-
Notifications
You must be signed in to change notification settings - Fork 29
/
Copy pathChatWaifu_marai.py
427 lines (360 loc) · 16.5 KB
/
ChatWaifu_marai.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
import time
from scipy.io.wavfile import write
from mel_processing import spectrogram_torch
from text import text_to_sequence, _clean_text
from models import SynthesizerTrn
import utils
import commons
import sys
import re
import os
import miraicle
from torch import no_grad, LongTensor
import logging
from winsound import PlaySound
sound_status=False
qq = # 你登录的机器人 QQ 号
verify_key = '' # 你在 setting.yml 中设置的 verifyKey
port =8080 # 你在 setting.yml 中设置的 port (http)
####################################
#CHATGPT INITIALIZE
from pyChatGPT import ChatGPT
import json
idmessage = """ID Speaker
0 綾地寧々
1 在原七海
2 小茸
3 唐乐吟
"""
speakerID = 0
def get_input():
# prompt for input
print("You:")
user_input = input()
return user_input
if os.path.exists("./token.txt") == True:
def get_token():
tok = open("token.txt",encoding = "utf-8")
token=tok.read()
print("you have cached your token, simpiflied your login!")
return token;
else :
def get_token():
token = input("Copy your token from ChatGPT and press Enter \n")
with open('token.txt','w') as tok:
tok.write(token)
return token;
################################################
logging.getLogger('numba').setLevel(logging.WARNING)
def ex_print(text, escape=False):
if escape:
print(text.encode('unicode_escape').decode())
else:
print(text)
def get_text(text, hps, cleaned=False):
if cleaned:
text_norm = text_to_sequence(text, hps.symbols, [])
else:
text_norm = text_to_sequence(text, hps.symbols, hps.data.text_cleaners)
if hps.data.add_blank:
text_norm = commons.intersperse(text_norm, 0)
text_norm = LongTensor(text_norm)
return text_norm
def ask_if_continue():
while True:
answer = input('Continue? (y/n): ')
if answer == 'y':
break
elif answer == 'n':
sys.exit(0)
def print_speakers(speakers, escape=False):
if len(speakers) > 100:
return
print('ID\tSpeaker')
for id, name in enumerate(speakers):
ex_print(str(id) + '\t' + name, escape)
def get_speaker_id(message):
speaker_id = input(message)
try:
speaker_id = int(speaker_id)
except:
print(str(speaker_id) + ' is not a valid ID!')
sys.exit(1)
return speaker_id
def get_label_value(text, label, default, warning_name='value'):
value = re.search(rf'\[{label}=(.+?)\]', text)
if value:
try:
text = re.sub(rf'\[{label}=(.+?)\]', '', text, 1)
value = float(value.group(1))
except:
print(f'Invalid {warning_name}!')
sys.exit(1)
else:
value = default
return value, text
def get_label(text, label):
if f'[{label}]' in text:
return True, text.replace(f'[{label}]', '')
else:
return False, text
def generateSound(inputString,language):
if '--escape' in sys.argv:
escape = True
else:
escape = False
#model = input('Path of a VITS model: ')
#config = input('Path of a config file: ')
if language=="ch":
model = r".\model\CN\model.pth"
config = r".\model\CN\config.json"
elif language=="jp":
model = r".\model\H_excluded.pth"
config = r".\model\config.json"
elif language=="multi":
model = r".\model\Multi\multi.pth"
config = r".\model\Multi\config.json"
hps_ms = utils.get_hparams_from_file(config)
n_speakers = hps_ms.data.n_speakers if 'n_speakers' in hps_ms.data.keys() else 0
n_symbols = len(hps_ms.symbols) if 'symbols' in hps_ms.keys() else 0
speakers = hps_ms.speakers if 'speakers' in hps_ms.keys() else ['0']
use_f0 = hps_ms.data.use_f0 if 'use_f0' in hps_ms.data.keys() else False
emotion_embedding = hps_ms.data.emotion_embedding if 'emotion_embedding' in hps_ms.data.keys() else False
net_g_ms = SynthesizerTrn(
n_symbols,
hps_ms.data.filter_length // 2 + 1,
hps_ms.train.segment_size // hps_ms.data.hop_length,
n_speakers=n_speakers,
emotion_embedding=emotion_embedding,
**hps_ms.model)
_ = net_g_ms.eval()
utils.load_checkpoint(model, net_g_ms)
def voice_conversion():
audio_path = input('Path of an audio file to convert:\n')
