Hi , The latest update for deepspeech0.6.0 works fine when I experimented in different aspects.
I have a requirement my ASR model need to concentrate on particular 100 words. For example the model should not recognise ‘rose’ instead it should be ‘rows’ all time.
I read some articles about generating binary.lm & trie file for the custom words.
Can I able to useful links to follow on to complete this requirement?
Just follow data/lm/README.rst ?
May I know uses of binary.lm,trie and alphabet.txt files with respect to the model??
This is all documented …