AudioText Linking Tool

Brett Baker, Michael Kovacs UNE

The AudioText Linking Tool (hereafter 'the linking tool') is an electronic system for managing, displaying/playing, searching and archiving language material in both text and audio formats.

The Linking Tool uses time-aligned transcripts of audio files as a way of displaying both the text and playing the associated audio. Time-aligned transcripts are files which contain a transcript of a named audio file, together with time values (in minutes and seconds) recording a start and end time in the audio file for a segment of audio corresponding to a segment of the transcript (a sentence for instance). These files are stored on a server in an SQL database, and displayed via a web interface.

Unlike many other instantiations of this basic audio-text linking function, the Linking Tool does not download an entire audio file at once, nor is the audio file segmented. Only those portions corresponding to a line of transcript are downloaded when selected by the user. There are several advantages to this: the original audio file retains its integrity, the researcher doesn't waste time 'slicing and dicing' segments of audio to associate with the text file, and downloading the portions of audio doesn't use up unacceptable amounts of bandwidth.

So far, this kind of thing has been done before by other researchers. The innovation in this particular tool lies in the fact that both these sets of files (time-aligned transcription, and associated audio file) can also be accessed by a dictionary database (or, other kinds of databases). One can search for words in the dictionary, and then, clicking on the word, can hear any instance of that word in the set of audio recordings. So the Linking Tool can also behave as an audio dictionary, which was its original aim.

The system also has an innovative 'admin' interface which allows users to directly upload more audio and transcription files to the server. The system will allow these files to be browsed immediately. The idea is that organisations or individuals (language teachers or students, Aboriginal language workers, linguists doing fieldwork) can use the Linking Tool as a server-side archive for their recordings and transcriptions.

The tool is platform-neutral (Mac, PC, UNIX) because the interface is web-based, and all the files and scripts are stored on the server. This means that the tool can be accessed anywhere with internet access, and that there are no version-control issues. At this stage, there are no comparable electronic tools which allow this range of functionality in a web-based format.