Pascale and Linking timecoded audio to other resources

[David Nathan]

Hi Pascale

I wrote to you separately, but I reply here to your "public" version

>I still find it unclear how MP3 conversion (which to me seems to be a way
>of reducing the size of a file without affecting quality/time codes etc.,
>nothing more) or audio streaming (which seems to make a single large file
>of all data to enable searching etc.) actually solve the issue of how to
>locate a few seconds of material within one audio file (whatever format).


You are right. What you actually need to do is to identify the language
content within the sound file and then store the data that identifies its
time extent. In other words, you are not making a link, you are creating
the data which underlies a link, or allows some application to implemnt the
link. As for applications, there are many, eg Quicktime, SMIL, web browser,
multimedia authoring software (eg Director), Transcriber, and others - each
of these would need (i) the timing data to be available and structured or
encoded in a particular way and (ii) the sound file to be in a particular
format. It is only for the latter that MP3 is relevant as part of a
solution, because formats like MP3 and Quicktime are technically much more
suited to streaming and time cuing, but this is not germane to your problem.

>Is it possible at all?
Of course it's possible. Lots of people do it all the time.

>Does it require some small program to be written?
Of course! Typically, of course, you use a program that someone else has
already created. I mentioned some of them above. You might need to find out
how to convert/restructure your timing data, but that is not fundamentally
important, what is important is that you represent your timing data in a
way which is robust, standard, and CAN be restructured.

>Or is the only real answer to segment/extract each of those small seconds
>of material for document linking purposes?

No. Although people have effectively done this for a long time, it is now
unnecessary and a weaker solution.


>I'd love to get a clear answer to this question (which I perhaps didn't
>phrase succinctly in the first place) and then hopefully I can stop
>bugging you all! : )


Understand the distinction between representing the relevant linguistic
data and later implementing it as some software/interface, and all should
be clear.

best regards

David