mirror of
https://github.com/RVC-Boss/GPT-SoVITS.git
synced 2025-10-08 16:00:01 +08:00
重构了webui，提供了另一种风格？
This commit is contained in:
XTer
parent
ccf832c4d8
commit
3d65166b35
@ -26,6 +26,7 @@ webui_port_main = 9874
|
||||
webui_port_uvr5 = 9873
|
||||
webui_port_infer_tts = 9872
|
||||
webui_port_subfix = 9871
|
||||
webui_port_srt_slicer = 9870
|
||||
|
||||
api_port = 9880
|
||||
|
||||
@ -62,5 +63,6 @@ class Config:
|
||||
self.webui_port_uvr5 = webui_port_uvr5
|
||||
self.webui_port_infer_tts = webui_port_infer_tts
|
||||
self.webui_port_subfix = webui_port_subfix
|
||||
self.webui_port_srt_slicer = webui_port_srt_slicer
|
||||
|
||||
self.api_port = api_port
|
||||
|
||||
@ -15,6 +15,11 @@ from srt_utils import (
|
||||
merge_list_folders
|
||||
)
|
||||
|
||||
port = 8991
|
||||
|
||||
if len(sys.argv) > 1:
|
||||
port = int(sys.argv[1])
|
||||
|
||||
|
||||
from i18n.i18n import I18nAuto
|
||||
|
||||
@ -371,4 +376,4 @@ with gr.Blocks() as app:
|
||||
[save_folder, character],
|
||||
[character_warning],
|
||||
)
|
||||
app.launch(inbrowser=True, server_port=8991, debug=True)
|
||||
app.launch(inbrowser=True, server_port=port, debug=True)
|
||||
|
||||
396
webui.py
396
webui.py
@ -48,7 +48,7 @@ import pdb
|
||||
import gradio as gr
|
||||
from subprocess import Popen
|
||||
import signal
|
||||
from config import python_exec,infer_device,is_half,exp_root,webui_port_main,webui_port_infer_tts,webui_port_uvr5,webui_port_subfix,is_share
|
||||
from config import python_exec,infer_device,is_half,exp_root,webui_port_main,webui_port_infer_tts,webui_port_uvr5,webui_port_subfix,is_share,webui_port_srt_slicer
|
||||
from tools.i18n.i18n import I18nAuto
|
||||
i18n = I18nAuto()
|
||||
from scipy.io import wavfile
|
||||
@ -120,6 +120,7 @@ p_uvr5=None
|
||||
p_asr=None
|
||||
p_denoise=None
|
||||
p_tts_inference=None
|
||||
p_srt_slicer=None
|
||||
|
||||
def kill_proc_tree(pid, including_parent=True):
|
||||
try:
|
||||
@ -162,6 +163,18 @@ def change_label(if_label,path_list):
|
||||
p_label=None
|
||||
yield i18n("打标工具WebUI已关闭")
|
||||
|
||||
def change_srt_slicer(if_srt_slicer):
|
||||
global p_srt_slicer
|
||||
if(if_srt_slicer==True and p_srt_slicer==None):
|
||||
cmd = '"%s" tools/srt_slicer/webui.py %s'%(python_exec,webui_port_srt_slicer)
|
||||
yield i18n("SRT切割工具WebUI已开启")
|
||||
print(cmd)
|
||||
p_srt_slicer = Popen(cmd, shell=True)
|
||||
elif(if_srt_slicer==False and p_srt_slicer!=None):
|
||||
kill_process(p_srt_slicer.pid)
|
||||
p_srt_slicer=None
|
||||
yield i18n("SRT切割工具WebUI已关闭")
|
||||
|
||||
def change_uvr5(if_uvr5):
|
||||
global p_uvr5
|
||||
if(if_uvr5==True and p_uvr5==None):
|
||||
@ -680,143 +693,168 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
|
||||
i18n("中文教程文档：https://www.yuque.com/baicaigongchang1145haoyuangong/ib3g1e")
|
||||
)
|
||||
|
||||
# 重构前端WebUI的展示方式，by XTer: https://github.com/X-T-E-R
|
||||
with gr.Tabs():
|
||||
with gr.TabItem(i18n("0-前置数据集获取工具")):#提前随机切片防止uvr5爆内存->uvr5->slicer->asr->打标
|
||||
gr.Markdown(value=i18n("0a-UVR5人声伴奏分离&去混响去延迟工具"))
|
||||
with gr.Tab(i18n("0-前置数据集获取工具")):#提前随机切片防止uvr5爆内存->uvr5->slicer->asr->打标
|
||||
with gr.Row():
|
||||
if_uvr5 = gr.Checkbox(label=i18n("是否开启UVR5-WebUI"),show_label=True)
|
||||
uvr5_info = gr.Textbox(label=i18n("UVR5进程输出信息"))
|
||||
gr.Markdown(value=i18n("0b-语音切分工具"))
|
||||
with gr.Row():
|
||||
with gr.Row():
|
||||
slice_inp_path=gr.Textbox(label=i18n("音频自动切分输入路径，可文件可文件夹"),value="")
|
||||
slice_opt_root=gr.Textbox(label=i18n("切分后的子音频的输出根目录"),value="output/slicer_opt")
