-
Notifications
You must be signed in to change notification settings - Fork 4
/
start.py
509 lines (474 loc) · 23.2 KB
/
start.py
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
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
import os
import numpy as np
os.environ['PPSPEECH_HOME'] = os.getcwd()
import json
import tarfile
import paddle
import soundfile as sf
import streamlit as st
from datetime import datetime
from tool.src.tts import TTSExecutor, pretrained_models, front_models
@st.cache(allow_output_mutation=True)
def load_model(am='fastspeech2_aishell3',
am_config=None,
am_ckpt=None,
am_stat=None,
am_tag="1.0",
phones_dict=None,
tones_dict=None,
speaker_dict=None,
voc='pwgan_aishell3',
voc_config=None,
voc_ckpt=None,
voc_stat=None,
voc_tag="1.0",
lang='zh',
front='pinyin',
device=paddle.get_device()):
tts_executor = TTSExecutor(am=am,
am_config=am_config,
am_ckpt=am_ckpt,
am_stat=am_stat,
am_tag=am_tag,
phones_dict=phones_dict,
tones_dict=tones_dict,
speaker_dict=speaker_dict,
voc=voc,
voc_config=voc_config,
voc_ckpt=voc_ckpt,
voc_stat=voc_stat,
voc_tag=voc_tag,
lang=lang,
front=front,
device=device)
return tts_executor
class Builder():
def __init__(self):
st.set_page_config(
page_title="Dubbing配音工具", # st.get_option(""),
page_icon=":shark",
layout="wide",
initial_sidebar_state="auto",
)
self.jsonfile_path = os.path.join(os.getcwd(), "history/dubbing_log")
if not os.path.exists(self.jsonfile_path):
os.makedirs(self.jsonfile_path)
def start_doc(self):
st.header("导语")
st.text("欢迎使用AI配音工具PaddleDubbing，本工具基于streamlit搭建。")
st.text("作者: phecda-xu")
st.markdown("模型能力由[paddlespeech](https://github.com/PaddlePaddle/PaddleSpeech)提供。")
st.subheader("操作引导")
st.markdown("先在`应用`中选择要使用的功能。")
st.markdown("然后页面右上角选择栏里面有 `说明`，`选我开始`，`历史记录`三个选项。")
st.markdown("参照`说明`内容配置左侧的内容。")
st.markdown("配置完成后，选择`选我开始`，然后继续按照`说明`的内容操作即可。")
st.markdown("所有合成的音频都可以在`历史记录`里面看到。")
st.markdown("任务执行过程中不要做其他操作，否则会打断当前任务进程。")
def tts_leader_doc(self):
st.subheader("语音合成")
st.markdown("**第一步**: 选择是否使用GPU（GPU默认只使用 device id 为0的设备）")
st.markdown("**第二步**: 选择语种，目前支持中文（`zh`）,英文（`en`）,不支持双语混合，可用中文音替换。")
st.markdown("**第三步**: 选择声学模型和声码器，比如： `fastspeech2_aishell3` 和 `pwgan_aishell3` 组合。")
st.markdown("**注意**:下划线后边的要保持一致比如 `aishell3` 这样合成音频的质量才是好的。")
st.markdown("选定后页面右侧会出现加载字样和图标，配置参数变动后都会自动重新加载模型。")
st.markdown("**第四步**: 选择说话人ID，每个ID对应一个说话风格，大部分是女声。")
st.markdown("**第五步**: 已支持语速、音高和音量调节。")
st.markdown("**第六步**: 保存音频的地址，默认为当前代码路劲下的 `output/dubbing` 。")
st.subheader("模型列表")
# 声学模型
ac_tag = ["speedyspeech","fastspeech","tacotron"]
# 声码器
voc_tag = ["gan","wavernn"]
st.markdown("#### **声学模型:**")
for key, value in list(pretrained_models.items()):
if ac_tag[0] in key or ac_tag[1] in key or ac_tag[2] in key:
cols1, cols2 = st.columns([2,2])
with cols1:
st.markdown(f"**model_name:** {key}")
with cols2:
tag_list = list(value.keys())
a = [f"[{i}]({value[i]['url']})" for i in tag_list]
st.markdown(f"**tag:** {a}")
st.markdown("#### **声码器:**")
for key, value in list(pretrained_models.items()):
if voc_tag[0] in key or voc_tag[1] in key:
cols1, cols2 = st.columns([2, 2])
with cols1:
st.markdown(f"**model_name:** {key}")
with cols2:
tag_list = list(value.keys())
a = [f"[{i}]({value[i]['url']})" for i in tag_list]
st.markdown(f"**tag:** {a}")
st.markdown("#### **Frontend model:**")
for key, value in list(front_models.items()):
cols1, cols2 = st.columns([2, 2])
with cols1:
st.markdown(f"**model_name:** {key}")
with cols2:
tag_list = list(value.keys())
a = [f"[{i}]({value[i]['url']})" for i in tag_list]
st.markdown(f"**tag:** {a}")
