Transana (Review)

From LinguisticAnnotation
Jump to: navigation, search

This is a review of Transana written by Thomas Schmidt for the working group "Transcription and annotation of primary data" at the E-MELD workshop 2006.


Transana is a tool for managing, transcribing and analyzing digital audio and video recordings. It runs on Windows and Macintosh platforms. It can handle video files in MPEG and AVI formats (no support for MOV files) and audio files in MP3, WAV, AU and SND formats. The main idea of the tool is to support the user in organizing long video files into smaller units of analysis, in assigning keywords to such units and in producing transcripts that can be (optionally) time-aligned with the recordings. The former two of these functionalities are, to my knowledge, not provided by any other transcription tool to the same extent. The latter functionality is comparable, for instance, to what the CLAN tool offers. The developers are David Woods and Chris Fassnacht from the University of Wisconsin. Transana is open source software, and its development is ongoing. The tool comes in a single-user and a multiple-user version. I tested version 2.10 of the single-user version.


I found the tool very well documented – it comes with an on-line tutorial to familiarize new users with the tool, and the website offers a forum for asking questions about specific aspects of the tool. My own questions to that forum were all answered within less than an hour (!).

I had absolutely no problems in installing the tool. Setting up a video for transcription took some time, mainly because I found the terminology in some places not very intuitive. This, however, was only an initial inconvenience – after I had understood the terminology it was perfectly straightforward.

The tool asks for a lot of categorization data during the initial setup of a data set. As Schwab (2006) points out, this is something that many users will find irritating from a methodological point of view, because many methodologies request that no pre-formed categories be applied to the data before an in-depth analysis. It is, however, possible to ignore this initial categorization and add such information at a later stage.

I have not extensively tested the tool's functionality for “identifying analytically interesting clips, assigning keywords to clips, arranging and rearranging clips, creating complex collections of interrelated clips, exploring relationships between applied keywords, and sharing an analysis with colleagues.” (quote from the tool's website). Rather, I have concentrated on its transcription functionality. Schwab's (2006) account of the tool's functionality in the other areas is, however, very positive.

I used the tool to make a transcription of a five-minute (MPG) video of a televised interview. Navigating in the recording (via keystrokes or by selecting a portion of a waveform) was straightforward, and the video and audio playback was smooth and error-free (on Windows, the tool uses the Windows Media Player for playback). Understanding the focus management (i.e. the way in which either the window for video display or the text window in which the transcription is edited are activated) was a bit tricky sometimes. Otherwise, I found the tool's functionality helpful for producing a fine-grained transcription of the video recording.

After switching to a Unicode-Font (Arial Unicode MS) in the transcript window, I had some problems entering German Umlauts, IPA characters, Cyrillic characters or Chinese characters (the text window would display other things than I had entered). This seemed to be display-related rather than caused by a “real” lack of Unicode support (saving and reopening the transcription yielded what I had intended to enter), but it made it difficult to use Non-Standard-Latin characters in the transcription.

After producing the transcription, I used Transana to time-align the transcription with the video-recording. This is done by inserting timestamps at freely selectable positions of the transcript text. This worked well, too. However, I found it inconvenient that the tool requires all timestamps to be in an increasing sequence. In cases of overlapping speech, this means that time alignment can sometimes not be done as precisely as one would like.

I encountered a few minor bugs (e.g. clicking on “Select clip text” produced an “Unhandled Exception”). With one exception (I produced a full crash with the function “Locate clip in episode”), these were non-fatal, i.e. I did not lose any data when they occurred. All in all, I found this a good transcription tool as far as its functionality is concerned. Users with little computer literacy will probably value its simplicity and user-friendliness as well as the extensive documentation. Compared to tools like ELAN or PRAAT, the functionality of Transana is somewhat limited, but this may be seen as a good thing for those users who want to keep their transcription process simple.

In its present form, however, the tool's data format fails to comply with current best practice standards. Transcripts, keywords, video segmentations etc. are stored internally in a MySQL database, and there is nothing wrong with that. What is problematic is that Transana does not offer a usable exchange format for its data. It does provide an XML export, but in this export, transcripts are represented as one big stretch of character data, interspersed with RTF formatting instructions. Thus, there is no real content-oriented markup of the transcription. This will make it difficult or even impossible for other transcription tools to provide adequate import or export filters for Transana data. Moreover, when the transcription contains timestamps, the tool produces a non-well-formed XML export (because it fails to substitute angle brackets with their corresponding escape sequences), which will – obviously – make an exchange of such data even more problematic. Thus, with Transana alone, users will presently not be capable of producing exchangeable and archivable data. The workaround would probably be to export transcripts as RTF and let some other tool transform this RTF into a more sustainable (XML) format, but since one of Transana's main pros is its simplicity, the need for such an additional procedure may severely diminish its attractiveness for many users.

Literature/References

  • Götz Schwab (2006): Transana - ein Tranksriptions- und Analyseprogramm zur Verarbeitung von Videodaten am Computer. In Gesprächsforschung 7, 70-78. Available online from http://www.gespraechsforschung-ozs.de.
Personal tools