blackteay/GPT-SoVITS
1
0
Fork 0
mirror of https://github.com/RVC-Boss/GPT-SoVITS.git synced 2025-10-07 07:02:57 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add inference seed control and optimize ui layout

Browse Source
This commit is contained in:
Miuna 2024-02-08 16:37:32 +08:00
parent 57dde51e11
commit f08041a4d1
1 changed files with 56 additions and 37 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

93
GPT_SoVITS/inference_webui.py
View File

@ -6,7 +6,7 @@
全部按英文识别
全部按日文识别
'''
import os, re, logging
import os, re, logging, random
import LangSegment
logging.getLogger("markdown_it").setLevel(logging.ERROR)
logging.getLogger("urllib3").setLevel(logging.ERROR)
@ -359,6 +359,15 @@ def merge_short_text_in_array(texts, threshold):
result[len(result) - 1] += text
return result
def set_seed(seed):
seed = seed if seed != -1 else random.randrange(1 << 32)
torch.manual_seed(seed)
print("Seed:", seed)
return seed
def reuse_seed(seed):
return seed
def get_tts_wav(ref_wav_path, prompt_text, prompt_language, text, text_language, how_to_cut=i18n("不切")):
t0 = ttime()
prompt_language = dict_language[prompt_language]
@ -587,58 +596,68 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
gr.Markdown(
value=i18n("本软件以MIT协议开源, 作者不对软件具备任何控制力, 使用软件者、传播软件导出的声音者自负全责. <br>如不认可该条款, 则不能使用或引用软件包内任何代码和文件. 详见根目录<b>LICENSE</b>.")
)
with gr.Group():
gr.Markdown(value=i18n("模型切换"))
with gr.Row():
GPT_dropdown = gr.Dropdown(label=i18n("GPT模型列表"), choices=sorted(GPT_names, key=custom_sort_key), value=gpt_path, interactive=True)
SoVITS_dropdown = gr.Dropdown(label=i18n("SoVITS模型列表"), choices=sorted(SoVITS_names, key=custom_sort_key), value=sovits_path, interactive=True)
refresh_button = gr.Button(i18n("刷新模型路径"), variant="primary")
refresh_button.click(fn=change_choices, inputs=[], outputs=[SoVITS_dropdown, GPT_dropdown])
SoVITS_dropdown.change(change_sovits_weights, [SoVITS_dropdown], [])
GPT_dropdown.change(change_gpt_weights, [GPT_dropdown], [])
gr.Markdown(value=i18n("*请上传并填写参考信息"))
with gr.Row():
inp_ref = gr.Audio(label=i18n("请上传3~10秒内参考音频，超过会报错！"), type="filepath")
prompt_text = gr.Textbox(label=i18n("参考音频的文本"), value="")
prompt_language = gr.Dropdown(
label=i18n("参考音频的语种"), choices=[i18n("中文"), i18n("英文"), i18n("日文"), i18n("中英混合"), i18n("日英混合"), i18n("多语种混合")], value=i18n("中文")
)
gr.Markdown(value=i18n("*请填写需要合成的目标文本。中英混合选中文，日英混合选日文，中日混合暂不支持，非目标语言文本自动遗弃。"))
with gr.Row():
text = gr.Textbox(label=i18n("需要合成的文本"), value="")
gr.Markdown(value=i18n("模型切换"))
with gr.Row():
GPT_dropdown = gr.Dropdown(label=i18n("GPT模型列表"), choices=sorted(GPT_names, key=custom_sort_key), value=gpt_path, interactive=True)
SoVITS_dropdown = gr.Dropdown(label=i18n("SoVITS模型列表"), choices=sorted(SoVITS_names, key=custom_sort_key), value=sovits_path, interactive=True)
refresh_button = gr.Button(i18n("刷新模型路径"), variant="primary")
refresh_button.click(fn=change_choices, inputs=[], outputs=[SoVITS_dropdown, GPT_dropdown])
SoVITS_dropdown.change(change_sovits_weights, [SoVITS_dropdown], [])
