deepgeek/CosyVoice
2
0
Fork 0
mirror of https://github.com/FunAudioLLM/CosyVoice.git synced 2026-02-04 17:39:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix inference_instruct2 speaker ID bug

Browse Source
This commit is contained in:
bearlu
2025-04-21 09:26:34 -07:00
parent e97cd1b655
commit 587604b2b4
3 changed files with 41 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
cosyvoice/cli/cosyvoice.py
View File

@@ -177,10 +177,10 @@ class CosyVoice2(CosyVoice):
def inference_instruct(self, *args, **kwargs):
raise NotImplementedError('inference_instruct is not implemented for CosyVoice2!')
def inference_instruct2(self, tts_text, instruct_text, prompt_speech_16k, stream=False, speed=1.0, text_frontend=True):
def inference_instruct2(self, tts_text, instruct_text, prompt_speech_16k, zero_shot_spk_id='', stream=False, speed=1.0, text_frontend=True):
assert isinstance(self.model, CosyVoice2Model), 'inference_instruct2 is only implemented for CosyVoice2!'
for i in tqdm(self.frontend.text_normalize(tts_text, split=True, text_frontend=text_frontend)):
model_input = self.frontend.frontend_instruct2(i, instruct_text, prompt_speech_16k, self.sample_rate)
model_input = self.frontend.frontend_instruct2(i, instruct_text, prompt_speech_16k, self.sample_rate, zero_shot_spk_id)
start_time = time.time()
logging.info('synthesis text {}'.format(i))
for model_output in self.model.tts(**model_input, stream=stream, speed=speed):

4
cosyvoice/cli/frontend.py
View File

@@ -196,8 +196,8 @@ class CosyVoiceFrontEnd:
model_input['prompt_text_len'] = instruct_text_token_len
return model_input
def frontend_instruct2(self, tts_text, instruct_text, prompt_speech_16k, resample_rate):
model_input = self.frontend_zero_shot(tts_text, instruct_text + '<|endofprompt|>', prompt_speech_16k, resample_rate)
def frontend_instruct2(self, tts_text, instruct_text, prompt_speech_16k, resample_rate, zero_shot_spk_id):
model_input = self.frontend_zero_shot(tts_text, instruct_text + '<|endofprompt|>', prompt_speech_16k, resample_rate, zero_shot_spk_id)
del model_input['llm_prompt_speech_token']
del model_input['llm_prompt_speech_token_len']
return model_input