netease-youdao / EmotiVoice

EmotiVoice 😊: a Multi-Voice and Prompt-Controlled TTS Engine
Apache License 2.0
7.46k stars 633 forks source link

Problem with the last phoneme not sounding in some cases #71

Closed clcarwin closed 11 months ago

clcarwin commented 11 months ago

I have found that some word do not pronounce if it in the end of the sentence. Can we have a method to work around this problem?

8051|Happy|[L] [AH1] [V] engsp1 [M] [IY1]|love me output_loveme.zip

11614|Happy|[HH] [AH0] [L] [OW1] engsp4 [K] [AE1] [N] engsp1 [Y] [UW1] engsp1 [HH] [EH1] [L] [P] engsp1 [M] [IY1]|hello, can you help me? output_hellocanyouhelpme.zip

8051|Happy|[AY1] engsp1 [AE1] [M] engsp1 [B] [EH1] [S] [T] engsp4 [TH] [AE1] [NG] [K] engsp1 [Y] [UW1]|i am best, thank you. output_iambestthankyou.zip

syq163 commented 11 months ago

I have confirmed this issue. It appears that the phonemes are not accurate. Could you please explain how you generated the phonemes?

clcarwin commented 11 months ago

Following the instructions in README.md, the phonemes have been generated by python frontend_en.py data/my_text.txt > data/my_text_for_tts.txt

syq163 commented 11 months ago

Following the instructions in README.md, the phonemes have been generated by python frontend_en.py data/my_text.txt > data/my_text_for_tts.txt

The issue has been identified and the code has been updated! Please make sure to update the code and check if it is functioning properly now. The problem was caused by the missing "sos/eos" in the phonemes.

clcarwin commented 11 months ago

It works now. Thanks.