forked from svc-develop-team/so-vits-svc
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathwebUI.py
More file actions
149 lines (131 loc) · 6.49 KB
/
webUI.py
File metadata and controls
149 lines (131 loc) · 6.49 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
import gradio as gr
import os
from inference.infer_tool import Svc, format_wav
from spkmix import spk_mix_map
import soundfile
import tempfile
def save_uploaded(file_obj):
os.makedirs("raw", exist_ok=True)
with open(file_obj.name, "rb") as src:
content = src.read()
filename = os.path.basename(file_obj.name)
save_path = os.path.join("raw", filename)
with open(save_path, "wb") as dst:
dst.write(content)
return filename
def infer(
model_path, config_path, uploaded_files, trans, spk_list,
clip, auto_predict_f0, cluster_model_path, cluster_infer_ratio,
linear_gradient, f0_predictor, enhance, shallow_diffusion, use_spk_mix,
loudness_envelope_adjustment, feature_retrieval, diffusion_model_path,
diffusion_config_path, k_step, second_encoding, only_diffusion,
slice_db, device, noice_scale, pad_seconds, wav_format,
linear_gradient_retain, enhancer_adaptive_key, f0_filter_threshold
):
clean_names = [save_uploaded(f) for f in uploaded_files]
trans = list(map(int, trans.strip().split())) if isinstance(trans, str) else trans
spk_list = spk_list.strip().split()
svc_model = Svc(model_path, config_path, device, cluster_model_path,
enhance, diffusion_model_path, diffusion_config_path,
shallow_diffusion, only_diffusion, use_spk_mix, feature_retrieval)
os.makedirs("results", exist_ok=True)
if len(spk_mix_map) <= 1:
use_spk_mix = False
if use_spk_mix:
spk_list = [spk_mix_map]
outputs = []
for name, tran in zip(clean_names, trans):
raw_audio_path = f"raw/{name}"
"""
if not raw_audio_path.endswith(".wav"):
raw_audio_path += ".wav"
"""
format_wav(raw_audio_path)
for spk in spk_list:
audio = svc_model.slice_inference(
raw_audio_path=raw_audio_path,
spk=spk,
tran=tran,
slice_db=slice_db,
cluster_infer_ratio=cluster_infer_ratio,
auto_predict_f0=auto_predict_f0,
noice_scale=noice_scale,
pad_seconds=pad_seconds,
clip_seconds=clip,
lg_num=linear_gradient,
lgr_num=linear_gradient_retain,
f0_predictor=f0_predictor,
enhancer_adaptive_key=enhancer_adaptive_key,
cr_threshold=f0_filter_threshold,
k_step=k_step,
use_spk_mix=use_spk_mix,
second_encoding=second_encoding,
loudness_envelope_adjustment=loudness_envelope_adjustment
)
key = "auto" if auto_predict_f0 else f"{tran}key"
cluster_tag = "" if cluster_infer_ratio == 0 else f"_{cluster_infer_ratio}"
mode = "sovits"
if shallow_diffusion: mode = "sovdiff"
if only_diffusion: mode = "diff"
if use_spk_mix: spk = "spk_mix"
out_path = f"results/{name}_{key}_{spk}{cluster_tag}_{mode}_{f0_predictor}.{wav_format}"
soundfile.write(out_path, audio, svc_model.target_sample, format=wav_format)
svc_model.clear_empty()
outputs.append(out_path)
return outputs
with gr.Blocks() as demo:
gr.Markdown("# So-VITS-SVC WebUI")
with gr.Row():
model_path = gr.Textbox("logs/44k/G_0.pth", label="模型路径")
config_path = gr.Textbox("logs/44k/config.json", label="配置路径")
cluster_model_path = gr.Textbox("", label="聚类模型路径（可为空）")
shallow_diffusion = gr.Checkbox(False, label="使用浅层扩散")
diffusion_model_path = gr.Textbox("logs/44k/diffusion/model_0.pt", label="扩散模型路径")
diffusion_config_path = gr.Textbox("logs/44k/diffusion/config.yaml", label="扩散配置")
with gr.Row():
#uploaded_files = gr.File(file_count="single", label="上传音频文件")
uploaded_files = gr.File(
file_count="multiple",
file_types=[".wav", ".mp3", ".m4a", ".aac", ".ogg"],
label="上传音频文件（可多选）"
)
trans = gr.Textbox("0", label="音高变化（半音） 空格分隔，对应每个文件")
spk_list = gr.Textbox("female", label="目标说话人 空格分隔")
with gr.Row():
clip = gr.Number(0, label="强制切片/s（0为自动）")
cluster_infer_ratio = gr.Slider(0, 1, step=0.01, value=0, label="聚类混合比例")
linear_gradient = gr.Number(0, label="淡入长度/s")
linear_gradient_retain = gr.Slider(0, 1, value=0.75, label="交叉保留比例")
loudness_envelope_adjustment = gr.Slider(0, 1, value=1, label="响度包络融合比例")
with gr.Row():
f0_predictor = gr.Dropdown(["pm", "crepe", "dio", "harvest", "rmvpe", "fcpe"], value="rmvpe", label="F0预测器")
wav_format = gr.Dropdown(["flac", "wav"], value="flac", label="输出格式")
device = gr.Dropdown(["cpu", "cuda"], value="cpu", label="推理设备")
k_step = gr.Number(100, label="扩散步数")
with gr.Row():
slice_db = gr.Number(-40, label="切片分贝阈值")
noice_scale = gr.Number(0.4, label="噪音等级")
pad_seconds = gr.Number(0.5, label="pad秒数")
enhancer_adaptive_key = gr.Number(0, label="增强器音域调整")
f0_filter_threshold = gr.Slider(0, 1, value=0.05, label="F0过滤阈值")
with gr.Row():
auto_predict_f0 = gr.Checkbox(False, label="自动预测F0")
enhance = gr.Checkbox(False, label="使用增强器")
only_diffusion = gr.Checkbox(False, label="纯扩散模式")
use_spk_mix = gr.Checkbox(False, label="角色融合")
second_encoding = gr.Checkbox(False, label="二次编码")
feature_retrieval = gr.Checkbox(False, label="特征检索")
btn = gr.Button("开始推理")
output = gr.Files(label="输出结果")
btn.click(fn=infer, inputs=[
model_path, config_path, uploaded_files, trans, spk_list,
clip, auto_predict_f0, cluster_model_path, cluster_infer_ratio,
linear_gradient, f0_predictor, enhance, shallow_diffusion, use_spk_mix,
loudness_envelope_adjustment, feature_retrieval, diffusion_model_path,
diffusion_config_path, k_step, second_encoding, only_diffusion,
slice_db, device, noice_scale, pad_seconds, wav_format,
linear_gradient_retain, enhancer_adaptive_key, f0_filter_threshold
], outputs=output)
if __name__ == '__main__':
#os.system("start http://127.0.0.1:7860")
demo.launch(server_name="0.0.0.0")