Spaces:
Sleeping
Sleeping
TheStinger
commited on
Commit
•
43af790
1
Parent(s):
a6c3559
Upload 9 files
Browse files- .gitattributes +36 -2
- app.py +354 -1884
- gitattributes +36 -0
- model.index +3 -0
- model.pth +3 -0
- packages.txt +1 -3
- requirements.txt +10 -23
- test.ogg +0 -0
- tts_voice.py +230 -0
.gitattributes
CHANGED
@@ -1,2 +1,36 @@
|
|
1 |
-
|
2 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
*.7z filter=lfs diff=lfs merge=lfs -text
|
2 |
+
*.arrow filter=lfs diff=lfs merge=lfs -text
|
3 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
4 |
+
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
5 |
+
*.ckpt filter=lfs diff=lfs merge=lfs -text
|
6 |
+
*.ftz filter=lfs diff=lfs merge=lfs -text
|
7 |
+
*.gz filter=lfs diff=lfs merge=lfs -text
|
8 |
+
*.h5 filter=lfs diff=lfs merge=lfs -text
|
9 |
+
*.joblib filter=lfs diff=lfs merge=lfs -text
|
10 |
+
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
11 |
+
*.mlmodel filter=lfs diff=lfs merge=lfs -text
|
12 |
+
*.model filter=lfs diff=lfs merge=lfs -text
|
13 |
+
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
14 |
+
*.npy filter=lfs diff=lfs merge=lfs -text
|
15 |
+
*.npz filter=lfs diff=lfs merge=lfs -text
|
16 |
+
*.onnx filter=lfs diff=lfs merge=lfs -text
|
17 |
+
*.ot filter=lfs diff=lfs merge=lfs -text
|
18 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
19 |
+
*.pb filter=lfs diff=lfs merge=lfs -text
|
20 |
+
*.pickle filter=lfs diff=lfs merge=lfs -text
|
21 |
+
*.pkl filter=lfs diff=lfs merge=lfs -text
|
22 |
+
*.pt filter=lfs diff=lfs merge=lfs -text
|
23 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
24 |
+
*.rar filter=lfs diff=lfs merge=lfs -text
|
25 |
+
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
26 |
+
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
27 |
+
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
28 |
+
*.tar filter=lfs diff=lfs merge=lfs -text
|
29 |
+
*.tflite filter=lfs diff=lfs merge=lfs -text
|
30 |
+
*.tgz filter=lfs diff=lfs merge=lfs -text
|
31 |
+
*.wasm filter=lfs diff=lfs merge=lfs -text
|
32 |
+
*.xz filter=lfs diff=lfs merge=lfs -text
|
33 |
+
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
+
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
+
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
model.index filter=lfs diff=lfs merge=lfs -text
|
app.py
CHANGED
@@ -1,1884 +1,354 @@
|
|
1 |
-
import
|
2 |
-
|
3 |
-
|
4 |
-
import
|
5 |
-
|
6 |
-
|
7 |
-
import
|
8 |
-
import
|
9 |
-
from
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
import
|
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 |
-
for
|
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 |
-
file_index,
|
356 |
-
file_index2,
|
357 |
-
# file_big_npy,
|
358 |
-
index_rate,
|
359 |
-
filter_radius,
|
360 |
-
resample_sr,
|
361 |
-
rms_mix_rate,
|
362 |
-
protect,
|
363 |
-
format1,
|
364 |
-
crepe_hop_length,
|
365 |
-
):
|
366 |
-
try:
|
367 |
-
dir_path = (
|
368 |
-
dir_path.strip(" ").strip('"').strip("\n").strip('"').strip(" ")
|
369 |
-
) # 防止小白拷路径头尾带了空格和"和回车
|
370 |
-
opt_root = opt_root.strip(" ").strip('"').strip("\n").strip('"').strip(" ")
|
371 |
-
os.makedirs(opt_root, exist_ok=True)
|
372 |
-
try:
|
373 |
-
if dir_path != "":
|
374 |
-
paths = [os.path.join(dir_path, name) for name in os.listdir(dir_path)]
|
375 |
-
else:
|
376 |
-
paths = [path.name for path in paths]
|
377 |
-
except:
|
378 |
-
traceback.print_exc()
|
379 |
-
paths = [path.name for path in paths]
|
380 |
-
infos = []
|
381 |
-
for path in paths:
|
382 |
-
info, opt = vc_single(
|
383 |
-
sid,
|
384 |
-
path,
|
385 |
-
f0_up_key,
|
386 |
-
None,
|
387 |
-
f0_method,
|
388 |
-
file_index,
|
389 |
-
# file_big_npy,
|
390 |
-
index_rate,
|
391 |
-
filter_radius,
|
392 |
-
resample_sr,
|
393 |
-
rms_mix_rate,
|
394 |
-
protect,
|
395 |
-
crepe_hop_length
|
396 |
-
)
|
397 |
-
if "Success" in info:
|
398 |
-
try:
|
399 |
-
tgt_sr, audio_opt = opt
|
400 |
-
if format1 in ["wav", "flac"]:
|
401 |
-
sf.write(
|
402 |
-
"%s/%s.%s" % (opt_root, os.path.basename(path), format1),
|
403 |
-
audio_opt,
|
404 |
-
tgt_sr,
|
405 |
-
)
|
406 |
-
else:
|
407 |
-
path = "%s/%s.wav" % (opt_root, os.path.basename(path))
|
408 |
-
sf.write(
|
409 |
-
path,
|
410 |
-
audio_opt,
|
411 |
-
tgt_sr,
|
412 |
-
)
|
413 |
-
if os.path.exists(path):
|
414 |
-
os.system(
|
415 |
-
"ffmpeg -i %s -vn %s -q:a 2 -y"
|
416 |
-
% (path, path[:-4] + ".%s" % format1)
|
417 |
-
)
|
418 |
-
except:
|
419 |
-
info += traceback.format_exc()
|
420 |
-
infos.append("%s->%s" % (os.path.basename(path), info))
|
421 |
-
yield "\n".join(infos)
|
422 |
-
yield "\n".join(infos)
|
423 |
-
except:
|
424 |
-
yield traceback.format_exc()
|
425 |
-
|
426 |
