tuna2134/sbv2-v2
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Merge pull request #214 from jiangyuxiaoxiao/master

Browse Source 
fix server_fastapi.py: 修复bug。
This commit is contained in:
Sora
2023-12-04 12:11:12 +08:00
committed by GitHub
parent 67de3baba2 2b178ba183
commit 15603dcf3f
1 changed files with 32 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

62
server_fastapi.py
View File

@@ -214,11 +214,34 @@ if __name__ == "__main__":
if auto_translate: if auto_translate:
text = trans.translate(Sentence=text, to_Language=language.lower()) text = trans.translate(Sentence=text, to_Language=language.lower())
if reference_audio is not None: if reference_audio is not None:
with BytesIO(await reference_audio.read()) as ref_audio: ref_audio = BytesIO(await reference_audio.read())
if not auto_split: else:
with torch.no_grad(): ref_audio = reference_audio
audio = infer( if not auto_split:
text=text, with torch.no_grad():
audio = infer(
text=text,
sdp_ratio=sdp_ratio,
noise_scale=noise,
noise_scale_w=noisew,
length_scale=length,
sid=speaker_name,
language=language,
hps=loaded_models.models[model_id].hps,
net_g=loaded_models.models[model_id].net_g,
device=loaded_models.models[model_id].device,
emotion=emotion,
reference_audio=ref_audio,
)
audio = gradio.processing_utils.convert_to_16_bit_wav(audio)
else:
texts = cut_sent(text)
audios = []
with torch.no_grad():
for t in texts:
audios.append(
infer(
text=t,
sdp_ratio=sdp_ratio, sdp_ratio=sdp_ratio,
noise_scale=noise, noise_scale=noise,
noise_scale_w=noisew, noise_scale_w=noisew,
@@ -231,31 +254,10 @@ if __name__ == "__main__":
emotion=emotion, emotion=emotion,
reference_audio=ref_audio, reference_audio=ref_audio,
) )
audio = gradio.processing_utils.convert_to_16_bit_wav(audio) )
else: audios.append(np.zeros(int(44100 * 0.2)))
texts = cut_sent(text) audio = np.concatenate(audios)
audios = [] audio = gradio.processing_utils.convert_to_16_bit_wav(audio)
with torch.no_grad():
for t in texts:
audios.append(
infer(
text=t,
sdp_ratio=sdp_ratio,
noise_scale=noise,
noise_scale_w=noisew,
length_scale=length,
sid=speaker_name,
language=language,
hps=loaded_models.models[model_id].hps,
net_g=loaded_models.models[model_id].net_g,
device=loaded_models.models[model_id].device,
emotion=emotion,
reference_audio=ref_audio,
)
)
audios.append(np.zeros(int(44100 * 0.2)))
audio = np.concatenate(audios)
audio = gradio.processing_utils.convert_to_16_bit_wav(audio)
with BytesIO() as wavContent: with BytesIO() as wavContent:
wavfile.write( wavfile.write(
wavContent, loaded_models.models[model_id].hps.data.sampling_rate, audio wavContent, loaded_models.models[model_id].hps.data.sampling_rate, audio