|
||||
threshold=gr.Textbox(label=i18n("threshold:音量小于这个值视作静音的备选切割点"),value="-34")
|
||||
min_length=gr.Textbox(label=i18n("min_length:每段最小多长，如果第一段太短一直和后面段连起来直到超过这个值"),value="4000")
|
||||
min_interval=gr.Textbox(label=i18n("min_interval:最短切割间隔"),value="300")
|
||||
hop_size=gr.Textbox(label=i18n("hop_size:怎么算音量曲线，越小精度越大计算量越高（不是精度越大效果越好）"),value="10")
|
||||
max_sil_kept=gr.Textbox(label=i18n("max_sil_kept:切完后静音最多留多长"),value="500")
|
||||
with gr.Row():
|
||||
open_slicer_button=gr.Button(i18n("开启语音切割"), variant="primary",visible=True)
|
||||
close_slicer_button=gr.Button(i18n("终止语音切割"), variant="primary",visible=False)
|
||||
_max=gr.Slider(minimum=0,maximum=1,step=0.05,label=i18n("max:归一化后最大值多少"),value=0.9,interactive=True)
|
||||
alpha=gr.Slider(minimum=0,maximum=1,step=0.05,label=i18n("alpha_mix:混多少比例归一化后音频进来"),value=0.25,interactive=True)
|
||||
n_process=gr.Slider(minimum=1,maximum=n_cpu,step=1,label=i18n("切割使用的进程数"),value=4,interactive=True)
|
||||
slicer_info = gr.Textbox(label=i18n("语音切割进程输出信息"))
|
||||
gr.Markdown(value=i18n("0bb-语音降噪工具"))
|
||||
with gr.Row():
|
||||
open_denoise_button = gr.Button(i18n("开启语音降噪"), variant="primary",visible=True)
|
||||
close_denoise_button = gr.Button(i18n("终止语音降噪进程"), variant="primary",visible=False)
|
||||
denoise_input_dir=gr.Textbox(label=i18n("降噪音频文件输入文件夹"),value="")
|
||||
denoise_output_dir=gr.Textbox(label=i18n("降噪结果输出文件夹"),value="output/denoise_opt")
|
||||
denoise_info = gr.Textbox(label=i18n("语音降噪进程输出信息"))
|
||||
gr.Markdown(value=i18n("0c-中文批量离线ASR工具"))
|
||||
with gr.Row():
|
||||
open_asr_button = gr.Button(i18n("开启离线批量ASR"), variant="primary",visible=True)
|
||||
close_asr_button = gr.Button(i18n("终止ASR进程"), variant="primary",visible=False)
|
||||
with gr.Column():
|
||||
with gr.Row():
|
||||
asr_inp_dir = gr.Textbox(
|
||||
label=i18n("输入文件夹路径"),
|
||||
value="D:\\GPT-SoVITS\\raw\\xxx",
|
||||
interactive=True,
|
||||
)
|
||||
asr_opt_dir = gr.Textbox(
|
||||
label = i18n("输出文件夹路径"),
|
||||
value = "output/asr_opt",
|
||||
interactive = True,
|
||||
)
|
||||
with gr.Row():
|
||||
asr_model = gr.Dropdown(
|
||||
label = i18n("ASR 模型"),
|
||||
choices = list(asr_dict.keys()),
|
||||
interactive = True,
|
||||
value="达摩 ASR (中文)"
|
||||
)
|
||||
asr_size = gr.Dropdown(
|
||||
label = i18n("ASR 模型尺寸"),
|
||||
choices = ["large"],
|
||||
interactive = True,
|
||||
value="large"
|
||||
)
|
||||
asr_lang = gr.Dropdown(
|
||||
label = i18n("ASR 语言设置"),
|
||||
choices = ["zh"],
|
||||
interactive = True,
|
||||
value="zh"
|
||||
)
|
||||
with gr.Row():
|
||||
asr_info = gr.Textbox(label=i18n("ASR进程输出信息"))
|
||||
with gr.Column(scale=2):
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("0a-UVR5人声伴奏分离&去混响去延迟工具")):
|
||||
if_uvr5 = gr.Checkbox(label=i18n("是否开启UVR5-WebUI"),show_label=True)
|
||||
uvr5_info = gr.Textbox(label=i18n("UVR5进程输出信息"))
|
||||
with gr.Tab(i18n("0b-语音切分工具")):
|
||||
with gr.Row():
|
||||
slice_inp_path=gr.Textbox(label=i18n("音频自动切分输入路径，可文件可文件夹"),value="")
|
||||
slice_opt_root=gr.Textbox(label=i18n("切分后的子音频的输出根目录"),value="output/slicer_opt")
|
||||
threshold=gr.Textbox(label=i18n("threshold:音量小于这个值视作静音的备选切割点"),value="-34")
|
||||
min_length=gr.Textbox(label=i18n("min_length:每段最小多长，如果第一段太短一直和后面段连起来直到超过这个值"),value="4000")
|
||||
min_interval=gr.Textbox(label=i18n("min_interval:最短切割间隔"),value="300")
|
||||
hop_size=gr.Textbox(label=i18n("hop_size:怎么算音量曲线，越小精度越大计算量越高（不是精度越大效果越好）"),value="10")
|
||||
