VoxForge
VoxForge
There are over 40 hours of MP3 audio with transcription here:
https://stackoverflow.fogbugz.com/default.asp?W4
The transcription was created by volunteer listeners with the intention of making the podcasts available to people who can't listen to them, and making them appear in search engines.
Is data like this useful to VoxForge?
--- (Edited on 2/10/2009 5:48 pm [GMT-0600] by rjmunro) ---
Hi rjmunro,
>There are over 40 hours of MP3 audio with transcription here:
>Is data like this useful to VoxForge?
Yes, especially if there are transcriptions of the speech recordings.
Although we are currently focusing on collecting uncompressed audio, transcribed compressed audio (such as those recorded with the MP3 codec) is also useful in creating acoustic models.
Thanks for letting us know about this resource,
Ken
--- (Edited on 2/12/2009 11:28 am [GMT-0500] by kmaclean) ---