Datasets:

bookbot

/

ljspeech_phonemes

like 3

The model is likely to only perform well on the LJSpeech dataset and the phoneme annotation that we used. We don't expect the model to generalize on out-of-domain speech data.

Thanks for reply. I wanna know what should I do to identify the stress mark and long vowel. In short, I wanna recognize the original phoneme detail.Thanks for ur helping.

You might need to make changes and use a better training set that has accurate phoneme annotations.

Thanks!!!

Feel free to reach out to me if you need more help. I'll close this thread for the time being, cheers!

OK!cheers!