max_sil_kept=gr.Textbox(label=i18n("max_sil_kept:切完后静音最多留多长"),value="500")
|
||||
with gr.Row():
|
||||
_max=gr.Slider(minimum=0,maximum=1,step=0.05,label=i18n("max:归一化后最大值多少"),value=0.9,interactive=True)
|
||||
alpha=gr.Slider(minimum=0,maximum=1,step=0.05,label=i18n("alpha_mix:混多少比例归一化后音频进来"),value=0.25,interactive=True)
|
||||
n_process=gr.Slider(minimum=1,maximum=n_cpu,step=1,label=i18n("切割使用的进程数"),value=4,interactive=True)
|
||||
with gr.Row():
|
||||
with gr.Group():
|
||||
slicer_info = gr.Textbox(label=i18n("语音切割进程输出信息"))
|
||||
open_slicer_button=gr.Button(i18n("开启语音切割"), variant="primary",visible=True)
|
||||
close_slicer_button=gr.Button(i18n("终止语音切割"), variant="primary",visible=False)
|
||||
with gr.Tab(i18n("0bb-语音降噪工具")):
|
||||
with gr.Row():
|
||||
|
||||
def change_lang_choices(key): #根据选择的模型修改可选的语言
|
||||
# return gr.Dropdown(choices=asr_dict[key]['lang'])
|
||||
return {"__type__": "update", "choices": asr_dict[key]['lang'],"value":asr_dict[key]['lang'][0]}
|
||||
def change_size_choices(key): # 根据选择的模型修改可选的模型尺寸
|
||||
# return gr.Dropdown(choices=asr_dict[key]['size'])
|
||||
return {"__type__": "update", "choices": asr_dict[key]['size']}
|
||||
asr_model.change(change_lang_choices, [asr_model], [asr_lang])
|
||||
asr_model.change(change_size_choices, [asr_model], [asr_size])
|
||||
denoise_input_dir=gr.Textbox(label=i18n("降噪音频文件输入文件夹"),value="")
|
||||
denoise_output_dir=gr.Textbox(label=i18n("降噪结果输出文件夹"),value="output/denoise_opt")
|
||||
with gr.Row():
|
||||
with gr.Group():
|
||||
denoise_info = gr.Textbox(label=i18n("语音降噪进程输出信息"))
|
||||
open_denoise_button = gr.Button(i18n("开启语音降噪"), variant="primary",visible=True)
|
||||
close_denoise_button = gr.Button(i18n("终止语音降噪进程"), variant="primary",visible=False)
|
||||
with gr.Column(scale=2):
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("0c-离线批量ASR工具")):
|
||||
|
||||
gr.Markdown(value=i18n("0d-语音文本校对标注工具"))
|
||||
with gr.Row():
|
||||
if_label = gr.Checkbox(label=i18n("是否开启打标WebUI"),show_label=True)
|
||||
path_list = gr.Textbox(
|
||||
label=i18n(".list标注文件的路径"),
|
||||
value="D:\\RVC1006\\GPT-SoVITS\\raw\\xxx.list",
|
||||
interactive=True,
|
||||
)
|
||||
label_info = gr.Textbox(label=i18n("打标工具进程输出信息"))
|
||||
if_label.change(change_label, [if_label,path_list], [label_info])
|
||||
if_uvr5.change(change_uvr5, [if_uvr5], [uvr5_info])
|
||||
open_asr_button.click(open_asr, [asr_inp_dir, asr_opt_dir, asr_model, asr_size, asr_lang], [asr_info,open_asr_button,close_asr_button])
|
||||
close_asr_button.click(close_asr, [], [asr_info,open_asr_button,close_asr_button])
|
||||
open_slicer_button.click(open_slice, [slice_inp_path,slice_opt_root,threshold,min_length,min_interval,hop_size,max_sil_kept,_max,alpha,n_process], [slicer_info,open_slicer_button,close_slicer_button])
|
||||
close_slicer_button.click(close_slice, [], [slicer_info,open_slicer_button,close_slicer_button])
|
||||
open_denoise_button.click(open_denoise, [denoise_input_dir,denoise_output_dir], [denoise_info,open_denoise_button,close_denoise_button])
|
||||
close_denoise_button.click(close_denoise, [], [denoise_info,open_denoise_button,close_denoise_button])
|
||||
with gr.Column():
|
||||
with gr.Row():
|
||||
asr_inp_dir = gr.Textbox(
|
||||
label=i18n("输入文件夹路径"),
|
||||
value="D:\\GPT-SoVITS\\raw\\xxx",
|
||||
interactive=True,
|
||||
)
|
||||
asr_opt_dir = gr.Textbox(
|
||||
label = i18n("输出文件夹路径"),
|
||||
value = "output/asr_opt",
|
||||
interactive = True,
|
||||
)
|
||||
with gr.Row():
|
||||