st.markdown("```\n"
"模型文件结构：\n"
f"├──models \n"
f"│ ├──fastspeech2_aishell3-zh \n"
f"│ │ ├──fastspeech2_nosil_aishell3_ckpt_0.4 \n"
f"│ │ │ ├──default.yaml \n"
f"│ │ │ ├──... \n"
f"│ │ │ └──speech_stats.npy \n"
f"│ │ └──fastspeech2_nosil_aishell3_ckpt_0.4.zip \n"
f"│ ├──pwgan_aishell3-zh \n"
f"│ │ ├──pwg_aishell3_ckpt_0.5 \n"
f"│ │ │ ├──default.yaml \n"
f"│ │ │ ├──... \n"
f"│ │ │ └──speech_stats.npy \n"
f"│ │ └──pwg_aishell3_ckpt_0.5.zip \n"
f"... \n"
"```")
st.markdown("**注**：点击tag 可以下载对应的模型！")
def gpu_setting(self):
col1, col2 = st.sidebar.columns([1, 3])
with col1:
gpu_option = st.checkbox('GPU')
with col2:
if gpu_option:
st.markdown("<font size=3.5>(当前使用 gpu:0)</font>", unsafe_allow_html=True)
os.environ["CUDA_VISIBLE_DEVICES"] = '0'
else:
st.markdown("<font size=3.5>(当前使用 cpu)</font> ", unsafe_allow_html=True)
os.environ["CUDA_VISIBLE_DEVICES"] = ''
def model_setting(self):
self.am_list = []
self.voc_list = []
ac_tag = ["speedyspeech", "fastspeech", "tacotron"]
voc_tag = ["gan", "wavernn"]
# 搜集本地 finetune 模型
for new_model_name in os.listdir(os.path.join(os.environ['PPSPEECH_HOME'], "models")):
if "fintune" in new_model_name:
file_path = os.path.join(os.environ['PPSPEECH_HOME'], "models", new_model_name)
for folder_name in os.listdir(file_path):
if not folder_name.endswith(".zip"):
for i in os.listdir(os.path.join(file_path, folder_name)):
if i.endswith('pdz'):
ckpt_name = i
if ac_tag[0] in new_model_name or ac_tag[1] in new_model_name or ac_tag[2] in new_model_name:
pretrained_models[new_model_name] = {
"1.0": {
'url':
f'models/{new_model_name}/{folder_name}.zip',
'md5':
'0',
'config':
'default.yaml',
'ckpt':
ckpt_name,
'speech_stats':
'speech_stats.npy',
'phones_dict':
'phone_id_map.txt',
'speaker_dict':
'speaker_id_map.txt',
}
}
if voc_tag[0] in new_model_name or voc_tag[1] in new_model_name:
pretrained_models[new_model_name] = {
"1.0": {
'url':
f'models/{new_model_name}/{folder_name}.zip',
'md5':
'0',
'config':
'default.yaml',
'ckpt':
ckpt_name,
'speech_stats':
'feats_stats.npy',
}
}
for key, value in list(pretrained_models.items()):
if ac_tag[0] in key or ac_tag[1] in key or ac_tag[2] in key:
self.am_list.append(key)
for key, value in list(pretrained_models.items()):
if voc_tag[0] in key or voc_tag[1] in key:
self.voc_list.append(key)
# st.sidebar.header("语种")
lang_option = st.sidebar.selectbox(
'选择语言',
['zh', 'en', 'mix']
)
am_zh_model_list = ['']
am_en_model_list = ['']
am_mix_model_list = ['']
for i in self.am_list:
if "zh" in i:
am_zh_model_list.append(i.replace("-zh", ""))
if "en" in i:
am_en_model_list.append(i.replace("-en", ""))
if "mix" in i:
am_mix_model_list.append(i.replace("-mix", ""))
voc_zh_model_list = ['']
voc_en_model_list = ['']
voc_mix_model_list = ['']
for i in self.voc_list:
if "zh" in i:
voc_zh_model_list.append(i.replace("-zh", ""))
if "en" in i:
voc_en_model_list.append(i.replace("-en", ""))
if "mix" in i:
voc_mix_model_list.append(i.replace("-mix", ""))
# st.sidebar.header("配置模型")
if lang_option == "zh":
am_option = st.sidebar.selectbox('声学模型',am_zh_model_list, key="am_zh_option")
if am_option != "":
am_tag_list = list(pretrained_models[am_option + '-zh'].keys())
am_tag_option = st.sidebar.selectbox('tag', am_tag_list, key="am_zh_tag_option")
else:
am_tag_option = ""
voc_option = st.sidebar.selectbox('声码器', voc_zh_model_list, key="voc_zh_option")
if voc_option != "":
voc_tag_list = list(pretrained_models[voc_option + '-zh'].keys())
voc_tag_option = st.sidebar.selectbox('tag', voc_tag_list, key="voc_zh_tag_option")
else:
voc_tag_option = ""
elif lang_option == "en":