-
# 一个选项卡全局只能有一个音色
|
427 |
-
def get_vc(sid):
|
428 |
-
global n_spk, tgt_sr, net_g, vc, cpt, version
|
429 |
-
if sid == "" or sid == []:
|
430 |
-
global hubert_model
|
431 |
-
if hubert_model != None: # 考虑到轮询, 需要加个判断看是否 sid 是由有模型切换到无模型的
|
432 |
-
print("clean_empty_cache")
|
433 |
-
del net_g, n_spk, vc, hubert_model, tgt_sr # ,cpt
|
434 |
-
hubert_model = net_g = n_spk = vc = hubert_model = tgt_sr = None
|
435 |
-
if torch.cuda.is_available():
|
436 |
-
torch.cuda.empty_cache()
|
437 |
-
###楼下不这么折腾清理不干净
|
438 |
-
if_f0 = cpt.get("f0", 1)
|
439 |
-
version = cpt.get("version", "v1")
|
440 |
-
if version == "v1":
|
441 |
-
if if_f0 == 1:
|
442 |
-
net_g = SynthesizerTrnMs256NSFsid(
|
443 |
-
*cpt["config"], is_half=config.is_half
|
444 |
-
)
|
445 |
-
else:
|
446 |
-
net_g = SynthesizerTrnMs256NSFsid_nono(*cpt["config"])
|
447 |
-
elif version == "v2":
|
448 |
-
if if_f0 == 1:
|
449 |
-
net_g = SynthesizerTrnMs768NSFsid(
|
450 |
-
*cpt["config"], is_half=config.is_half
|
451 |
-
)
|
452 |
-
else:
|
453 |
-
net_g = SynthesizerTrnMs768NSFsid_nono(*cpt["config"])
|
454 |
-
del net_g, cpt
|
455 |
-
if torch.cuda.is_available():
|
456 |
-
torch.cuda.empty_cache()
|
457 |
-
cpt = None
|
458 |
-
return {"visible": False, "__type__": "update"}
|
459 |
-
person = "%s/%s" % (weight_root, sid)
|
460 |
-
print("loading %s" % person)
|
461 |
-
cpt = torch.load(person, map_location="cpu")
|
462 |
-
tgt_sr = cpt["config"][-1]
|
463 |
-
cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0] # n_spk
|
464 |
-
if_f0 = cpt.get("f0", 1)
|
465 |
-
version = cpt.get("version", "v1")
|
466 |
-
if version == "v1":
|
467 |
-
if if_f0 == 1:
|
468 |
-
net_g = SynthesizerTrnMs256NSFsid(*cpt["config"], is_half=config.is_half)
|
469 |
-
else:
|
470 |
-
net_g = SynthesizerTrnMs256NSFsid_nono(*cpt["config"])
|
471 |
-
elif version == "v2":
|
472 |
-
if if_f0 == 1:
|
473 |
-
net_g = SynthesizerTrnMs768NSFsid(*cpt["config"], is_half=config.is_half)
|
474 |
-
else:
|
475 |
-
net_g = SynthesizerTrnMs768NSFsid_nono(*cpt["config"])
|
476 |
-
del net_g.enc_q
|
477 |
-
print(net_g.load_state_dict(cpt["weight"], strict=False))
|
478 |
-
net_g.eval().to(config.device)
|
479 |
-
if config.is_half:
|
480 |
-
net_g = net_g.half()
|
481 |
-
else:
|
482 |
-
net_g = net_g.float()
|
483 |
-
vc = VC(tgt_sr, config)
|
484 |
-
n_spk = cpt["config"][-3]
|
485 |
-
return {"visible": False, "maximum": n_spk, "__type__": "update"}
|
486 |
-
|
487 |
-
|
488 |
-
def change_choices():
|
489 |
-
names = []
|
490 |
-
for name in os.listdir(weight_root):
|
491 |
-
if name.endswith(".pth"):
|
492 |
-
names.append(name)
|
493 |
-
index_paths = []
|
494 |
-
for root, dirs, files in os.walk(index_root, topdown=False):
|
495 |
-
for name in files:
|
496 |
-
if name.endswith(".index") and "trained" not in name:
|
497 |
-
index_paths.append("%s/%s" % (root, name))
|
498 |
-
return {"choices": sorted(names), "__type__": "update"}, {
|
499 |
-
"choices": sorted(index_paths),
|
500 |
-
"__type__": "update",
|
501 |
-
}
|
502 |
-
|
503 |
-
|
504 |
-
def clean():
|
505 |
-
return {"value": "", "__type__": "update"}
|
506 |
-
|
507 |
-
|
508 |
-
sr_dict = {
|
509 |
-
"32k": 32000,
|
510 |
-
"40k": 40000,
|
511 |
-
"48k": 48000,
|
512 |
-
}
|
513 |
-
|
514 |
-
|
515 |
-
def if_done(done, p):
|
516 |
-
while 1:
|
517 |
-
if p.poll() == None:
|
518 |
-
sleep(0.5)
|
519 |
-
else:
|
520 |
-
break
|
521 |
-
done[0] = True
|
522 |
-
|
523 |
-
|
524 |
-
def if_done_multi(done, ps):
|
525 |
-
while 1:
|
526 |
-
# poll==None代表进程未结束
|
527 |
-
# 只要有一个进程未结束都不停
|
528 |
-
flag = 1
|
529 |
-
for p in ps:
|
530 |
-
if p.poll() == None:
|
531 |
-
flag = 0
|
532 |
-
sleep(0.5)
|
533 |
-
break
|
534 |
-
if flag == 1:
|
535 |
-
break
|
536 |
-
done[0] = True
|
537 |
-
|
538 |
-
|
539 |
-
def preprocess_dataset(trainset_dir, exp_dir, sr, n_p):
|
540 |
-
sr = sr_dict[sr]
|
541 |
-
os.makedirs("%s/logs/%s" % (now_dir, exp_dir), exist_ok=True)
|
542 |
-
f = open("%s/logs/%s/preprocess.log" % (now_dir, exp_dir), "w")
|
543 |
-
f.close()
|
544 |
-
cmd = (
|
545 |
-
config.python_cmd
|
546 |
-
+ " trainset_preprocess_pipeline_print.py %s %s %s %s/logs/%s "
|
547 |
-
% (trainset_dir, sr, n_p, now_dir, exp_dir)
|
548 |
-
+ str(config.noparallel)
|
549 |
-
)
|
550 |
-
print(cmd)
|
551 |
-
p = Popen(cmd, shell=True) # , stdin=PIPE, stdout=PIPE,stderr=PIPE,cwd=now_dir
|
552 |
-
###煞笔gr, popen read都非得全跑完了再一次性读取, 不用gr就正常读一句输出一句;只能额外弄出一个文本流定时读
|
553 |
-
done = [False]
|
554 |
-
threading.Thread(
|
555 |
-
target=if_done,
|
556 |
-
args=(
|
557 |
-
done,
|
558 |
-
p,
|
559 |
-
),
|
560 |
-
).start()
|
561 |
-
while 1:
|
562 |
-
with open("%s/logs/%s/preprocess.log" % (now_dir, exp_dir), "r") as f:
|
563 |
-
yield (f.read())