asr_model = gr.Dropdown(
|
||||
label = i18n("ASR 模型"),
|
||||
choices = list(asr_dict.keys()),
|
||||
interactive = True,
|
||||
value="达摩 ASR (中文)"
|
||||
)
|
||||
asr_size = gr.Dropdown(
|
||||
label = i18n("ASR 模型尺寸"),
|
||||
choices = ["large"],
|
||||
interactive = True,
|
||||
value="large"
|
||||
)
|
||||
asr_lang = gr.Dropdown(
|
||||
label = i18n("ASR 语言设置"),
|
||||
choices = ["zh"],
|
||||
interactive = True,
|
||||
value="zh"
|
||||
)
|
||||
with gr.Row():
|
||||
with gr.Group():
|
||||
asr_info = gr.Textbox(label=i18n("ASR进程输出信息"))
|
||||
open_asr_button = gr.Button(i18n("开启离线批量ASR"), variant="primary",visible=True)
|
||||
close_asr_button = gr.Button(i18n("终止ASR进程"), variant="primary",visible=False)
|
||||
|
||||
with gr.TabItem(i18n("1-GPT-SoVITS-TTS")):
|
||||
def change_lang_choices(key): #根据选择的模型修改可选的语言
|
||||
# return gr.Dropdown(choices=asr_dict[key]['lang'])
|
||||
return {"__type__": "update", "choices": asr_dict[key]['lang'],"value":asr_dict[key]['lang'][0]}
|
||||
def change_size_choices(key): # 根据选择的模型修改可选的模型尺寸
|
||||
# return gr.Dropdown(choices=asr_dict[key]['size'])
|
||||
return {"__type__": "update", "choices": asr_dict[key]['size']}
|
||||
asr_model.change(change_lang_choices, [asr_model], [asr_lang])
|
||||
asr_model.change(change_size_choices, [asr_model], [asr_size])
|
||||
|
||||
with gr.Tab(i18n("0d-语音文本校对标注工具")):
|
||||
if_label = gr.Checkbox(label=i18n("是否开启打标WebUI"),show_label=True)
|
||||
path_list = gr.Textbox(
|
||||
label=i18n(".list标注文件的路径"),
|
||||
value="D:\\RVC1006\\GPT-SoVITS\\raw\\xxx.list",
|
||||
interactive=True,
|
||||
)
|
||||
label_info = gr.Textbox(label=i18n("打标工具进程输出信息"))
|
||||
|
||||
with gr.Tab(i18n("0c-基于SRT的音频切分工具")):
|
||||
if_srt_slicer = gr.Checkbox(label=i18n("是否开启SRT切分工具"),show_label=True)
|
||||
srt_slicer_info = gr.Textbox(label=i18n("SRT切分工具进程输出信息"))
|
||||
if_label.change(change_label, [if_label,path_list], [label_info])
|
||||
if_srt_slicer.change(change_srt_slicer, [if_srt_slicer], [srt_slicer_info])
|
||||
if_uvr5.change(change_uvr5, [if_uvr5], [uvr5_info])
|
||||
open_asr_button.click(open_asr, [asr_inp_dir, asr_opt_dir, asr_model, asr_size, asr_lang], [asr_info,open_asr_button,close_asr_button])
|
||||
close_asr_button.click(close_asr, [], [asr_info,open_asr_button,close_asr_button])
|
||||
open_slicer_button.click(open_slice, [slice_inp_path,slice_opt_root,threshold,min_length,min_interval,hop_size,max_sil_kept,_max,alpha,n_process], [slicer_info,open_slicer_button,close_slicer_button])
|
||||
close_slicer_button.click(close_slice, [], [slicer_info,open_slicer_button,close_slicer_button])
|
||||
open_denoise_button.click(open_denoise, [denoise_input_dir,denoise_output_dir], [denoise_info,open_denoise_button,close_denoise_button])
|
||||
close_denoise_button.click(close_denoise, [], [denoise_info,open_denoise_button,close_denoise_button])
|
||||
|
||||
with gr.Tab(i18n("1-GPT-SoVITS-TTS")):
|
||||
with gr.Row():
|
||||
exp_name = gr.Textbox(label=i18n("*实验/模型名"), value="xxx", interactive=True)
|
||||
gpu_info = gr.Textbox(label=i18n("显卡信息"), value=gpu_info, visible=True, interactive=False)
|
||||
pretrained_s2G = gr.Textbox(label=i18n("预训练的SoVITS-G模型路径"), value="GPT_SoVITS/pretrained_models/s2G488k.pth", interactive=True)
|
||||
pretrained_s2D = gr.Textbox(label=i18n("预训练的SoVITS-D模型路径"), value="GPT_SoVITS/pretrained_models/s2D488k.pth", interactive=True)
|
||||
pretrained_s1 = gr.Textbox(label=i18n("预训练的GPT模型路径"), value="GPT_SoVITS/pretrained_models/s1bert25hz-2kh-longer-epoch=68e-step=50232.ckpt", interactive=True)
