Speech recognition

Amedee Van Gasse amedee-ubuntu at amedee.be
Sat Jun 12 20:06:51 UTC 2010


I'm looking for speech recognition software that has the following 
properties:

* work without a gui, no user interaction, scriptable
* accept a WAV as input (other formats are fine, I can do conversions)
* send the text to stdout or file.

My use case:
1) the answering machine of my SIP provider sends WAV files to my email 
address
2) A combination of procmail and munpack save the WAV files on the 
filesystem
2a) Convert the WAv files to any other convenient audio format.
3) A script runs software SpeechFoo, that creates a text file. This is 
the transcript of the audio file.
4) The transcript, together with the metadata of the original call 
(time, phone number, length,...) is sent as a new email.

So far I have been able to do 1, 2, 2a and 4. I have searched for 3, but 
all I can find for Linux are GUI apps. And this is supposed to run on a 
headless server (without X).

-- 
Amedee




More information about the ubuntu-users mailing list