Re: [Trans-devel] Transcriber Next Gen

[Alan S. <ast...@gm...>]

Hi Karim,

I don't know much about portaudio or libavformat but do know that they
are used by lots of open source programs that I also use. So using them
makes a lot of sense to me given that Snack seems to have run its course
and these other tools are embedded in a lot of other very active and
popular open source projects. Glad to hear end-users will still have the
ability to create export filters. I hope it isn't too difficult--my
programming skills aren't very sophisticated! I'd be happy to help test
the betas when they are available. At the moment I'm running several
boxes with Windows XP and OpenSuse 11.

Note that I think the time drift issue in the current version is an
issue with Snack. There have been a number of threads on this issue in
Trans-devel. In my testing it only occurs with MP3 files. WAV and OGG
files work fine. As I had a lot of data in the MP3 format that had been
transcribed before I realized there was an issue, I created a work
around by modifying my export filters (see details by following the link
I provided earlier). I just set up the filter to adjust the times on
transcriptions created from MP3 files by a factor of 0.9953. This
allowed my to keep the transcript and audio properly synchronized when I
imported the data into another program like NVivo 8.

Cheers, Alan.

Karim Boudahmane wrote:
> Alan,
>
> Mathieu has left the team since end of June and I am now in charge of
> Transcriber (especially the new one).
> I have answers to your questions. Sorry for the delay, I had to check
> some points with Bertin Technologies.
>
> 3. The new version of Transcriber is developed from scratch. Our
> purpose was to make a fresh start. As you wrote, development on Snack
> seems to have stopped, Tcl/Tk community is more and more weak... In
> this context it was very tricky to make Transcriber evolve. The
> redevelopment is not really related to Quaero, it is mainly a question
> of durability of the software.
>
> 1. The new version will use libavformat and portaudio to support
> audio. I have not enough detachment to answer to the second point of
> your question. I will check the drift you mentioned.
>
> 2. With the new version of Transcriber, end-users will still have the
> ability to create export filters. To do so, he just has to create a
> derived class from our I/O interface and to override load and store
> methods. But the new internal representation (annotation graph)
> imposes to be more rigorous in the development of these filters,
> because you have to browse a graph (and it is not the same as parsing
> a XML file).
>
> I hope, my answers help you. I you need more informations, send me
> back an email. Thank you for the interest you have for this project
>
> Karim