|
||||
with gr.TabItem(i18n("1A-训练集格式化工具")):
|
||||
gr.Markdown(value=i18n("输出logs/实验名目录下应有23456开头的文件和文件夹"))
|
||||
with gr.Tab(i18n("1A-训练集格式化工具")):
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("输出logs/实验名目录下应有23456开头的文件和文件夹")):
|
||||
with gr.Group():
|
||||
inp_text = gr.Textbox(label=i18n("*文本标注文件"),value=r"D:\RVC1006\GPT-SoVITS\raw\xxx.list",interactive=True)
|
||||
inp_wav_dir = gr.Textbox(
|
||||
label=i18n("*训练集音频文件目录"),
|
||||
# value=r"D:\RVC1006\GPT-SoVITS\raw\xxx",
|
||||
interactive=True,
|
||||
placeholder=i18n("填切割后音频所在目录！读取的音频文件完整路径=该目录-拼接-list文件里波形对应的文件名（不是全路径）。如果留空则使用.list文件里的绝对全路径。")
|
||||
)
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("1Aabc-训练集格式化一键三连")):
|
||||
with gr.Group():
|
||||
info1abc=gr.Textbox(label=i18n("一键三连进程输出信息"))
|
||||
button1abc_open = gr.Button(i18n("开启一键三连"), variant="primary",visible=True)
|
||||
button1abc_close = gr.Button(i18n("终止一键三连"), variant="primary",visible=False)
|
||||
with gr.Row():
|
||||
inp_text = gr.Textbox(label=i18n("*文本标注文件"),value=r"D:\RVC1006\GPT-SoVITS\raw\xxx.list",interactive=True)
|
||||
inp_wav_dir = gr.Textbox(
|
||||
label=i18n("*训练集音频文件目录"),
|
||||
# value=r"D:\RVC1006\GPT-SoVITS\raw\xxx",
|
||||
interactive=True,
|
||||
placeholder=i18n("填切割后音频所在目录！读取的音频文件完整路径=该目录-拼接-list文件里波形对应的文件名（不是全路径）。如果留空则使用.list文件里的绝对全路径。")
|
||||
)
|
||||
gr.Markdown(value=i18n("1Aa-文本内容"))
|
||||
with gr.Row():
|
||||
gpu_numbers1a = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"),value="%s-%s"%(gpus,gpus),interactive=True)
|
||||
bert_pretrained_dir = gr.Textbox(label=i18n("预训练的中文BERT模型路径"),value="GPT_SoVITS/pretrained_models/chinese-roberta-wwm-ext-large",interactive=False)
|
||||
button1a_open = gr.Button(i18n("开启文本获取"), variant="primary",visible=True)
|
||||
button1a_close = gr.Button(i18n("终止文本获取进程"), variant="primary",visible=False)
|
||||
info1a=gr.Textbox(label=i18n("文本进程输出信息"))
|
||||
gr.Markdown(value=i18n("1Ab-SSL自监督特征提取"))
|
||||
with gr.Row():
|
||||
gpu_numbers1Ba = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"),value="%s-%s"%(gpus,gpus),interactive=True)
|
||||
cnhubert_base_dir = gr.Textbox(label=i18n("预训练的SSL模型路径"),value="GPT_SoVITS/pretrained_models/chinese-hubert-base",interactive=False)
|
||||
button1b_open = gr.Button(i18n("开启SSL提取"), variant="primary",visible=True)
|
||||
button1b_close = gr.Button(i18n("终止SSL提取进程"), variant="primary",visible=False)
|
||||
info1b=gr.Textbox(label=i18n("SSL进程输出信息"))
|
||||
gr.Markdown(value=i18n("1Ac-语义token提取"))
|
||||
with gr.Row():
|
||||
gpu_numbers1c = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"),value="%s-%s"%(gpus,gpus),interactive=True)
|
||||
button1c_open = gr.Button(i18n("开启语义token提取"), variant="primary",visible=True)
|
||||
button1c_close = gr.Button(i18n("终止语义token提取进程"), variant="primary",visible=False)
|
||||
info1c=gr.Textbox(label=i18n("语义token提取进程输出信息"))
|
||||
gr.Markdown(value=i18n("1Aabc-训练集格式化一键三连"))
|
||||
with gr.Row():
|
||||
button1abc_open = gr.Button(i18n("开启一键三连"), variant="primary",visible=True)
|
||||
button1abc_close = gr.Button(i18n("终止一键三连"), variant="primary",visible=False)
|
||||
info1abc=gr.Textbox(label=i18n("一键三连进程输出信息"))
|
||||
with gr.Column():
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("1Aa-文本内容")):
|
||||
gpu_numbers1a = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"),value="%s-%s"%(gpus,gpus),interactive=True)
|
||||