|
564 |
-
sleep(1)
|
565 |
-
if done[0] == True:
|
566 |
-
break
|
567 |
-
with open("%s/logs/%s/preprocess.log" % (now_dir, exp_dir), "r") as f:
|
568 |
-
log = f.read()
|
569 |
-
print(log)
|
570 |
-
yield log
|
571 |
-
|
572 |
-
# but2.click(extract_f0,[gpus6,np7,f0method8,if_f0_3,trainset_dir4],[info2])
|
573 |
-
def extract_f0_feature(gpus, n_p, f0method, if_f0, exp_dir, version19, echl):
|
574 |
-
gpus = gpus.split("-")
|
575 |
-
os.makedirs("%s/logs/%s" % (now_dir, exp_dir), exist_ok=True)
|
576 |
-
f = open("%s/logs/%s/extract_f0_feature.log" % (now_dir, exp_dir), "w")
|
577 |
-
f.close()
|
578 |
-
if if_f0:
|
579 |
-
cmd = config.python_cmd + " extract_f0_print.py %s/logs/%s %s %s %s" % (
|
580 |
-
now_dir,
|
581 |
-
exp_dir,
|
582 |
-
n_p,
|
583 |
-
f0method,
|
584 |
-
echl,
|
585 |
-
)
|
586 |
-
print(cmd)
|
587 |
-
p = Popen(cmd, shell=True, cwd=now_dir) # , stdin=PIPE, stdout=PIPE,stderr=PIPE
|
588 |
-
###煞笔gr, popen read都非得全跑完了再一次性读取, 不用gr就正常读一句输出一句;只能额外弄出一个文本流定时读
|
589 |
-
done = [False]
|
590 |
-
threading.Thread(
|
591 |
-
target=if_done,
|
592 |
-
args=(
|
593 |
-
done,
|
594 |
-
p,
|
595 |
-
),
|
596 |
-
).start()
|
597 |
-
while 1:
|
598 |
-
with open(
|
599 |
-
"%s/logs/%s/extract_f0_feature.log" % (now_dir, exp_dir), "r"
|
600 |
-
) as f:
|
601 |
-
yield (f.read())
|
602 |
-
sleep(1)
|
603 |
-
if done[0] == True:
|
604 |
-
break
|
605 |
-
with open("%s/logs/%s/extract_f0_feature.log" % (now_dir, exp_dir), "r") as f:
|
606 |
-
log = f.read()
|
607 |
-
print(log)
|
608 |
-
yield log
|
609 |
-
####对不同part分别开多进程
|
610 |
-
"""
|
611 |
-
n_part=int(sys.argv[1])
|
612 |
-
i_part=int(sys.argv[2])
|
613 |
-
i_gpu=sys.argv[3]
|
614 |
-
exp_dir=sys.argv[4]
|
615 |
-
os.environ["CUDA_VISIBLE_DEVICES"]=str(i_gpu)
|
616 |
-
"""
|
617 |
-
leng = len(gpus)
|
618 |
-
ps = []
|
619 |
-
for idx, n_g in enumerate(gpus):
|
620 |
-
cmd = (
|
621 |
-
config.python_cmd
|
622 |
-
+ " extract_feature_print.py %s %s %s %s %s/logs/%s %s"
|
623 |
-
% (
|
624 |
-
config.device,
|
625 |
-
leng,
|
626 |
-
idx,
|
627 |
-
n_g,
|
628 |
-
now_dir,
|
629 |
-
exp_dir,
|
630 |
-
version19,
|
631 |
-
)
|
632 |
-
)
|
633 |
-
print(cmd)
|
634 |
-
p = Popen(
|
635 |
-
cmd, shell=True, cwd=now_dir
|
636 |
-
) # , shell=True, stdin=PIPE, stdout=PIPE, stderr=PIPE, cwd=now_dir
|
637 |
-
ps.append(p)
|
638 |
-
###煞笔gr, popen read都非得全跑完了再一次性读取, 不用gr就正常读一句输出一句;只能额外弄出一个文本流定时读
|
639 |
-
done = [False]
|
640 |
-
threading.Thread(
|
641 |
-
target=if_done_multi,
|
642 |
-
args=(
|
643 |
-
done,
|
644 |
-
ps,
|
645 |
-
),
|
646 |
-
).start()
|
647 |
-
while 1:
|
648 |
-
with open("%s/logs/%s/extract_f0_feature.log" % (now_dir, exp_dir), "r") as f:
|
649 |
-
yield (f.read())
|
650 |
-
sleep(1)
|
651 |
-
if done[0] == True:
|
652 |
-
break
|
653 |
-
with open("%s/logs/%s/extract_f0_feature.log" % (now_dir, exp_dir), "r") as f:
|
654 |
-
log = f.read()
|
655 |
-
print(log)
|
656 |
-
yield log
|
657 |
-
|
658 |
-
|
659 |
-
def change_sr2(sr2, if_f0_3, version19):
|
660 |
-
path_str = "" if version19 == "v1" else "_v2"
|
661 |
-
f0_str = "f0" if if_f0_3 else ""
|
662 |
-
if_pretrained_generator_exist = os.access("pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2), os.F_OK)
|
663 |
-
if_pretrained_discriminator_exist = os.access("pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2), os.F_OK)
|
664 |
-
if (if_pretrained_generator_exist == False):
|
665 |
-
print("pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2), "not exist, will not use pretrained model")
|
666 |
-
if (if_pretrained_discriminator_exist == False):
|
667 |
-
print("pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2), "not exist, will not use pretrained model")
|
668 |
-
return (
|
669 |
-
("pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2)) if if_pretrained_generator_exist else "",
|
670 |
-
("pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2)) if if_pretrained_discriminator_exist else "",
|
671 |
-
{"visible": True, "__type__": "update"}
|
672 |
-
)
|
673 |
-
|
674 |
-
def change_version19(sr2, if_f0_3, version19):
|
675 |
-
path_str = "" if version19 == "v1" else "_v2"
|
676 |
-
f0_str = "f0" if if_f0_3 else ""
|
677 |
-
if_pretrained_generator_exist = os.access("pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2), os.F_OK)
|
678 |
-
if_pretrained_discriminator_exist = os.access("pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2), os.F_OK)
|
679 |
-
if (if_pretrained_generator_exist == False):
|
680 |
-
print("pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2), "not exist, will not use pretrained model")
|
681 |
-
if (if_pretrained_discriminator_exist == False):
|
682 |
-
print("pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2), "not exist, will not use pretrained model")
