About the stress
#2
by
jz703
- opened
I just want to know that if the model can identify the stress. I set ignore_stress = False, but the predict result won't have any stress mark. I just want to know if I have missed anything here.
Hi
@jz703
, you're right, the prediction result won't have any stress marks. The model's vocabulary doesn't include phonemes with stress marks. The decode_phonemes function just happens to have ignore_stress parameter from another of our internal phoneme model, sorry for the confusion!
Thank you for the clarification. Cheers.
jz703
changed discussion status to
closed