print_speakers(speakers)
audio = utils.load_audio_to_torch(
audio_path, hps_ms.data.sampling_rate)
originnal_id = get_speaker_id('Original speaker ID: ')
target_id = get_speaker_id('Target speaker ID: ')
out_path = input('Path to save: ')
y = audio.unsqueeze(0)
spec = spectrogram_torch(y, hps_ms.data.filter_length,
hps_ms.data.sampling_rate, hps_ms.data.hop_length, hps_ms.data.win_length,
center=False)
spec_lengths = LongTensor([spec.size(-1)])
sid_src = LongTensor([originnal_id])
with no_grad():
sid_tgt = LongTensor([target_id])
audio = net_g_ms.voice_conversion(spec, spec_lengths, sid_src=sid_src, sid_tgt=sid_tgt)[
0][0, 0].data.cpu().float().numpy()
return audio, out_path
if n_symbols != 0:
if not emotion_embedding:
#while True:
if(1==1):
#choice = input('TTS or VC? (t/v):')
choice = 't'
if choice == 't':
#text = input('Text to read: ')
text = inputString
if text == '[ADVANCED]':
#text = input('Raw text:')
text = "我不会说"
#print('Cleaned text is:')
#ex_print(_clean_text(
# text, hps_ms.data.text_cleaners), escape)
#continue
length_scale, text = get_label_value(
text, 'LENGTH', 1, 'length scale')
noise_scale, text = get_label_value(
text, 'NOISE', 0.667, 'noise scale')
noise_scale_w, text = get_label_value(
text, 'NOISEW', 0.8, 'deviation of noise')
cleaned, text = get_label(text, 'CLEANED')
stn_tst = get_text(text, hps_ms, cleaned=cleaned)
#print_speakers(speakers, escape)
#speaker_id = get_speaker_id('Speaker ID: ')
speaker_id = speakerID
#out_path = input('Path to save: ')
out_path = "output.wav"
with no_grad():
x_tst = stn_tst.unsqueeze(0)
x_tst_lengths = LongTensor([stn_tst.size(0)])
sid = LongTensor([speaker_id])
audio = net_g_ms.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale,
noise_scale_w=noise_scale_w, length_scale=length_scale)[0][0, 0].data.cpu().float().numpy()
elif choice == 'v':
audio, out_path = voice_conversion()
write(out_path, hps_ms.data.sampling_rate, audio)
print('Successfully saved!')
#ask_if_continue()
else:
import os
import librosa
import numpy as np
from torch import FloatTensor
import audonnx
w2v2_folder = input('Path of a w2v2 dimensional emotion model: ')
w2v2_model = audonnx.load(os.path.dirname(w2v2_folder))
#while True:
if(1==1):
#choice = input('TTS or VC? (t/v):')
choice = 't'
if choice == 't':
#text = input('Text to read: ')
text = inputString
if text == '[ADVANCED]':
#text = input('Raw text:')
text = "我不会说"
#print('Cleaned text is:')
#ex_print(_clean_text(
# text, hps_ms.data.text_cleaners), escape)
#continue
length_scale, text = get_label_value(
text, 'LENGTH', 1, 'length scale')
noise_scale, text = get_label_value(
text, 'NOISE', 0.667, 'noise scale')
noise_scale_w, text = get_label_value(
text, 'NOISEW', 0.8, 'deviation of noise')
cleaned, text = get_label(text, 'CLEANED')
stn_tst = get_text(text, hps_ms, cleaned=cleaned)
#print_speakers(speakers, escape)
#speaker_id = get_speaker_id('Speaker ID: ')
speaker_id = speakerID
emotion_reference = input('Path of an emotion reference: ')
if emotion_reference.endswith('.npy'):
emotion = np.load(emotion_reference)
emotion = FloatTensor(emotion).unsqueeze(0)
else:
audio16000, sampling_rate = librosa.load(
emotion_reference, sr=16000, mono=True)
emotion = w2v2_model(audio16000, sampling_rate)[
'hidden_states']
emotion_reference = re.sub(
r'\..*$', '', emotion_reference)
np.save(emotion_reference, emotion.squeeze(0))
emotion = FloatTensor(emotion)
#out_path = input('Path to save: ')
out_path = "output.wav"
with no_grad():
x_tst = stn_tst.unsqueeze(0)
x_tst_lengths = LongTensor([stn_tst.size(0)])
sid = LongTensor([speaker_id])
audio = net_g_ms.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale, noise_scale_w=noise_scale_w,
length_scale=length_scale, emotion_embedding=emotion)[0][0, 0].data.cpu().float().numpy()
elif choice == 'v':
audio, out_path = voice_conversion()
write(out_path, hps_ms.data.sampling_rate, audio)
print('Successfully saved!')
