deepgeek/CosyVoice
2
0
Fork 0
mirror of https://github.com/FunAudioLLM/CosyVoice.git synced 2026-02-04 09:29:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

add_zero_shot_spk

Browse Source
This commit is contained in:
lyuxiang.lx
2025-04-15 11:45:00 +08:00
parent 9dc559fc2a
commit 9fea0f0836
3 changed files with 38 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
cosyvoice/cli/cosyvoice.py
View File

@@ -66,6 +66,14 @@ class CosyVoice:
spks = list(self.frontend.spk2info.keys())
return spks
def add_zero_shot_spk(self, prompt_text, prompt_speech_16k, zero_shot_spk_id):
assert zero_shot_spk_id != '', 'do not use empty zero_shot_spk_id'
model_input = self.frontend.frontend_zero_shot('', prompt_text, prompt_speech_16k, self.sample_rate, '')
del model_input['text']
del model_input['text_len']
self.frontend.spk2info[zero_shot_spk_id] = model_input
return True
def inference_sft(self, tts_text, spk_id, stream=False, speed=1.0, text_frontend=True):
for i in tqdm(self.frontend.text_normalize(tts_text, split=True, text_frontend=text_frontend)):
model_input = self.frontend.frontend_sft(i, spk_id)
@@ -77,12 +85,12 @@ class CosyVoice:
yield model_output
start_time = time.time()
def inference_zero_shot(self, tts_text, prompt_text, prompt_speech_16k, stream=False, speed=1.0, text_frontend=True):
def inference_zero_shot(self, tts_text, prompt_text, prompt_speech_16k, zero_shot_spk_id='', stream=False, speed=1.0, text_frontend=True):
prompt_text = self.frontend.text_normalize(prompt_text, split=False, text_frontend=text_frontend)
for i in tqdm(self.frontend.text_normalize(tts_text, split=True, text_frontend=text_frontend)):
if (not isinstance(i, Generator)) and len(i) < 0.5 * len(prompt_text):
logging.warning('synthesis text {} too short than prompt text {}, this may lead to bad performance'.format(i, prompt_text))
model_input = self.frontend.frontend_zero_shot(i, prompt_text, prompt_speech_16k, self.sample_rate)
model_input = self.frontend.frontend_zero_shot(i, prompt_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):