|
683 |
-
return (
|
684 |
-
("pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2)) if if_pretrained_generator_exist else "",
|
685 |
-
("pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2)) if if_pretrained_discriminator_exist else "",
|
686 |
-
)
|
687 |
-
|
688 |
-
|
689 |
-
def change_f0(if_f0_3, sr2, version19): # f0method8,pretrained_G14,pretrained_D15
|
690 |
-
path_str = "" if version19 == "v1" else "_v2"
|
691 |
-
if_pretrained_generator_exist = os.access("pretrained%s/f0G%s.pth" % (path_str, sr2), os.F_OK)
|
692 |
-
if_pretrained_discriminator_exist = os.access("pretrained%s/f0D%s.pth" % (path_str, sr2), os.F_OK)
|
693 |
-
if (if_pretrained_generator_exist == False):
|
694 |
-
print("pretrained%s/f0G%s.pth" % (path_str, sr2), "not exist, will not use pretrained model")
|
695 |
-
if (if_pretrained_discriminator_exist == False):
|
696 |
-
print("pretrained%s/f0D%s.pth" % (path_str, sr2), "not exist, will not use pretrained model")
|
697 |
-
if if_f0_3:
|
698 |
-
return (
|
699 |
-
{"visible": True, "__type__": "update"},
|
700 |
-
"pretrained%s/f0G%s.pth" % (path_str, sr2) if if_pretrained_generator_exist else "",
|
701 |
-
"pretrained%s/f0D%s.pth" % (path_str, sr2) if if_pretrained_discriminator_exist else "",
|
702 |
-
)
|
703 |
-
return (
|
704 |
-
{"visible": False, "__type__": "update"},
|
705 |
-
("pretrained%s/G%s.pth" % (path_str, sr2)) if if_pretrained_generator_exist else "",
|
706 |
-
("pretrained%s/D%s.pth" % (path_str, sr2)) if if_pretrained_discriminator_exist else "",
|
707 |
-
)
|
708 |
-
|
709 |
-
|
710 |
-
global log_interval
|
711 |
-
|
712 |
-
|
713 |
-
def set_log_interval(exp_dir, batch_size12):
|
714 |
-
log_interval = 1
|
715 |
-
|
716 |
-
folder_path = os.path.join(exp_dir, "1_16k_wavs")
|
717 |
-
|
718 |
-
if os.path.exists(folder_path) and os.path.isdir(folder_path):
|
719 |
-
wav_files = [f for f in os.listdir(folder_path) if f.endswith(".wav")]
|
720 |
-
if wav_files:
|
721 |
-
sample_size = len(wav_files)
|
722 |
-
log_interval = math.ceil(sample_size / batch_size12)
|
723 |
-
if log_interval > 1:
|
724 |
-
log_interval += 1
|
725 |
-
return log_interval
|
726 |
-
|
727 |
-
# but3.click(click_train,[exp_dir1,sr2,if_f0_3,save_epoch10,total_epoch11,batch_size12,if_save_latest13,pretrained_G14,pretrained_D15,gpus16])
|
728 |
-
def click_train(
|
729 |
-
exp_dir1,
|
730 |
-
sr2,
|
731 |
-
if_f0_3,
|
732 |
-
spk_id5,
|
733 |
-
save_epoch10,
|
734 |
-
total_epoch11,
|
735 |
-
batch_size12,
|
736 |
-
if_save_latest13,
|
737 |
-
pretrained_G14,
|
738 |
-
pretrained_D15,
|
739 |
-
gpus16,
|
740 |
-
if_cache_gpu17,
|
741 |
-
if_save_every_weights18,
|
742 |
-
version19,
|
743 |
-
):
|
744 |
-
CSVutil('csvdb/stop.csv', 'w+', 'formanting', False)
|
745 |
-
# 生成filelist
|
746 |
-
exp_dir = "%s/logs/%s" % (now_dir, exp_dir1)
|
747 |
-
os.makedirs(exp_dir, exist_ok=True)
|
748 |
-
gt_wavs_dir = "%s/0_gt_wavs" % (exp_dir)
|
749 |
-
feature_dir = (
|
750 |
-
"%s/3_feature256" % (exp_dir)
|
751 |
-
if version19 == "v1"
|
752 |
-
else "%s/3_feature768" % (exp_dir)
|
753 |
-
)
|
754 |
-
|
755 |
-
log_interval = set_log_interval(exp_dir, batch_size12)
|
756 |
-
|
757 |
-
if if_f0_3:
|
758 |
-
f0_dir = "%s/2a_f0" % (exp_dir)
|
759 |
-
f0nsf_dir = "%s/2b-f0nsf" % (exp_dir)
|
760 |
-
names = (
|
761 |
-
set([name.split(".")[0] for name in os.listdir(gt_wavs_dir)])
|
762 |
-
& set([name.split(".")[0] for name in os.listdir(feature_dir)])
|
763 |
-
& set([name.split(".")[0] for name in os.listdir(f0_dir)])
|
764 |
-
& set([name.split(".")[0] for name in os.listdir(f0nsf_dir)])
|
765 |
-
)
|
766 |
-
else:
|
767 |
-
names = set([name.split(".")[0] for name in os.listdir(gt_wavs_dir)]) & set(
|
768 |
-
[name.split(".")[0] for name in os.listdir(feature_dir)]
|
769 |
-
)
|
770 |
-
opt = []
|
771 |
-
for name in names:
|
772 |
-
if if_f0_3:
|
773 |
-
opt.append(
|
774 |
-
"%s/%s.wav|%s/%s.npy|%s/%s.wav.npy|%s/%s.wav.npy|%s"
|
775 |
-
% (
|
776 |
-
gt_wavs_dir.replace("\\", "\\\\"),
|
777 |
-
name,
|
778 |
-
feature_dir.replace("\\", "\\\\"),
|
779 |
-
name,
|
780 |
-
f0_dir.replace("\\", "\\\\"),
|
781 |
-
name,
|
782 |
-
f0nsf_dir.replace("\\", "\\\\"),
|
783 |
-
name,
|
784 |
-
spk_id5,
|
785 |
-
)
|
786 |
-
)
|
787 |
-
else:
|
788 |
-
opt.append(
|
789 |
-
"%s/%s.wav|%s/%s.npy|%s"
|
790 |
-
% (
|
791 |
-
gt_wavs_dir.replace("\\", "\\\\"),
|
792 |
-
name,
|
793 |
-
feature_dir.replace("\\", "\\\\"),
|
794 |
-
name,
|
795 |
-
spk_id5,
|
796 |
-
)
|
797 |
-
)
|
798 |
-
fea_dim = 256 if version19 == "v1" else 768
|
799 |
-
if if_f0_3:
|
800 |
-
for _ in range(2):
|
801 |
-
opt.append(
|
802 |
-
