|
It seems that all English speech recognition datasets are non-free, starting with the 1986 TIMIT ($250) and ending with the more recent ones that cost about a grand. This is probably a non-issue for speech researchers, but it's an unpleasant surprise for hobbyists. Are there any free speech-recognition datasets? |
|
You Can check on Voxforge. http://voxforge.org/home/listen |
|
There is also the TED-LIUM corpus: http://www-lium.univ-lemans.fr/en/content/ted-lium-corpus |
Which zones are closest to TIMIT? I mean which ones are free and we can expect to test methods published on TIMIT?