VoxForge
Hey Guys...
In this file you can find an acoustic model trained with sphinxtrain, for 8khz recordings.
The WER was 3% and the sentence error rate 14.2% .
Enjoy..
Akane
Good job, nice to see we advance even if slowly.