bert_pretrained_dir = gr.Textbox(label=i18n("预训练的中文BERT模型路径"),value="GPT_SoVITS/pretrained_models/chinese-roberta-wwm-ext-large",interactive=False)
|
||||
with gr.Group():
|
||||
info1a=gr.Textbox(label=i18n("文本进程输出信息"))
|
||||
button1a_open = gr.Button(i18n("开启文本获取"), variant="primary",visible=True)
|
||||
button1a_close = gr.Button(i18n("终止文本获取进程"), variant="primary",visible=False)
|
||||
with gr.Column():
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("1Ab-SSL自监督特征提取")):
|
||||
gpu_numbers1Ba = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"),value="%s-%s"%(gpus,gpus),interactive=True)
|
||||
cnhubert_base_dir = gr.Textbox(label=i18n("预训练的SSL模型路径"),value="GPT_SoVITS/pretrained_models/chinese-hubert-base",interactive=False)
|
||||
with gr.Group():
|
||||
info1b=gr.Textbox(label=i18n("SSL进程输出信息"))
|
||||
button1b_open = gr.Button(i18n("开启SSL提取"), variant="primary",visible=True)
|
||||
button1b_close = gr.Button(i18n("终止SSL提取进程"), variant="primary",visible=False)
|
||||
with gr.Column():
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("1Ac-语义token提取")):
|
||||
|
||||
gpu_numbers1c = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"),value="%s-%s"%(gpus,gpus),interactive=True)
|
||||
with gr.Group():
|
||||
info1c=gr.Textbox(label=i18n("语义token提取进程输出信息"))
|
||||
button1c_open = gr.Button(i18n("开启语义token提取"), variant="primary",visible=True)
|
||||
button1c_close = gr.Button(i18n("终止语义token提取进程"), variant="primary",visible=False)
|
||||
|
||||
button1a_open.click(open1a, [inp_text,inp_wav_dir,exp_name,gpu_numbers1a,bert_pretrained_dir], [info1a,button1a_open,button1a_close])
|
||||
button1a_close.click(close1a, [], [info1a,button1a_open,button1a_close])
|
||||
button1b_open.click(open1b, [inp_text,inp_wav_dir,exp_name,gpu_numbers1Ba,cnhubert_base_dir], [info1b,button1b_open,button1b_close])
|
||||
@ -825,54 +863,72 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
|
||||
button1c_close.click(close1c, [], [info1c,button1c_open,button1c_close])
|
||||
button1abc_open.click(open1abc, [inp_text,inp_wav_dir,exp_name,gpu_numbers1a,gpu_numbers1Ba,gpu_numbers1c,bert_pretrained_dir,cnhubert_base_dir,pretrained_s2G], [info1abc,button1abc_open,button1abc_close])
|
||||
button1abc_close.click(close1abc, [], [info1abc,button1abc_open,button1abc_close])
|
||||
with gr.TabItem(i18n("1B-微调训练")):
|
||||
gr.Markdown(value=i18n("1Ba-SoVITS训练。用于分享的模型文件输出在SoVITS_weights下。"))
|
||||
with gr.Tab(i18n("1B-微调训练")):
|
||||
with gr.Row():
|
||||
batch_size = gr.Slider(minimum=1,maximum=40,step=1,label=i18n("每张显卡的batch_size"),value=default_batch_size,interactive=True)
|
||||
total_epoch = gr.Slider(minimum=1,maximum=25,step=1,label=i18n("总训练轮数total_epoch，不建议太高"),value=8,interactive=True)
|
||||
text_low_lr_rate = gr.Slider(minimum=0.2,maximum=0.6,step=0.05,label=i18n("文本模块学习率权重"),value=0.4,interactive=True)
|
||||
save_every_epoch = gr.Slider(minimum=1,maximum=25,step=1,label=i18n("保存频率save_every_epoch"),value=4,interactive=True)
|
||||
if_save_latest = gr.Checkbox(label=i18n("是否仅保存最新的ckpt文件以节省硬盘空间"), value=True, interactive=True, show_label=True)
|
||||
if_save_every_weights = gr.Checkbox(label=i18n("是否在每次保存时间点将最终小模型保存至weights文件夹"), value=True, interactive=True, show_label=True)
|
||||
gpu_numbers1Ba = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"), value="%s" % (gpus), interactive=True)
|
||||
with gr.Row():
|
||||
button1Ba_open = gr.Button(i18n("开启SoVITS训练"), variant="primary",visible=True)
|
||||
button1Ba_close = gr.Button(i18n("终止SoVITS训练"), variant="primary",visible=False)
|