#ask_if_continue()
else:
model = input('Path of a hubert-soft model: ')
from hubert_model import hubert_soft
hubert = hubert_soft(model)
while True:
audio_path = input('Path of an audio file to convert:\n')
if audio_path != '[VC]':
import librosa
if use_f0:
audio, sampling_rate = librosa.load(
audio_path, sr=hps_ms.data.sampling_rate, mono=True)
audio16000 = librosa.resample(
audio, orig_sr=sampling_rate, target_sr=16000)
else:
audio16000, sampling_rate = librosa.load(
audio_path, sr=16000, mono=True)
#print_speakers(speakers, escape)
target_id = get_speaker_id('Target speaker ID: ')
out_path = input('Path to save: ')
length_scale, out_path = get_label_value(
out_path, 'LENGTH', 1, 'length scale')
noise_scale, out_path = get_label_value(
out_path, 'NOISE', 0.1, 'noise scale')
noise_scale_w, out_path = get_label_value(
out_path, 'NOISEW', 0.1, 'deviation of noise')
from torch import inference_mode, FloatTensor
import numpy as np
with inference_mode():
units = hubert.units(FloatTensor(audio16000).unsqueeze(
0).unsqueeze(0)).squeeze(0).numpy()
if use_f0:
f0_scale, out_path = get_label_value(
out_path, 'F0', 1, 'f0 scale')
f0 = librosa.pyin(audio, sr=sampling_rate,
fmin=librosa.note_to_hz('C0'),
fmax=librosa.note_to_hz('C7'),
frame_length=1780)[0]
target_length = len(units[:, 0])
f0 = np.nan_to_num(np.interp(np.arange(0, len(f0)*target_length, len(f0))/target_length,
np.arange(0, len(f0)), f0)) * f0_scale
units[:, 0] = f0 / 10
stn_tst = FloatTensor(units)
with no_grad():
x_tst = stn_tst.unsqueeze(0)
x_tst_lengths = LongTensor([stn_tst.size(0)])
sid = LongTensor([target_id])
audio = net_g_ms.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=noise_scale,
noise_scale_w=noise_scale_w, length_scale=length_scale)[0][0, 0].data.float().numpy()
else:
audio, out_path = voice_conversion()
write(out_path, hps_ms.data.sampling_rate, audio)
print('Successfully saved!')
#ask_if_continue()
if __name__ == "__main__":
session_token = get_token()
api = ChatGPT(session_token)
bot = miraicle.Mirai(qq=qq, verify_key=verify_key, port=port)
print(idmessage)
peaker_id = input()
os_dir=os.getcwd()
@miraicle.Mirai.receiver('GroupMessage')
def reply_to_group(bot: miraicle.Mirai, msg: miraicle.GroupMessage):
global sound_status
if msg.at_me() and msg.plain.lstrip()!="" and sound_status==False:
if msg.plain.lstrip() == "重置对话":
api.reset_conversation()
bot.send_group_msg(group=msg.group, msg="重置对话成功")
else:
return_voice = True
sound_status = True
bot.send_group_msg(group=msg.group, msg="正在处理中请稍后...")
text = msg.plain.strip()
resp = api.send_message(text)
answer = resp["message"].replace('\n', '')
if re.match('[\u4e00-\u9fa5]',answer):
generateSound("[ZH]" + answer + "[ZH]",language="ch")
else:
generateSound(answer,language="jp")
#generateSound(answer,language="multi")
if False:
try:
os.remove('output.pcm')
os.remove('output.silk')
except:
pass
ffmpeg=os.popen(r'"D:\Program Files (x86)\ffmpeg\bin\ffmpeg.exe" -i '+os_dir+'"\output.wav" -f s16le -ar 24000 -ac 1 -acodec pcm_s16le '+os_dir+'"\output.pcm" -loglevel quiet')
#print(ffmpeg)
while return_voice:
if os.path.exists('output.pcm'):
return_voice=False
return_voice = True
voice_status=os.system(os_dir+r'\silk_v3_encoder.exe '+os_dir+ r'\output.pcm '+os_dir+r'\output.silk -tencent -quiet')
while return_voice:
if voice_status==0:
return_voice=False
sound_win=miraicle.Voice(base64='output.silk')
else:
sound = miraicle.Voice(base64='output.wav')
print("ChatGPT:")
print(answer)
bot.send_group_msg(group=msg.group,msg=sound)
bot.send_group_msg(group=msg.group,msg=answer)
#PlaySound(r'.\output.wav', flags=1)
sound_status=False
elif msg.at_me() and msg.plain.lstrip()!="" and sound_status==True:
bot.send_group_msg(group=msg.group,msg="上条信息还没处理完，请等一下再试吧！")
elif msg.at_me() and msg.plain.lstrip()=="":
bot.send_group_msg(group=msg.group, msg="哎呦，你干嘛~")
else:
pass
bot.run()