignment was wrong. Is now fixed.
  • In Shift mode, when resizing an annotation such that it overlapped another annotation this annotation was deleted instead of shifted.

Know issue:

  • Colors and special characters render not nicely in the Preview window on the Mac.

November 17, 2004: Release 2.3

The main new features and improvements are:

  • Search functionality is improved and extended: continued search on previous results, export of results, saving and restoring of queries and search across all tiers are now possible.
  • Regular expression search on annotation documents in a “basket” of local folders.
  • Further improvements on media playback on Mac.
  • For experimental use it is now possible to annotate 2D regions of a video signal as well. Since this is implemented on basis of Quicktime, the price to be payed is that it is not possible to use native media playback at the same time. To use this feature Quicktime has to be installed, including Quicktime for Java.
  • Shoebox/Toolbox import is substantially improved: users can define and reuse their own marker or tier setups (as alternative for using a Shoebox typ file), Toolbox import is supported, including (interlinearized) Unicode encoded fields, users can define their own participant marker and exclude markers from import. A number of special text alignment problems is solved.
    Temporarily Shoebox export is unavailable. In ELAN 2.4 revised export to Shoebox/Toolbox will be added again.
  • PRAAT connection: when a wave file is opened in ELAN PRAAT can be opened from the Signal Viewer on either the whole media file or on the current selection in ELAN. PRAAT and sendpraat have to be installed.
  • Undo is implemented for all edit actions where an annotation, tier, type or controlled vocabulary is modified. Also ‘clear selection’ is undoable. ELAN currently remembers the last 25 undoable commands.
  • Built-in tokenizer: it is now possible to automatically generate new annotations for every token in the annotation of some source tier. Any tier can be used as a source tier, time subdivision tiers or symbolic subdivision tiers can be used as target. This can for example be used to decompose utterances into individual words. Delimiters for tokens are user definable. Default delimiters are newlines, tabs and spaces.
  • On Macs the Command (Apple) key is now used for shortcuts instead of the Control key (with the exception of Control-Space).
  • The order of entries in a Controlled Vocabulary can now be changed. All modifications in Controlled Vocabularies are made undoable as well.
  • A number of small additions and improvements.
  • We intend to make localized versions of ELAN available for English, Dutch, German, Swedish and possible Finnish a few weeks after the official 2.3 release.
  • The manual is made up-to-date for this version.

The most important bug fixes for this release are:

  • A number of bugs concerning the manipulation of complex graphs of alignable, sometimes unaligned, annotations are fixed.
  • The Annotation Density Viewer now also shows annotations for long media files correctly.
  • Input methods on the Mac are now available.
  • Changing font size in Text and Subtitle Viewers now works correctly.
  • Control-Q on Mac does not quit without save anymore.

July 9, 2004: Release 2.2, for Native Windows Media, JMF and Quicktime

Main features and changes:

  • On Windows, ELAN now by default uses native Windows DirectX libraries to handle media playback. This finally gives smooth and accurate video playback for MPEG-1 video. The annoying little ‘jump’ at the end of selection playback is also gone. We continue to work on native media handling for the Mac version of ELAN. MPEG-2 may work, but, dependent on which MPEG-2 codecs the user has installed, problems may occur. We are looking for an MPEG-2 codec that functions properly.
  • The search user interface and functionality are substantially improved
  • The Grid Viewer is extended. It can now optionally show annotations from a number of dependent tiers in a tabular way, where different columns contain annotation values from different tiers. This can be done for all tiers where the annotations are one-to-one associated with annotations on dependent tiers. Where no child annotations exist they can be created by just filling in an empty cell.
  • The Timeline Viewer and Interlinear Viewer now support several new ways to hide or display tiers and to sort tiers.
  • Controlled Vocabularies can be defined and associated with Linguistic Types. When annotation values of such a Linguistic Type are created or modified a list of alternative values is shown for selection. CVs are stored inside the EAF files. They can be reused for other documents as part of a template document, or by importing CVs from existing template documents using ELAN’s CV Editor.
  • Export of ELAN’s current document view to an image.
  • Export of an annotation document, or a part of it, to a SMIL ‘clip’.
  • When the user has m2edit installed, ELAN offers the possibility to cut out an MPEG clip as well.
  • Import of CHAT files, both pre-UTF-8 and UTF-8. Time alignment following a couple time conventions is maintained. Where time alignment applies to a number of consecutive CHAT utterances the available time interval is equally divided over the utterances. When no time alignment is present by default one second time intervals are assigned to consecutive utterances.
  • Import of Shoebox/Toolbox is improved: As alternative to using a .typ file the user can now manually specify which tier markers are used in the imported Shoebox file. In the next release this will be improved to support Unicode (from Toolbox), import as time subdivision, store and loading of specified shoebox markers
  • The RTR input method for IPA is revised a bit: more characters are made available under lower case keys.