"%s/logs/mute/0_gt_wavs/mute%s.wav|%s/logs/mute/3_feature%s/mute.npy|%s/logs/mute/2a_f0/mute.wav.npy|%s/logs/mute/2b-f0nsf/mute.wav.npy|%s"
|
803 |
-
% (now_dir, sr2, now_dir, fea_dim, now_dir, now_dir, spk_id5)
|
804 |
-
)
|
805 |
-
else:
|
806 |
-
for _ in range(2):
|
807 |
-
opt.append(
|
808 |
-
"%s/logs/mute/0_gt_wavs/mute%s.wav|%s/logs/mute/3_feature%s/mute.npy|%s"
|
809 |
-
% (now_dir, sr2, now_dir, fea_dim, spk_id5)
|
810 |
-
)
|
811 |
-
shuffle(opt)
|
812 |
-
with open("%s/filelist.txt" % exp_dir, "w") as f:
|
813 |
-
f.write("\n".join(opt))
|
814 |
-
print("write filelist done")
|
815 |
-
# 生成config#无需生成config
|
816 |
-
# cmd = python_cmd + " train_nsf_sim_cache_sid_load_pretrain.py -e mi-test -sr 40k -f0 1 -bs 4 -g 0 -te 10 -se 5 -pg pretrained/f0G40k.pth -pd pretrained/f0D40k.pth -l 1 -c 0"
|
817 |
-
print("use gpus:", gpus16)
|
818 |
-
if pretrained_G14 == "":
|
819 |
-
print("no pretrained Generator")
|
820 |
-
if pretrained_D15 == "":
|
821 |
-
print("no pretrained Discriminator")
|
822 |
-
if gpus16:
|
823 |
-
cmd = (
|
824 |
-
config.python_cmd
|
825 |
-
+ " train_nsf_sim_cache_sid_load_pretrain.py -e %s -sr %s -f0 %s -bs %s -g %s -te %s -se %s %s %s -l %s -c %s -sw %s -v %s -li %s"
|
826 |
-
% (
|
827 |
-
exp_dir1,
|
828 |
-
sr2,
|
829 |
-
1 if if_f0_3 else 0,
|
830 |
-
batch_size12,
|
831 |
-
gpus16,
|
832 |
-
total_epoch11,
|
833 |
-
save_epoch10,
|
834 |
-
("-pg %s" % pretrained_G14) if pretrained_G14 != "" else "",
|
835 |
-
("-pd %s" % pretrained_D15) if pretrained_D15 != "" else "",
|
836 |
-
1 if if_save_latest13 == True else 0,
|
837 |
-
1 if if_cache_gpu17 == True else 0,
|
838 |
-
1 if if_save_every_weights18 == True else 0,
|
839 |
-
version19,
|
840 |
-
log_interval,
|
841 |
-
)
|
842 |
-
)
|
843 |
-
else:
|
844 |
-
cmd = (
|
845 |
-
config.python_cmd
|
846 |
-
+ " train_nsf_sim_cache_sid_load_pretrain.py -e %s -sr %s -f0 %s -bs %s -te %s -se %s %s %s -l %s -c %s -sw %s -v %s -li %s"
|
847 |
-
% (
|
848 |
-
exp_dir1,
|
849 |
-
sr2,
|
850 |
-
1 if if_f0_3 else 0,
|
851 |
-
batch_size12,
|
852 |
-
total_epoch11,
|
853 |
-
save_epoch10,
|
854 |
-
("-pg %s" % pretrained_G14) if pretrained_G14 != "" else "\b",
|
855 |
-
("-pd %s" % pretrained_D15) if pretrained_D15 != "" else "\b",
|
856 |
-
1 if if_save_latest13 == True else 0,
|
857 |
-
1 if if_cache_gpu17 == True else 0,
|
858 |
-
1 if if_save_every_weights18 == True else 0,
|
859 |
-
version19,
|
860 |
-
log_interval,
|
861 |
-
)
|
862 |
-
)
|
863 |
-
print(cmd)
|
864 |
-
p = Popen(cmd, shell=True, cwd=now_dir)
|
865 |
-
global PID
|
866 |
-
PID = p.pid
|
867 |
-
p.wait()
|
868 |
-
return ("训练结束, 您可查看控制台训练日志或实验文件夹下的train.log", {"visible": False, "__type__": "update"}, {"visible": True, "__type__": "update"})
|
869 |
-
|
870 |
-
|
871 |
-
# but4.click(train_index, [exp_dir1], info3)
|
872 |
-
def train_index(exp_dir1, version19):
|
873 |
-
exp_dir = "%s/logs/%s" % (now_dir, exp_dir1)
|
874 |
-
os.makedirs(exp_dir, exist_ok=True)
|
875 |
-
feature_dir = (
|
876 |
-
"%s/3_feature256" % (exp_dir)
|
877 |
-
if version19 == "v1"
|
878 |
-
else "%s/3_feature768" % (exp_dir)
|
879 |
-
)
|
880 |
-
if os.path.exists(feature_dir) == False:
|
881 |
-
return "请先进行特征提取!"
|
882 |
-
listdir_res = list(os.listdir(feature_dir))
|
883 |
-
if len(listdir_res) == 0:
|
884 |
-
return "请先进行特征提取!"
|
885 |
-
npys = []
|
886 |
-
for name in sorted(listdir_res):
|
887 |
-
phone = np.load("%s/%s" % (feature_dir, name))
|
888 |
-
npys.append(phone)
|
889 |
-
big_npy = np.concatenate(npys, 0)
|
890 |
-
big_npy_idx = np.arange(big_npy.shape[0])
|
891 |
-
np.random.shuffle(big_npy_idx)
|
892 |
-
big_npy = big_npy[big_npy_idx]
|
893 |
-
np.save("%s/total_fea.npy" % exp_dir, big_npy)
|
894 |
-
# n_ivf = big_npy.shape[0] // 39
|
895 |
-
n_ivf = min(int(16 * np.sqrt(big_npy.shape[0])), big_npy.shape[0] // 39)
|
896 |
-
infos = []
|
897 |
-
infos.append("%s,%s" % (big_npy.shape, n_ivf))
|
898 |
-
yield "\n".join(infos)
|
899 |
-
index = faiss.index_factory(256 if version19 == "v1" else 768, "IVF%s,Flat" % n_ivf)
|
900 |
-
# index = faiss.index_factory(256if version19=="v1"else 768, "IVF%s,PQ128x4fs,RFlat"%n_ivf)
|
901 |
-
infos.append("training")
|
902 |
-
yield "\n".join(infos)
|
903 |
-
index_ivf = faiss.extract_index_ivf(index) #
|
904 |
-
index_ivf.nprobe = 1
|
905 |
-
index.train(big_npy)
|
906 |
-
faiss.write_index(
|
907 |
-
index,
|
908 |
-
"%s/trained_IVF%s_Flat_nprobe_%s_%s_%s.index"
|
909 |
-
% (exp_dir, n_ivf, index_ivf.nprobe, exp_dir1, version19),
|
910 |
-
)
|
911 |
-
# faiss.write_index(index, '%s/trained_IVF%s_Flat_FastScan_%s.index'%(exp_dir,n_ivf,version19))
|
912 |
-
infos.append("adding")
|
913 |
-
yield "\n".join(infos)
|
914 |
-
batch_size_add = 8192
|
915 |
-
for i in range(0, big_npy.shape[0], batch_size_add):