||||
info1Ba=gr.Textbox(label=i18n("SoVITS训练进程输出信息"))
|
||||
gr.Markdown(value=i18n("1Bb-GPT训练。用于分享的模型文件输出在GPT_weights下。"))
|
||||
with gr.Row():
|
||||
batch_size1Bb = gr.Slider(minimum=1,maximum=40,step=1,label=i18n("每张显卡的batch_size"),value=default_batch_size,interactive=True)
|
||||
total_epoch1Bb = gr.Slider(minimum=2,maximum=50,step=1,label=i18n("总训练轮数total_epoch"),value=15,interactive=True)
|
||||
if_dpo = gr.Checkbox(label=i18n("是否开启dpo训练选项(实验性)"), value=False, interactive=True, show_label=True)
|
||||
if_save_latest1Bb = gr.Checkbox(label=i18n("是否仅保存最新的ckpt文件以节省硬盘空间"), value=True, interactive=True, show_label=True)
|
||||
if_save_every_weights1Bb = gr.Checkbox(label=i18n("是否在每次保存时间点将最终小模型保存至weights文件夹"), value=True, interactive=True, show_label=True)
|
||||
save_every_epoch1Bb = gr.Slider(minimum=1,maximum=50,step=1,label=i18n("保存频率save_every_epoch"),value=5,interactive=True)
|
||||
gpu_numbers1Bb = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"), value="%s" % (gpus), interactive=True)
|
||||
with gr.Row():
|
||||
button1Bb_open = gr.Button(i18n("开启GPT训练"), variant="primary",visible=True)
|
||||
button1Bb_close = gr.Button(i18n("终止GPT训练"), variant="primary",visible=False)
|
||||
info1Bb=gr.Textbox(label=i18n("GPT训练进程输出信息"))
|
||||
with gr.Column():
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("1Ba-SoVITS训练。用于分享的模型文件输出在SoVITS_weights下。")):
|
||||
with gr.Group():
|
||||
batch_size = gr.Slider(minimum=1,maximum=40,step=1,label=i18n("每张显卡的batch_size"),value=default_batch_size,interactive=True)
|
||||
total_epoch = gr.Slider(minimum=1,maximum=25,step=1,label=i18n("总训练轮数total_epoch，不建议太高"),value=8,interactive=True)
|
||||
text_low_lr_rate = gr.Slider(minimum=0.2,maximum=0.6,step=0.05,label=i18n("文本模块学习率权重"),value=0.4,interactive=True)
|
||||
save_every_epoch = gr.Slider(minimum=1,maximum=25,step=1,label=i18n("保存频率save_every_epoch"),value=4,interactive=True)
|
||||
if_save_latest = gr.Checkbox(label=i18n("是否仅保存最新的ckpt文件以节省硬盘空间"), value=True, interactive=True, show_label=True)
|
||||
if_save_every_weights = gr.Checkbox(label=i18n("是否在每次保存时间点将最终小模型保存至weights文件夹"), value=True, interactive=True, show_label=True)
|
||||
gpu_numbers1Ba = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"), value="%s" % (gpus), interactive=True)
|
||||
with gr.Group():
|
||||
info1Ba=gr.Textbox(label=i18n("SoVITS训练进程输出信息"))
|
||||
button1Ba_open = gr.Button(i18n("开启SoVITS训练"), variant="primary",visible=True)
|
||||
button1Ba_close = gr.Button(i18n("终止SoVITS训练"), variant="primary",visible=False)
|
||||
with gr.Column():
|
||||
with gr.Tabs():
|
||||
with gr.Tab(i18n("1Bb-GPT训练。用于分享的模型文件输出在GPT_weights下。")):
|
||||
with gr.Group():
|
||||
batch_size1Bb = gr.Slider(minimum=1,maximum=40,step=1,label=i18n("每张显卡的batch_size"),value=default_batch_size,interactive=True)
|
||||
total_epoch1Bb = gr.Slider(minimum=2,maximum=50,step=1,label=i18n("总训练轮数total_epoch"),value=15,interactive=True)
|
||||
if_dpo = gr.Checkbox(label=i18n("是否开启dpo训练选项(实验性)"), value=False, interactive=True, show_label=True)
|
||||
if_save_latest1Bb = gr.Checkbox(label=i18n("是否仅保存最新的ckpt文件以节省硬盘空间"), value=True, interactive=True, show_label=True)
|
||||
if_save_every_weights1Bb = gr.Checkbox(label=i18n("是否在每次保存时间点将最终小模型保存至weights文件夹"), value=True, interactive=True, show_label=True)
|
||||
save_every_epoch1Bb = gr.Slider(minimum=1,maximum=50,step=1,label=i18n("保存频率save_every_epoch"),value=5,interactive=True)
|
||||
gpu_numbers1Bb = gr.Textbox(label=i18n("GPU卡号以-分割，每个卡号一个进程"), value="%s" % (gpus), interactive=True)
