The AudioText Linking Tool
Brett Baker and Michael Kovacs (UNE,
Armidale)
How we came to ask the question.
- dissatisfaction
with current formats for electronic dictionaries.
- incorporating
audio required firstly, digitising analog recordings, and secondly incorporating
them somehow with a dictionary database
- problems
with the FMP format: proprietary software, x-platform sound issues,
scripting stupidity
- suggestion
to go to browser
- loses
functionality on the database side, but this is more a problem for the researcher
than the client base
- advantage
is that we can use open code, well-understood tools for building
dictionaries
- and we can
use currently existing tools for playing sections of audio
The current project.
The components of the linking tool are the following:
server-side database and associated interfaces:
'audio', 'dictionary', 'admin', ('help')
URL: http://turing.une.edu.au/~linking
What's it good for?
·
The linking tool is basically an archive for
digitised audio together with transcripts which are time-aligned to the audio,
which look like this:
15.8.00.02.xml
The linking tool has a number of uses:
archive for digital recordings
- because
these are transcribed, information in the recordings can be quickly located
and downloaded
- the tool has
an admin interface for users, allowing them to upload audio, transcript
and dictionary files and (depending on their competence) query the
database in SQL language
- files
uploaded to the database may be viewed immediately
- advantages
over other audio archives
·
the tool relies almost entirely on open source
code and applications which most people already have on their machines. No one
needs to download and install new software to use the tool. Completely
x-platform.
the dictionary interface can be used as an audio
dictionary
- words in
the dictionary which have corresponding recorded versions in the audio
files can use those segments as audio examples of the word
advantages over other online, audio dictionaries
- no slicing
and dicing. The audio downloads only the minimal segment which contains
the word in question, hence saving bandwidth
- the audio
is in mp3 format, to further enhance bandwidth savings. This is perfectly
acceptable for most linguistic purposes
·
there are no issues with file names, because
there is no direct association between a dictionary entry and a segment of
audio. The audio file names do not have to be updated when/if the dictionary
changes
language learning tool
- the
dictionary/audio-transcript interface used in reverse acts as a language
learning tool
- students
can be presented with texts in any language, play the audio, and look-up
definitions and other examples of words they don't recognise
advantages over other language learning tools
·
the linking tool is not dedicated software, it
can be update continuously, never loses links between the various parts, and
can be used for other purposes besides language teaching. This makes it quite
different to the majority of CALL tools out there at the moment
Compatibility
- At the
moment, the linking tool accepts .mp3 audio files, and transcript and
dictionary files in a given XML format.
- There is a
conversion tool on the admin interface to convert transcripts from
SoundIndex files into the XML format, and to convert standard CSV format
dictionaries into the XML format. We are currently working on allowing
transcript files from CLAN and from Transcriber to be converted and
uploaded, since these seem to be the most commonly used transcription
linking applications in Australia, at least.
- It should
be possible (in the future) to allow conversion of (standard) 'back-slash
coded' dictionary files, which seem to be everywhere. The problem is that
everyone uses their own codes.
- The only
compatibility issue we've come across is that older versions of browsers
without a current media player version that accepts .mp3 format will not
be able to play the audio files.
For the future
- Right now,
Michael is working on a few improvements:
- Ability to
upload data from multiple languages, and possibility of multiple users
with multiple languages, some access issues yet to be resolved
- Addition
of CLAN and Transcriber format transcripts to available types
- Regex
searches on the database
·
Better layout of the dictionary: sliding side bar
with all headwords