|
916 |
-
index.add(big_npy[i : i + batch_size_add])
|
917 |
-
faiss.write_index(
|
918 |
-
index,
|
919 |
-
"%s/added_IVF%s_Flat_nprobe_%s_%s_%s.index"
|
920 |
-
% (exp_dir, n_ivf, index_ivf.nprobe, exp_dir1, version19),
|
921 |
-
)
|
922 |
-
infos.append(
|
923 |
-
"成功构建索引,added_IVF%s_Flat_nprobe_%s_%s_%s.index"
|
924 |
-
% (n_ivf, index_ivf.nprobe, exp_dir1, version19)
|
925 |
-
)
|
926 |
-
# faiss.write_index(index, '%s/added_IVF%s_Flat_FastScan_%s.index'%(exp_dir,n_ivf,version19))
|
927 |
-
# infos.append("成功构建索引,added_IVF%s_Flat_FastScan_%s.index"%(n_ivf,version19))
|
928 |
-
yield "\n".join(infos)
|
929 |
-
|
930 |
-
|
931 |
-
# but5.click(train1key, [exp_dir1, sr2, if_f0_3, trainset_dir4, spk_id5, gpus6, np7, f0method8, save_epoch10, total_epoch11, batch_size12, if_save_latest13, pretrained_G14, pretrained_D15, gpus16, if_cache_gpu17], info3)
|
932 |
-
def train1key(
|
933 |
-
exp_dir1,
|
934 |
-
sr2,
|
935 |
-
if_f0_3,
|
936 |
-
trainset_dir4,
|
937 |
-
spk_id5,
|
938 |
-
np7,
|
939 |
-
f0method8,
|
940 |
-
save_epoch10,
|
941 |
-
total_epoch11,
|
942 |
-
batch_size12,
|
943 |
-
if_save_latest13,
|
944 |
-
pretrained_G14,
|
945 |
-
pretrained_D15,
|
946 |
-
gpus16,
|
947 |
-
if_cache_gpu17,
|
948 |
-
if_save_every_weights18,
|
949 |
-
version19,
|
950 |
-
echl
|
951 |
-
):
|
952 |
-
infos = []
|
953 |
-
|
954 |
-
def get_info_str(strr):
|
955 |
-
infos.append(strr)
|
956 |
-
return "\n".join(infos)
|
957 |
-
|
958 |
-
model_log_dir = "%s/logs/%s" % (now_dir, exp_dir1)
|
959 |
-
preprocess_log_path = "%s/preprocess.log" % model_log_dir
|
960 |
-
extract_f0_feature_log_path = "%s/extract_f0_feature.log" % model_log_dir
|
961 |
-
gt_wavs_dir = "%s/0_gt_wavs" % model_log_dir
|
962 |
-
feature_dir = (
|
963 |
-
"%s/3_feature256" % model_log_dir
|
964 |
-
if version19 == "v1"
|
965 |
-
else "%s/3_feature768" % model_log_dir
|
966 |
-
)
|
967 |
-
|
968 |
-
os.makedirs(model_log_dir, exist_ok=True)
|
969 |
-
#########step1:处理数据
|
970 |
-
open(preprocess_log_path, "w").close()
|
971 |
-
cmd = (
|
972 |
-
config.python_cmd
|
973 |
-
+ " trainset_preprocess_pipeline_print.py %s %s %s %s "
|
974 |
-
% (trainset_dir4, sr_dict[sr2], np7, model_log_dir)
|
975 |
-
+ str(config.noparallel)
|
976 |
-
)
|
977 |
-
yield get_info_str(i18n("step1:正在处理数据"))
|
978 |
-
yield get_info_str(cmd)
|
979 |
-
p = Popen(cmd, shell=True)
|
980 |
-
p.wait()
|
981 |
-
with open(preprocess_log_path, "r") as f:
|
982 |
-
print(f.read())
|
983 |
-
#########step2a:提取音高
|
984 |
-
open(extract_f0_feature_log_path, "w")
|
985 |
-
if if_f0_3:
|
986 |
-
yield get_info_str("step2a:正在提取音高")
|
987 |
-
cmd = config.python_cmd + " extract_f0_print.py %s %s %s %s" % (
|
988 |
-
model_log_dir,
|
989 |
-
np7,
|
990 |
-
f0method8,
|
991 |
-
echl
|
992 |
-
)
|
993 |
-
yield get_info_str(cmd)
|
994 |
-
p = Popen(cmd, shell=True, cwd=now_dir)
|
995 |
-
p.wait()
|
996 |
-
with open(extract_f0_feature_log_path, "r") as f:
|
997 |
-
print(f.read())
|
998 |
-
else:
|
999 |
-
yield get_info_str(i18n("step2a:无需提取音高"))
|
1000 |
-
#######step2b:提取特征
|
1001 |
-
yield get_info_str(i18n("step2b:正在提取特征"))
|
1002 |
-
gpus = gpus16.split("-")
|
1003 |
-
leng = len(gpus)
|
1004 |
-
ps = []
|
1005 |
-
for idx, n_g in enumerate(gpus):
|
1006 |
-
cmd = config.python_cmd + " extract_feature_print.py %s %s %s %s %s %s" % (
|
1007 |
-
config.device,
|
1008 |
-
leng,
|
1009 |
-
idx,
|
1010 |
-
n_g,
|
1011 |
-
model_log_dir,
|
1012 |
-
version19,
|
1013 |
-
)
|
1014 |
-
yield get_info_str(cmd)
|
1015 |
-
p = Popen(
|
1016 |
-
cmd, shell=True, cwd=now_dir
|
1017 |
-
) # , shell=True, stdin=PIPE, stdout=PIPE, stderr=PIPE, cwd=now_dir
|
1018 |
-
ps.append(p)
|
1019 |
-
for p in ps:
|
1020 |
-
p.wait()
|
1021 |
-
with open(extract_f0_feature_log_path, "r") as f:
|
1022 |
-
print(f.read())
|
1023 |
-
#######step3a:训练模型
|
1024 |
-
yield get_info_str(i18n("step3a:正在训练模型"))
|
1025 |
-
# 生成filelist
|
1026 |
-
if if_f0_3:
|
1027 |
-
f0_dir = "%s/2a_f0" % model_log_dir
|
1028 |
-
f0nsf_dir = "%s/2b-f0nsf" % model_log_dir
|
1029 |
-
names = (
|
1030 |
-
set([name.split(".")[0] for name in os.listdir(gt_wavs_dir)])
|
1031 |
-
& set([name.split(".")[0] for name in os.listdir(feature_dir)])
|
1032 |
-
& set([name.split(".")[0] for name in os.listdir(f0_dir)])
|
1033 |
-
& set([name.split(".")[0] for name in os.listdir(f0nsf_dir)])
|
1034 |
-
)
|
1035 |
-
else:
|
1036 |
-
names = set([name.split(".")[0] for name in os.listdir(gt_wavs_dir)]) & set(
|
1037 |
-
[name.split(".")[0] for name in os.listdir(feature_dir)]
|
1038 |
-
)
|
1039 |
-
opt = []
|
1040 |
-
for name in names:
|
1041 |
-
if if_f0_3:
|
1042 |
-
opt.append(
|
1043 |
-
"%s/%s.wav|%s/%s.npy|%s/%s.wav.npy|%s/%s.wav.npy|%s"
