I was working with deep speech2 to convert speech to text. Data I used is Indian English.
But the Accuracy is not at all good as it doesn’t pick the word right, although is sound or looks somewhat same.
My question is say I have enough data (Indian English ) and I want to train it
will I require to make any other changes in the code or something like vocab file, language model. Or it’ll give me proper or good result with just training.
Thanks in Advance.