Remark: during webstart download and first startup, there may be an additional dialog asking you if you want to install software from MIIK Ltd. Answer yes. This software is needed to play MPEG natively.

April 8, 2004: Release 2.1 for JMF and Quicktime

This version of Elan REQUIRES JAVA RUNTIME ENVIRONMENT 1.4,
it will not run on JRE 1.3 anymore.

Main features and changes:

  • Support for template documents to make reuse of tier setups easier.
  • New Unicode input methods for Korean, Georgian and Turkish.
  • Preferences are now stored between Elan working sessions. These are both preferences for Elan (like last used directories for eaf files, media files, shoebox type files) and preferences for individual documents (like media time, selection, active tier, etcetera).
  • Even if media files for some eaf file are completely missing the document can still be opened for inspection and modification.
  • A ‘shift’ mode is added to help alignment of imported data. Unlike the already existing ‘bulldozer’ mode gaps between annotations are maintained.
  • Minor improvements to the ‘Find’ function were made
  • Several small fixes and additions.

Known issues:

  • In Shift mode and in Bulldozer mode, inserting a new annotation that overlaps existing annotations does not work properly. Try to avoid this by first creating the new annotation and then adjust it’s time interval, or by creating the new annotation in ‘normal’ mode.

January 30, 2004: Release 2.0.2 for JMF and Quicktime

This version is a major upgrade. Elan’s viewer and media handling internals are completely re-engineered, as is the handling of user commands. The user interface is completely redesigned, including shortcut keys.
Preparations for undo functionality are made. Elan 2.0.2 runs on both Java Runtime Environment version 1.3 and version 1.4 (recommended are 1.3.1 or higher, 1.4.02 or higher). The next Elan release will NOT RUN ON JRE version 1.3 ANYMORE.

Main new features and changes:

  • All viewers for one annotation document are now shown in one document window. The video panel can be detached into a second window. This can for example be useful to display MPEG-2 video on a separate monitor. Most other panels
    in the document window will be made detachable in future releases.
  • Several new and/or revised viewers:
    • Waveform panel: supports a few alternative ways to show stereo signals, can be coupled to the time axis of the time line viewer or not, and has a wide range of zoom levels. Drawing performance is improved a lot.
    • Both the JMF and the Quicktime version use the same Elan-specific control panel, with buttons for navigating in, and playing the video and/or audio signal. The panel also has buttons for manipulating and playing the time selection, for jumping to the previous or next annotation on some tier, and for toggling selection mode and loop mode.
    • The time line viewer is completely rewritten, has improved drawing performance and has a number of user interface improvements
    • A ‘context panel’ is added, showing the current time and selection along the complete media time axis, as well as the density of annotations along media time. This panel is also time-synchronized and can be used to modify time and selection as well.
    • The interlinear viewer is more robust and has better drawing performance.
    • The gridviewer shows a list of all annotations of a selectable tier. Optional columns for sequence numbers and annotation duration are added. Times can be shown both as time code and in milliseconds. Most columns can be shown or hidden. Annotation text values can be modified by double-clicking on a cell in the gridviewer’s table.
    • A text viewer is added. It shows all annotation text values of the annotations of a selectable tier as ongoing scrollable text. Annotation boundaries can optionally be shown as dots embedded in the text. The text viewer is time synchronized and shows selection and active annotation. Annotation values can be modified directly be double-clicking on the relevant text.
    • Subtitle viewers: four concurrent subtitles can be shown for any combination of tiers. The subtitles are independently time-synchronized. Annotation values can be modified by double-clicking on a subtitle.
  • ‘Save As’ is now supported.
  • Time selections are now made or modified in a completely new way. Next to dragging and shift-click in the time line viewer or wave form panel Elan now has a special ‘selection mode’: all time navigation and playback buttons modify either the begin or the end of the selection when in selection mode.
  • Two time-synchronized video panels are supported now. The user can specify the begin time for each of the two separately.
  • Media files do not have to have the same name as the matching .eaf file anymore, and do not have to be in the same directory either. When files can not be found at the locations stored in the .eaf file, first the eaf file’s directory is checked, then the user is prompted to specify a location.
  • Elan’s user interface can be localized on the fly. Currently supported languages are English and Dutch. It is now easy and straightforward to support other languages. Volunteers for translation of English user interface texts to some other language are welcome.
  • Formats
    • New EAF format (version 2.1). The only change is that Media Descriptors are added to store media file locations, mime types, time origins and, typically in case of wav files, the source file that the media file is an extraction or conversion of. Elan 1.4.1 can still read EAF 2.1 files, but will ignore the Media Desciptors. Elan 2.0.2 can read old EAF files, and will add Media Descriptors when saving.
    • Shoebox import bug fixes: Elan did not read interlinear text hierarchies properly when the first tier in the first shoebox block was not the ‘top tier’ of the interlinear text (\tx, \word, … tier). Now it does. Tiers that use SIL IPA characters are converted to Unicode when for the tier a \lng field exists int the .typ file with the values “IPA” or “Phonetic”.
  • Time accuracy: all times in all viewers are correct, and synchronized at all times. There is one annoying issue that can not be fixed on the short term: when playing a time selection, playback of the video continues a few frames after the end of the selection. How much depends on the computer or operating system running Elan. Right after this ‘overshoot’ the media time is set to the exact end time of the selection, resulting in a little jump in the video playback. Audio does NOT have this problem.
  • A number of bugs is fixed. Important ones:
    • Linguistic types that were stored with more than one document open were always those of the document that was opened last. When saving the document that was opened first the wrong linguistic types were written, resulting in files that could only be opened again after manual correction.

Known issues:

  • Some problems exist only on Macs:
    • Playing audio/video with a gridviewer showing that contains many annotations has a large negative impact on performance. Clicking the tab of some other viewer solves this
    • There are redrawing issues in the time line viewer when playing in selection mode. This can be corrected by resizing the document window.
    • Sometimes the video initially does not show. Solve by resizing the document window or playing the video
    • When opening a second document window the video of this document shows at a smaller size. For the moment, there is no fix.
    • With an MPEG video opened in one document and a second wav-only document the audio in this second document mutes after a few seconds. The only solution is to open the wav-only document as the only document in a new Elan session.
    • Detaching the video panel on the Mac does not work yet.
    • Unicode input methods do not work yet, because they have to be installed using ‘root’ privileges.

Prerequisites:

  • Java Runtime Environment version 1.3.1, or preferably, JRE 1.4.02 or higher.
    This Elan release will be the last one not requiring JRE 1.4
  • On Macintosh: Quicktime should be installed.
  • On Macintosh: Mac OS 10.3 is highly recommended, because it offers substantially better performance of Elan.