|
1044 |
-
% (
|
1045 |
-
gt_wavs_dir.replace("\\", "\\\\"),
|
1046 |
-
name,
|
1047 |
-
feature_dir.replace("\\", "\\\\"),
|
1048 |
-
name,
|
1049 |
-
f0_dir.replace("\\", "\\\\"),
|
1050 |
-
name,
|
1051 |
-
f0nsf_dir.replace("\\", "\\\\"),
|
1052 |
-
name,
|
1053 |
-
spk_id5,
|
1054 |
-
)
|
1055 |
-
)
|
1056 |
-
else:
|
1057 |
-
opt.append(
|
1058 |
-
"%s/%s.wav|%s/%s.npy|%s"
|
1059 |
-
% (
|
1060 |
-
gt_wavs_dir.replace("\\", "\\\\"),
|
1061 |
-
name,
|
1062 |
-
feature_dir.replace("\\", "\\\\"),
|
1063 |
-
name,
|
1064 |
-
spk_id5,
|
1065 |
-
)
|
1066 |
-
)
|
1067 |
-
fea_dim = 256 if version19 == "v1" else 768
|
1068 |
-
if if_f0_3:
|
1069 |
-
for _ in range(2):
|
1070 |
-
opt.append(
|
1071 |
-
"%s/logs/mute/0_gt_wavs/mute%s.wav|%s/logs/mute/3_feature%s/mute.npy|%s/logs/mute/2a_f0/mute.wav.npy|%s/logs/mute/2b-f0nsf/mute.wav.npy|%s"
|
1072 |
-
% (now_dir, sr2, now_dir, fea_dim, now_dir, now_dir, spk_id5)
|
1073 |
-
)
|
1074 |
-
else:
|
1075 |
-
for _ in range(2):
|
1076 |
-
opt.append(
|
1077 |
-
"%s/logs/mute/0_gt_wavs/mute%s.wav|%s/logs/mute/3_feature%s/mute.npy|%s"
|
1078 |
-
% (now_dir, sr2, now_dir, fea_dim, spk_id5)
|
1079 |
-
)
|
1080 |
-
shuffle(opt)
|
1081 |
-
with open("%s/filelist.txt" % model_log_dir, "w") as f:
|
1082 |
-
f.write("\n".join(opt))
|
1083 |
-
yield get_info_str("write filelist done")
|
1084 |
-
if gpus16:
|
1085 |
-
cmd = (
|
1086 |
-
config.python_cmd
|
1087 |
-
+" train_nsf_sim_cache_sid_load_pretrain.py -e %s -sr %s -f0 %s -bs %s -g %s -te %s -se %s %s %s -l %s -c %s -sw %s -v %s"
|
1088 |
-
% (
|
1089 |
-
exp_dir1,
|
1090 |
-
sr2,
|
1091 |
-
1 if if_f0_3 else 0,
|
1092 |
-
batch_size12,
|
1093 |
-
gpus16,
|
1094 |
-
total_epoch11,
|
1095 |
-
save_epoch10,
|
1096 |
-
("-pg %s" % pretrained_G14) if pretrained_G14 != "" else "",
|
1097 |
-
("-pd %s" % pretrained_D15) if pretrained_D15 != "" else "",
|
1098 |
-
1 if if_save_latest13 == True else 0,
|
1099 |
-
1 if if_cache_gpu17 == True else 0,
|
1100 |
-
1 if if_save_every_weights18 == True else 0,
|
1101 |
-
version19,
|
1102 |
-
)
|
1103 |
-
)
|
1104 |
-
else:
|
1105 |
-
cmd = (
|
1106 |
-
config.python_cmd
|
1107 |
-
+ " train_nsf_sim_cache_sid_load_pretrain.py -e %s -sr %s -f0 %s -bs %s -te %s -se %s %s %s -l %s -c %s -sw %s -v %s"
|
1108 |
-
% (
|
1109 |
-
exp_dir1,
|
1110 |
-
sr2,
|
1111 |
-
1 if if_f0_3 else 0,
|
1112 |
-
batch_size12,
|
1113 |
-
total_epoch11,
|
1114 |
-
save_epoch10,
|
1115 |
-
("-pg %s" % pretrained_G14) if pretrained_G14 != "" else "",
|
1116 |
-
("-pd %s" % pretrained_D15) if pretrained_D15 != "" else "",
|
1117 |
-
1 if if_save_latest13 == True else 0,
|
1118 |
-
1 if if_cache_gpu17 == True else 0,
|
1119 |
-
1 if if_save_every_weights18 == True else 0,
|
1120 |
-
version19,
|
1121 |
-
)
|
1122 |
-
)
|
1123 |
-
yield get_info_str(cmd)
|
1124 |
-
p = Popen(cmd, shell=True, cwd=now_dir)
|
1125 |
-
p.wait()
|
1126 |
-
yield get_info_str(i18n("训练结束, 您可查看控制台训练日志或实验文件夹下的train.log"))
|
1127 |
-
#######step3b:训练索引
|
1128 |
-
npys = []
|
1129 |
-
listdir_res = list(os.listdir(feature_dir))
|
1130 |
-
for name in sorted(listdir_res):
|
1131 |
-
phone = np.load("%s/%s" % (feature_dir, name))
|
1132 |
-
npys.append(phone)
|
1133 |
-
big_npy = np.concatenate(npys, 0)
|
1134 |
-
|
1135 |
-
big_npy_idx = np.arange(big_npy.shape[0])
|
1136 |
-
np.random.shuffle(big_npy_idx)
|
1137 |
-
big_npy = big_npy[big_npy_idx]
|
1138 |
-
np.save("%s/total_fea.npy" % model_log_dir, big_npy)
|
1139 |
-
|
1140 |
-
# n_ivf = big_npy.shape[0] // 39
|
1141 |
-
n_ivf = min(int(16 * np.sqrt(big_npy.shape[0])), big_npy.shape[0] // 39)
|
1142 |
-
yield get_info_str("%s,%s" % (big_npy.shape, n_ivf))
|
1143 |
-
index = faiss.index_factory(256 if version19 == "v1" else 768, "IVF%s,Flat" % n_ivf)
|
1144 |
-
yield get_info_str("training index")
|
1145 |
-
index_ivf = faiss.extract_index_ivf(index) #
|
1146 |
-
index_ivf.nprobe = 1
|
1147 |
-
index.train(big_npy)
|
1148 |
-
faiss.write_index(
|
1149 |
-
index,
|
1150 |
-
"%s/trained_IVF%s_Flat_nprobe_%s_%s_%s.index"
|
1151 |
-
% (model_log_dir, n_ivf, index_ivf.nprobe, exp_dir1, version19),
|
1152 |
-
)
|
1153 |
-
yield get_info_str("adding index")
|
1154 |
-
batch_size_add = 8192
|
1155 |
-
for i in range(0, big_npy.shape[0], batch_size_add):
|
1156 |
-
index.add(big_npy[i : i + batch_size_add])
|
1157 |
-
faiss.write_index(
|
1158 |
-
index,
|
1159 |
-
"%s/added_IVF%s_Flat_nprobe_%s_%s_%s.index"
|
1160 |
-
% (model_log_dir, n_ivf, index_ivf.nprobe, exp_dir1, version19),
|
1161 |
-
)
|
1162 |
-
yield get_info_str(
|
1163 |
-
"成功构建索引, added_IVF%s_Flat_nprobe_%s_%s_%s.index"
|
1164 |
-
% (n_ivf, index_ivf.nprobe, exp_dir1, version19)
|
1165 |
-
)
|
1166 |
-
yield get_info_str(i18n("全流程结束!"))