am_option = st.sidebar.selectbox('声学模型', am_en_model_list, key="am_en_option")
if am_option != "":
am_tag_list = list(pretrained_models[am_option + '-en'].keys())
am_tag_option = st.sidebar.selectbox('tag', am_tag_list, key="am_en_tag_option")
else:
am_tag_option = ""
voc_option = st.sidebar.selectbox('声码器', voc_en_model_list, key="voc_en_option")
if voc_option != "":
voc_tag_list = list(pretrained_models[voc_option + '-en'].keys())
voc_tag_option = st.sidebar.selectbox('tag', voc_tag_list, key="voc_en_tag_option")
else:
voc_tag_option = ""
elif lang_option == "mix":
am_option = st.sidebar.selectbox('声学模型', am_mix_model_list, key="am_mix_option")
if am_option != "":
am_tag_list = list(pretrained_models[am_option + '-mix'].keys())
am_tag_option = st.sidebar.selectbox('tag', am_tag_list, key="am_mix_tag_option")
else:
am_tag_option = ""
voc_option = st.sidebar.selectbox('声码器', voc_zh_model_list, key="voc_mix_option")
if voc_option != "":
voc_tag_list = list(pretrained_models[voc_option + '-zh'].keys())
voc_tag_option = st.sidebar.selectbox('tag', voc_tag_list, key="voc_mix_tag_option")
else:
voc_tag_option = ""
else:
raise ValueError("")
frontend_option = st.sidebar.selectbox("frontend选择", ["g2pM", "g2pW"])
# 自动加载模型
if am_option != '' and voc_option != '':
with st.spinner("模型加载中..."):
# print(am_option, voc_option, lang_option)
self.tts_executor = load_model(am=am_option,
am_tag=am_tag_option,
voc=voc_option,
voc_tag=voc_tag_option,
lang=lang_option,
front=frontend_option)
def other_setting(self):
st.sidebar.header("参数配置")
self.spk_option = st.sidebar.number_input(label='说话人ID', min_value=0, max_value=200, step=1, format='%d')
self.speed_option = st.sidebar.slider(label='语速', value=1.0, min_value=0.7, max_value=2.0, step=0.1)
col1, col2 = st.sidebar.columns([2, 2])
with col1:
self.energy_option = st.slider(label='音量', value=1.0, min_value=0.3, max_value=3.0, step=0.1)
with col2:
self.pitch_option = st.slider(label='音高', value=1.0, min_value=0.7, max_value=1.3, step=0.1)
output_path = os.path.join(os.getcwd(), 'output', 'dubbing')
self.save_option = st.sidebar.text_input("保存路劲", output_path)
if not os.path.exists(self.save_option):
os.makedirs(self.save_option)
def get_text_input(self):
uploader_file = st.file_uploader("输入文件：")
if uploader_file is not None:
res = uploader_file.getvalue().decode("utf-8")
text = st.text_area("输入文本：",
value=res, key=None)
else:
text = st.text_area("输入文本：(在这里输入多行文本,每行文本字数不限,文本框右下角可以拖动。 )",
value='', key=None)
text_list = text.split("\n")
return text_list
@staticmethod
def get_wav_data(bin_file, file_label='File'):
with open(bin_file, 'rb') as f:
data = f.read()
return data
@staticmethod
def get_batch_wav_data(bin_file_list, file_label='File'):
data = b''
for bin_file in bin_file_list:
# signal, samplerate = sf.read(bin_file)
with open(bin_file, 'rb') as f:
bin_data = f.read()
data += bin_data
print(bin_data[:48])
return data
def make_targz_one_by_one(self, output_filename, pathfile):
tar = tarfile.open(output_filename, "w:gz")
tar.add(pathfile)
tar.close()
def update_history(self):
for json_file in os.listdir(self.jsonfile_path):
json_name = os.path.basename(json_file)
# print(json_name)
with st.expander(json_name[:-5], expanded=False):
with open(os.path.join(self.jsonfile_path, json_file)) as f:
json_dicts = json.load(f)
col1, col2, col3 = st.columns([3, 3, 1])
col1.subheader("文本")
col2.subheader("音频")
col3.subheader("下载")
array_list = []
array_list.append(np.zeros(300))
# with st.spinner("加载中..."):
for key, value in json_dicts.items():
json_dict = value
