Linking digitised archival versions of fieldtapes to transcripts is an important step in providing access for analysis, for return of the material to speakers and for lodging with repositories. This linked data allows us to rework our transcripts incrementally as we learn more about the language. Currently there are a number of transcription tools that link a transcript to an audio file. Having used one of these tools, the researcher needs to be able to search through all of their fieldtapes and have instant access to the audio data.
After using SoundIndex (http://michel.jacobson.free.fr/) for marking audio start and end points within a transcript file there did not appear to be a tool for combining these into a searchable corpus. As I did not want to segment the 20 audio files in the sample I used QuickTime commands in a HyperCard application (called Audiamus) to instantiate the links to start/end points within an audio file.
Having a working tool for interacting with audio data via its transcript provides a method for tagging all derived data (like texts or example sentences) with its location in the audio file. As the audio files are not segmented users can hear as much of the context as they need. The transcript can be exported with timecodes in whatever format is desired: tab-delimited; Quicktime (for creation of subtitled movies); or in a format ready for glossing in Shoebox. In this presentation I will demonstrate this tool and discuss the functions that flow from having a linked dataset of this kind, including: establishing playlists of example sentences; clipping examples, or just timecodes for pasting to text documents; providing a concordance point of entry to the transcripts and hence the audio data. By making it easy to work with audio files we can improve our practice and provide reusable archival material for posterity.