search

MontrealCorpusTools / mfa-models

Collection of pretrained models for the Montreal Forced Aligner
Creative Commons Attribution 4.0 International
103 stars 19 forks source link

For Mandarin alignment, are the PINYIN phonemes still functioning? #6

Open GalaxieT opened 2 years ago

GalaxieT commented 2 years ago

So great for the new pretrained acoustic models! Noticed that the new models now accept the IPA-like phonemes, I wonder whether it still works if I take pinyin instead of Chinese characters as the input labels for alignment using the pretrained v2.0 models.

GalaxieT commented 2 years ago

image I tried but it looks like all the pinyin are recognized as 'spn', although I used the mandarin_pinyin dictionary.

The input labeling file is a .txt file with its content: jin1 tian2 de5 tian1 qi5 hai2 shi4 ting3 bu2 cuo3 de5

GalaxieT commented 2 years ago

Found the problem. The Mandarin PINYIN (instead of MFA) acoustic model was missing. Wondering if that could be released.

GalaxieT commented 2 years ago

Partly solved this problem by making a pinyin to mfa phone-set dictionary: MFA-mandarin-pinyin-dict-for-pretrained-mfa-model-v2.0