June 26, 2003: Release 1.4.1 for JMF and Quicktime

This version fixes a number of bugs, most of them introduced in version 1.4:

  • time selections were reset to unwanted values during a number of different user tasks
  • time subdivisions did not work properly when the user used ‘insert annotation before’ or ‘insert annotation after’
  • annotations with longer duration than fitted in the time line viewer were not drawn when neither begin nor end time were in the visible area
  • cursor highlight (dark blue) disappeared when the user started to make a new selection
  • when creating new annotations sometimes existing annotations were overwritten
  • when entering annotations with longer duration than fits in the time line viewer the current time now jumps to the beginning of a newly entered annotation, to make it possible to starttyping an annotation value right away

May 22, 2003: Release 1.4 for JMF and Quicktime

Added functionality and fixes since version 1.3:

  • it is now possible to add frames to the selection by playing them, escape to deselect
  • ‘play selection’ plays from current timeif in selection, instead of from selection begin
  • color coding of tier names now is maintained when tier order changes
  • set and display active tier in time line viewer by clicking on tier name
  • File – Open activated again
  • Entering new annotations for long selections is improved: the time line viewer now scrolls to the begin of the selection, not to the end anymore.
  • ELAN can now work with Quicktime as alternative to the Java Media Framework (JMF)
  • ELAN for Quicktime works also on MacOS 10, however:
    • very fast start/stops can hang the program, try to avoid this
    • scroll in time line viewer during playback can hang the program, try to avoid this
    • Ctrl-Space as shortcut for play/pause is (temporarily) replaced by Ctrl-p on the Mac
    • There is currently only limited support for Unicode script systems
  • from now on ELAN is tested and supported on Windows XP as well
  • editing of time subdivision annotations is now correct
  • a number of bug fixes

Known issues

  • editing time subdivisions of time subdivisions is still buggy
  • when importing a WAC file, the dialog does not show .wac.txt files, select "all files" to make them visible

April 4, 2003: Release 1.3

Added functionality and fixes since version 1.2:
 

  • workarounds for JMF (Java Media Framework) bugs
    • JMF forgets a lot of it’s state at ‘stop’. ELAN now remembers it for JMF and restores it.
    • JMF is slow in handling start/stops. It can even crash when start and stop is too fast.
    • ELAN intercepts and discards start/stops that are too fast for JMF to handle in time.
    • JMF is too slow to be able to play very short segments.
    • ELAN now prevents playback of too short segments (under 80 msec)
    • ELAN prevents possible errors at end of a media file
    • Reported errors that were fixed in this way
      • delay between media and viewer playback
      • "3 seconds problem": audio starts playing a couple of seconds later than video
      • media plays at high playback rate without a rate change by the user
  • optimization and improved playback of viewers. ELAN is now substantially less demanding on CPU usage.
  • save shoebox imported file to eaf
  • import wac (word annotation converter) files
  • save wac imported files to eaf
  • another XML parser is used (now the same as for the IMDI tools: Xerces)
    • validation is now done when the eaf schema is available
    • eaf in now rendered in a more readable way
  • an autosave function is added
  • shoebox import: SIL IPA characters are now converted to Unicode during import
  • an ‘active tier’ item is added to the right mouse menu in the time line viewer
  • panels do not resize unexpectedly any more after edit operations
  • the JMF media controller is replaced by a single ‘play’ button
    • 1. to have control over what happens when you press it,
    • 2. to prepare for a next Elan release,
    • 3. because all removed functions are already available otherwise.
  • a number of small improvements and fixes

Known issues

  • slow motion playback on some machines does not work, since JMF behaves buggy. Please test if this is the case on your machine by playing slow motion and inspecting synchronization between crosshair in audio wave form and audio as you hear it. If your system has this problem, note that there can also be a 1 % error in the end time of selected and annotated intervals. (approx. 1 video frame for a 5 second interval). Please report the characteristics of your machine when you notice this problem.
  • File – Open is disabled, because currently dealing with two video files simultaneously causes problems.
  • during playback on slower machines sometimes the wavepanel does not update properly when going to the next page during video playback.
  • Resizing the ELAN main window during playback can cause JMF to behave unpredictably. Please resize only when playback is stopped.
  • JMF can still behave unpredictably and even crash ELAN if your system CPU load goes over 100%. For this version of ELAN that should only happen on slower machines. Use your Windows Task Manager if you want to check this.