col1_1, col2_1, col3_1 = st.columns([3, 3, 1])
with col1_1:
st.write("-" * 60)
st.write("句 {} : {} ".format(key, json_dict["text"]))
output = json_dict["audio_path"]
buffer = self.get_wav_data(output)
signal, sr = sf.read(output)
array_list.append(signal)
array_list.append(np.zeros(1000))
with col2_1:
st.write("-" * 60)
st.audio(buffer, format='audio/wav')
with col3_1:
st.write("-" * 60)
st.download_button(
label="Download",
data=buffer,
file_name=os.path.basename(output),
mime="application/octet-stream",
key=json_name[:-5] + '_' + str(key)
)
batch_array = np.concatenate(array_list)
bath_wav_filepath = '{}/{}/{}.wav'.format(self.save_option, json_name[:-5], json_name[:-5])
sf.write(bath_wav_filepath, batch_array, sr)
def process(self, text_list):
jsonfile_name = "{}.json".format(datetime.now().strftime('%Y_%m_%d%Z_%H_%M_%S'))
p_col1, p_col2, p_col3, p_col4 = st.columns([16, 1, 1, 1])
with p_col1:
st.write("\n")
process_bar = st.progress(0)
step = int((1 / len(text_list)) * 100)
with p_col2:
if text_list[0] == '':
st.markdown("<font size=5.5> {} 条</font>".format(len(text_list) - 1), unsafe_allow_html=True)
else:
st.markdown("<font size=5.5> {} 条</font>".format(len(text_list)), unsafe_allow_html=True)
with p_col3:
start_button = st.button("开始")
with p_col4:
stop_button = st.button("终止")
# 开始合成后的过程
if start_button:
with st.expander("详细信息", expanded=True):
col1, col2, col3 = st.columns([3, 3, 1])
col1.subheader("文本")
col2.subheader("音频")
col3.subheader("下载")
process_dic = {}
n = 0
break_state = 0
for i in text_list:
text_i = i.strip('\r')
col1, col2, col3 = st.columns([3, 3, 1])
n += 1
if text_i == '':
st.error("请输入要合成的文本内容！！")
break_state = 2
break
if stop_button:
st.stop()
break_state = 2
break
with col1:
st.write("-" * 60)
st.write("句 {} : {} ".format(n, text_i))
outfile_name = '{}_spk_{}_{}.wav'.format(n, self.spk_option, text_i)
output_path = os.path.join(self.save_option, jsonfile_name[:-5])
if not os.path.exists(output_path):
os.makedirs(output_path)
output = os.path.join(output_path, outfile_name)
with st.spinner("合成中..."):
wav_file = self.tts_executor(
text=text_i,
output=output,
spk_id=self.spk_option,
speed_degree=self.speed_option,
pitch_degree=self.pitch_option,
energy_degree=self.energy_option,
robot=False
)
process_dic[n] = {"text": text_i, "audio_path": output}
with open(os.path.join(self.jsonfile_path, jsonfile_name), 'w') as f:
f.write(json.dumps(process_dic, ensure_ascii=False, indent=4))
process_bar.progress(n * step)
buffer = self.get_wav_data(wav_file)
with col2:
st.write("-" * 60)
st.audio(buffer, format='audio/wav')
with col3:
st.write("-" * 60)
st.download_button(
label="Download",
data=buffer,
file_name=outfile_name,
mime="application/octet-stream",
key=n
)
break_state = 1
if break_state == 1:
process_bar.progress(100)
st.balloons()
def __call__(self, *args, **kwargs):
# 侧边栏应用分页设置
st.sidebar.subheader("配置栏")
app_option = st.sidebar.selectbox(
'应用',
["导语", "语音合成", "语音克隆"]
)
if app_option == "语音合成":
# 侧边栏GPU相关显示设置
self.gpu_setting()
# 侧边栏语种、声学模型、声码器选择设置
self.model_setting()
# 侧边栏合成参数、保存路劲配置
self.other_setting()
# 显示区显示设置
col1, col2 = st.columns([17, 2])
with col1:
st.empty()
with col2:
area_option = st.selectbox(
'area',
['说明', '选我开始', '历史记录'],
label_visibility="collapsed"
)
# 主页操作区
if area_option == '选我开始':
# 取输入文本
text_list = self.get_text_input()
# 合成
self.process(text_list)
# 历史记录区
elif area_option == "历史记录":
self.update_history()
else:
self.tts_leader_doc()
elif app_option == "语音克隆":
st.text("模型能力由paddlespeech提供。")
else:
self.start_doc()
if __name__ == "__main__":
app = Builder()
app()