GPT_dropdown.change(change_gpt_weights, [GPT_dropdown], [])
gr.Markdown(value=i18n("*请上传并填写参考信息"))
with gr.Row():
inp_ref = gr.Audio(label=i18n("请上传3~10秒内参考音频，超过会报错！"), type="filepath")
prompt_text = gr.Textbox(label=i18n("参考音频的文本"), value="")
prompt_language = gr.Dropdown(
label=i18n("参考音频的语种"), choices=[i18n("中文"), i18n("英文"), i18n("日文"), i18n("中英混合"), i18n("日英混合"), i18n("多语种混合")], value=i18n("中文")
)
gr.Markdown(value=i18n("*请填写需要合成的目标文本。中英混合选中文，日英混合选日文，中日混合暂不支持，非目标语言文本自动遗弃。"))
with gr.Row():
with gr.Column():
text = gr.Textbox(label=i18n("需要合成的文本"), value="", lines=5)
text_language = gr.Dropdown(
label=i18n("需要合成的语种"), choices=[i18n("中文"), i18n("英文"), i18n("日文"), i18n("中英混合"), i18n("日英混合"), i18n("多语种混合")], value=i18n("中文")
)
with gr.Column():
how_to_cut = gr.Radio(
label=i18n("怎么切"),
choices=[i18n("不切"), i18n("凑四句一切"), i18n("凑50字一切"), i18n("按中文句号。切"), i18n("按英文句号.切"), i18n("按标点符号切"), ],
value=i18n("凑四句一切"),
interactive=True,
)
with gr.Row():
seed = gr.Number(label=i18n("种子"), value=-1, precision=0)
last_seed = gr.State(value=-1)
reuse_button = gr.Button(value=i18n("种子复用"))
with gr.Column():
inference_button = gr.Button(i18n("合成语音"), variant="primary")
output = gr.Audio(label=i18n("输出的语音"))
inference_button.click(
inference_button.click(set_seed, seed, last_seed).then(
get_tts_wav,
[inp_ref, prompt_text, prompt_language, text, text_language, how_to_cut],
[output],
)
reuse_button.click(reuse_seed, last_seed, seed)
gr.Markdown(value=i18n("文本切分工具。太长的文本合成出来效果不一定好，所以太长建议先切。合成会根据文本的换行分开合成再拼起来。"))
with gr.Row():
text_inp = gr.Textbox(label=i18n("需要合成的切分前文本"), value="")
button1 = gr.Button(i18n("凑四句一切"), variant="primary")
button2 = gr.Button(i18n("凑50字一切"), variant="primary")
button3 = gr.Button(i18n("按中文句号。切"), variant="primary")
button4 = gr.Button(i18n("按英文句号.切"), variant="primary")
button5 = gr.Button(i18n("按标点符号切"), variant="primary")
text_opt = gr.Textbox(label=i18n("切分后文本"), value="")
button1.click(cut1, [text_inp], [text_opt])
button2.click(cut2, [text_inp], [text_opt])
button3.click(cut3, [text_inp], [text_opt])
button4.click(cut4, [text_inp], [text_opt])
button5.click(cut5, [text_inp], [text_opt])
gr.Markdown(value=i18n("后续将支持混合语种编码文本输入。"))
gr.Markdown(value=i18n("文本切分工具。太长的文本合成出来效果不一定好，所以太长建议先切。合成会根据文本的换行分开合成再拼起来。"))
with gr.Row():
text_inp = gr.Textbox(label=i18n("需要合成的切分前文本"), value="")
button1 = gr.Button(i18n("凑四句一切"), variant="primary")
button2 = gr.Button(i18n("凑50字一切"), variant="primary")
button3 = gr.Button(i18n("按中文句号。切"), variant="primary")
button4 = gr.Button(i18n("按英文句号.切"), variant="primary")
button5 = gr.Button(i18n("按标点符号切"), variant="primary")
text_opt = gr.Textbox(label=i18n("切分后文本"), value="")
button1.click(cut1, [text_inp], [text_opt])
button2.click(cut2, [text_inp], [text_opt])
button3.click(cut3, [text_inp], [text_opt])
button4.click(cut4, [text_inp], [text_opt])
button5.click(cut5, [text_inp], [text_opt])
gr.Markdown(value=i18n("后续将支持混合语种编码文本输入。"))
app.queue(concurrency_count=511, max_size=1022).launch(
server_name="0.0.0.0",