October 24, 2002:  Release 1.2

The transitional release replaces Release 1.1

August 19, 2002:  Release 1.2  in transition

Main features added and changed since Release 1.1:

  • Automatic creation of a backup copy when first saving a document.
  • Exporting a document as a tab-delimited text file.
  • Importing and exporting between Elan and Shoebox, allowing time alignment in ELAN and interlinearization in Shoebox.
  • The Interlinear Viewer.
  • Color coding to visualize complex hierarchies.
  • Slider for playing at rates between 5 and 200%.
  • More specific dependency relations between tiers using constraints.
  • Support of two modes of annotation, the Overwrite and the Bulldozer modes.
  • Entering annotations via the Inline Edit box, in addition to the already existing separate text entry window.
  • Subdividing annotations, i.e., entering annotations before/after existing annotations.
  • Modifying annotations from within the Grid Viewer.
  • Improved search options, allowing regular expression search and combination of multiple subqueries.

Most important known issues:

  • Time subdivisions are still buggy. Modifying annotation time alignment of annotations with time subdivisions often results in incorrectly ordered times. Use with extreme care.
  • It is not possible (yet) to save imported Shoebox files to EAF format, or to export them as tab-delimited text.
  • SIL IPA characters in Shoebox files are displayed wrongly.
  • There are still issues with exact playback of media segments on specific windows platforms. These issues are mainly due to known bugs in the Java Media Framework.

Release 1.1 remains available as "Elan", and
Release 1.2 is available as "Elan (new)" during transition.
 


February 28, 2002:   Release 1.1

Main features added since Release 1.0:

  • Renamed EAT (Eudico Annotation Tool) to ELAN (Eudico Linguistic Annotator).
  • Added copyright notice.
  • Support for Quicktime mov files.
  • Added shortcuts for all menu commands.
  • Added active tier concept.
  • Run ELAN from IMDI BC Browser.
  • Run ELAN as annotation document viewer (read-only mode).
  • Run ELAN as a viewer for mpg and wav files.
  • Tier names are now color coded per dependency hierarchy.
  • Video player follows current media time while setting a selection.
  • Added single step forward and backward.
  • Different opening dialog.
  • Warning if user tries to open media file instead of EAF file.
  • Grid viewer opens on right mouse click on tier image in time line viewer. This viewer is synchronized with media time, selection and editing. Can be used to navigate on basis of annotations.
  • EAF extension changed from .xml to .eaf
  • Single click on annotation in time line viewer now sets cursor (indicated by dark blue line).
  • Subtitle viewer now supports line wraps. Default number of lines shown is two. An additional split pane is added to resize the subtitle viewer.
  • New desktop icon for ELAN.
  • Save in order of display, instead of in alphabetic order.
  • Autoscrolling while selecting in time line viewer.
  • Empty annotations supported.
  • Warning if user wants to create a file that already exists.
  • Independent tiers can now be made dependent, and dependent tiers can be made independent or dependent on another parent. Annotation time intervals are adjusted as good as possible.

Main bug fixes since Release 1.0:

  • Properly clean up memory after quitting ELAN.
  • New document starts now completely empty.
  • Repaired play around selection/cross-hair.
  • Saving annotations in two steps now works properly.
  • Incorrect behavior after deletion of tier that is in the subtitle viewer is fixed.
  • Incorrect scrolling of tier names after editing of annotations is now fixed.
  • ‘new annotation’ command improved.
  • Opening a document with dependent tiers made ELAN freeze. Now fixed.
  • Time line viewer screen updates are less frequent, resulting in better playback performance.
  • Swapping of begin and end times when selecting now works properly.
  • Change of tier names or tier attributes is now immediately updated everywhere in the user interface.
  • Rename linguistic type. Now the new type is assigned with all tiers that used the old type.
  • Delete linguistic type. Now is only possible when no more tiers use the linguistic type.
  • Under some circumstances annotations on the same tier could overlap after editing. Not any more.
  • Consistent terminology in menus: ‘tag’ replaced by ‘annotation’.
  • For Chinese lookup: China/Taiwan replaced by Simplified/Traditional.
  • Dependent annotations sometimes appeared to get lost. They were not properly restored, now they are.
Facebooktwittergoogle_pluslinkedin