|
||||
with gr.Group():
|
||||
info1Bb=gr.Textbox(label=i18n("GPT训练进程输出信息"))
|
||||
button1Bb_open = gr.Button(i18n("开启GPT训练"), variant="primary",visible=True)
|
||||
button1Bb_close = gr.Button(i18n("终止GPT训练"), variant="primary",visible=False)
|
||||
button1Ba_open.click(open1Ba, [batch_size,total_epoch,exp_name,text_low_lr_rate,if_save_latest,if_save_every_weights,save_every_epoch,gpu_numbers1Ba,pretrained_s2G,pretrained_s2D], [info1Ba,button1Ba_open,button1Ba_close])
|
||||
button1Ba_close.click(close1Ba, [], [info1Ba,button1Ba_open,button1Ba_close])
|
||||
button1Bb_open.click(open1Bb, [batch_size1Bb,total_epoch1Bb,exp_name,if_dpo,if_save_latest1Bb,if_save_every_weights1Bb,save_every_epoch1Bb,gpu_numbers1Bb,pretrained_s1], [info1Bb,button1Bb_open,button1Bb_close])
|
||||
button1Bb_close.click(close1Bb, [], [info1Bb,button1Bb_open,button1Bb_close])
|
||||
with gr.TabItem(i18n("1C-推理")):
|
||||
with gr.Tab(i18n("1C-推理")):
|
||||
gr.Markdown(value=i18n("选择训练完存放在SoVITS_weights和GPT_weights下的模型。默认的一个是底模，体验5秒Zero Shot TTS用。"))
|
||||
with gr.Row():
|
||||
GPT_dropdown = gr.Dropdown(label=i18n("*GPT模型列表"), choices=sorted(GPT_names,key=custom_sort_key),value=pretrained_gpt_name,interactive=True)
|
||||
SoVITS_dropdown = gr.Dropdown(label=i18n("*SoVITS模型列表"), choices=sorted(SoVITS_names,key=custom_sort_key),value=pretrained_sovits_name,interactive=True)
|
||||
gpu_number_1C=gr.Textbox(label=i18n("GPU卡号,只能填1个整数"), value=gpus, 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])
|
||||
with gr.Row():
|
||||
if_tts = gr.Checkbox(label=i18n("是否开启TTS推理WebUI"), show_label=True)
|
||||
tts_info = gr.Textbox(label=i18n("TTS推理WebUI进程输出信息"))
|
||||
if_tts.change(change_tts_inference, [if_tts,bert_pretrained_dir,cnhubert_base_dir,gpu_number_1C,GPT_dropdown,SoVITS_dropdown], [tts_info])
|
||||
with gr.TabItem(i18n("2-GPT-SoVITS-变声")):gr.Markdown(value=i18n("施工中，请静候佳音"))
|
||||
app.queue(concurrency_count=511, max_size=1022).launch(
|
||||
server_name="0.0.0.0",
|
||||
inbrowser=True,
|
||||
share=is_share,
|
||||
server_port=webui_port_main,
|
||||
quiet=True,
|
||||
)
|
||||
with gr.Column():
|
||||
gpu_number_1C=gr.Textbox(label=i18n("GPU卡号,只能填1个整数"), value=gpus, interactive=True)
|
||||
with gr.Group():
|
||||
GPT_dropdown = gr.Dropdown(label=i18n("*GPT模型列表"), choices=sorted(GPT_names,key=custom_sort_key),value=pretrained_gpt_name,interactive=True)
|
||||
SoVITS_dropdown = gr.Dropdown(label=i18n("*SoVITS模型列表"), choices=sorted(SoVITS_names,key=custom_sort_key),value=pretrained_sovits_name,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])
|
||||
with gr.Column():
|
||||
if_tts = gr.Checkbox(label=i18n("是否开启TTS推理WebUI"), show_label=True)
|
||||
tts_info = gr.Textbox(label=i18n("TTS推理WebUI进程输出信息"))
|
||||
if_tts.change(change_tts_inference, [if_tts,bert_pretrained_dir,cnhubert_base_dir,gpu_number_1C,GPT_dropdown,SoVITS_dropdown], [tts_info])
|
||||
with gr.Tab(i18n("2-GPT-SoVITS-变声")):gr.Markdown(value=i18n("施工中，请静候佳音"))
|
||||
|
||||
if gr.__version__.split(".")[0] == "4":
|
||||
app.launch(
|
||||
server_name="0.0.0.0",
|
||||
inbrowser=True,
|
||||
share=is_share,
|
||||
server_port=webui_port_main,
|
||||
quiet=True,
|
||||
)
|
||||
else:
|
||||
app.queue(concurrency_count=511, max_size=1022).launch(
|
||||
server_name="0.0.0.0",
|
||||
inbrowser=True,
|
||||
share=is_share,
|
||||
server_port=webui_port_main,
|
||||
quiet=True,
|
||||
)
|
||||
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user