|
1167 |
-
|
1168 |
-
|
1169 |
-
def whethercrepeornah(radio):
|
1170 |
-
mango = True if radio == 'mangio-crepe' or radio == 'mangio-crepe-tiny' else False
|
1171 |
-
return ({"visible": mango, "__type__": "update"})
|
1172 |
-
|
1173 |
-
# ckpt_path2.change(change_info_,[ckpt_path2],[sr__,if_f0__])
|
1174 |
-
def change_info_(ckpt_path):
|
1175 |
-
if (
|
1176 |
-
os.path.exists(ckpt_path.replace(os.path.basename(ckpt_path), "train.log"))
|
1177 |
-
== False
|
1178 |
-
):
|
1179 |
-
return {"__type__": "update"}, {"__type__": "update"}, {"__type__": "update"}
|
1180 |
-
try:
|
1181 |
-
with open(
|
1182 |
-
ckpt_path.replace(os.path.basename(ckpt_path), "train.log"), "r"
|
1183 |
-
) as f:
|
1184 |
-
info = eval(f.read().strip("\n").split("\n")[0].split("\t")[-1])
|
1185 |
-
sr, f0 = info["sample_rate"], info["if_f0"]
|
1186 |
-
version = "v2" if ("version" in info and info["version"] == "v2") else "v1"
|
1187 |
-
return sr, str(f0), version
|
1188 |
-
except:
|
1189 |
-
traceback.print_exc()
|
1190 |
-
return {"__type__": "update"}, {"__type__": "update"}, {"__type__": "update"}
|
1191 |
-
|
1192 |
-
|
1193 |
-
from lib.infer_pack.models_onnx import SynthesizerTrnMsNSFsidM
|
1194 |
-
|
1195 |
-
|
1196 |
-
def export_onnx(ModelPath, ExportedPath, MoeVS=True):
|
1197 |
-
cpt = torch.load(ModelPath, map_location="cpu")
|
1198 |
-
cpt["config"][-3] = cpt["weight"]["emb_g.weight"].shape[0] # n_spk
|
1199 |
-
hidden_channels = 256 if cpt.get("version","v1")=="v1"else 768#cpt["config"][-2] # hidden_channels,为768Vec做准备
|
1200 |
-
|
1201 |
-
test_phone = torch.rand(1, 200, hidden_channels) # hidden unit
|
1202 |
-
test_phone_lengths = torch.tensor([200]).long() # hidden unit 长度(貌似没啥用)
|
1203 |
-
test_pitch = torch.randint(size=(1, 200), low=5, high=255) # 基频(单位赫兹)
|
1204 |
-
test_pitchf = torch.rand(1, 200) # nsf基频
|
1205 |
-
test_ds = torch.LongTensor([0]) # 说话人ID
|
1206 |
-
test_rnd = torch.rand(1, 192, 200) # 噪声(加入随机因子)
|
1207 |
-
|
1208 |
-
device = "cpu" # 导出时设备(不影响使用模型)
|
1209 |
-
|
1210 |
-
|
1211 |
-
net_g = SynthesizerTrnMsNSFsidM(
|
1212 |
-
*cpt["config"], is_half=False,version=cpt.get("version","v1")
|
1213 |
-
) # fp32导出(C++要支持fp16必须手动将内存重新排列所以暂时不用fp16)
|
1214 |
-
net_g.load_state_dict(cpt["weight"], strict=False)
|
1215 |
-
input_names = ["phone", "phone_lengths", "pitch", "pitchf", "ds", "rnd"]
|
1216 |
-
output_names = [
|
1217 |
-
"audio",
|
1218 |
-
]
|
1219 |
-
# net_g.construct_spkmixmap(n_speaker) 多角色混合轨道导出
|
1220 |
-
torch.onnx.export(
|
1221 |
-
net_g,
|
1222 |
-
(
|
1223 |
-
test_phone.to(device),
|
1224 |
-
test_phone_lengths.to(device),
|
1225 |
-
test_pitch.to(device),
|
1226 |
-
test_pitchf.to(device),
|
1227 |
-
test_ds.to(device),
|
1228 |
-
test_rnd.to(device),
|
1229 |
-
),
|
1230 |
-
ExportedPath,
|
1231 |
-
dynamic_axes={
|
1232 |
-
"phone": [1],
|
1233 |
-
"pitch": [1],
|
1234 |
-
"pitchf": [1],
|
1235 |
-
"rnd": [2],
|
1236 |
-
},
|
1237 |
-
do_constant_folding=False,
|
1238 |
-
opset_version=16,
|
1239 |
-
verbose=False,
|
1240 |
-
input_names=input_names,
|
1241 |
-
output_names=output_names,
|
1242 |
-
)
|
1243 |
-
return "Finished"
|
1244 |
-
|
1245 |
-
#region RVC WebUI App
|
1246 |
-
|
1247 |
-
def get_presets():
|
1248 |
-
data = None
|
1249 |
-
with open('../inference-presets.json', 'r') as file:
|
1250 |
-
data = json.load(file)
|
1251 |
-
preset_names = []
|
1252 |
-
for preset in data['presets']:
|
1253 |
-
preset_names.append(preset['name'])
|
1254 |
-
|
1255 |
-
return preset_names
|
1256 |
-
|
1257 |
-
def change_choices2():
|
1258 |
-
audio_files=[]
|
1259 |
-
for filename in os.listdir("./audios"):
|
1260 |
-
if filename.endswith(('.wav','.mp3','.ogg','.flac','.m4a','.aac','.mp4')):
|
1261 |
-
audio_files.append(os.path.join('./audios',filename).replace('\\', '/'))
|
1262 |
-
return {"choices": sorted(audio_files), "__type__": "update"}, {"__type__": "update"}
|
1263 |
-
|
1264 |
-
audio_files=[]
|
1265 |
-
for filename in os.listdir("./audios"):
|
1266 |
-
if filename.endswith(('.wav','.mp3','.ogg','.flac','.m4a','.aac','.mp4')):
|
1267 |
-
audio_files.append(os.path.join('./audios',filename).replace('\\', '/'))
|
1268 |
-
|
1269 |
-
def get_index():
|
1270 |
-
if check_for_name() != '':
|
1271 |
-
chosen_model=sorted(names)[0].split(".")[0]
|
1272 |
-
logs_path="./logs/"+chosen_model
|
1273 |
-
if os.path.exists(logs_path):
|
1274 |
-
for file in os.listdir(logs_path):
|
1275 |
-
if file.endswith(".index"):
|
1276 |
-
return os.path.join(logs_path, file)
|
1277 |
-
return ''
|
1278 |
-
else:
|
1279 |
-
return ''
|
1280 |
-
|
1281 |
-
def get_indexes():
|
1282 |
-
indexes_list=[]
|
1283 |
-
for dirpath, dirnames, filenames in os.walk("./logs/"):
|
1284 |
-
for filename in filenames:
|
1285 |
-
if filename.endswith(".index"):
|
1286 |
-
indexes_list.append(os.path.join(dirpath,filename))
|
1287 |
-
if len(indexes_list) > 0:
|
1288 |
-
return indexes_list
|
1289 |
-
else:
|
1290 |
-
return ''
|
1291 |
-
|
1292 |
-
def get_name():
|
1293 |
-
if len(audio_files) > 0:
|
1294 |
-
return sorted(audio_files)[0]
|
1295 |
-
else:
|
1296 |
-
return ''
|
1297 |
-
|
1298 |
-
def save_to_wav(record_button):
|
1299 |
-
if record_button is None:
|
1300 |
-
pass
|
1301 |
-
else:
|
1302 |
-
path_to_file=record_button
|
1303 |
-
new_name = datetime.datetime.now().strftime("%Y-%m-%d_%H-%M-%S")+'.wav'
|
1304 |
-
new_path='./audios/'+new_name
|
1305 |
-
shutil.move(path_to_file,new_path)
|
1306 |
-
return new_path
|
1307 |
-
|
1308 |
-
def save_to_wav2(dropbox):
|
1309 |
-
file_path=dropbox.name
|
1310 |
-
shutil.move(file_path,'./audios')
|
1311 |
-
return os.path.join('./audios',os.path.basename(file_path))
|
1312 |
-
|
1313 |
-
def match_index(sid0):
|
1314 |
-
folder=sid0.split(".")[0]
|
1315 |
-
parent_dir="./logs/"+folder
|
1316 |
-
if os.path.exists(parent_dir):
|
1317 |
-
for filename in os.listdir(parent_dir):
|
1318 |
-
if filename.endswith(".index"):
|
1319 |
-
index_path=os.path.join(parent_dir,filename)
|
1320 |
-
return index_path
|
1321 |
-
else:
|
1322 |
-
return ''
|
1323 |
-
|
1324 |
-
def check_for_name():
|
1325 |
-
if len(names) > 0:
|
1326 |
-
return sorted(names)[0]
|
1327 |
-
else:
|
1328 |
-
return ''
|
1329 |
-
|
1330 |
-
def download_from_url(url, model):
|
1331 |
-
if url == '':
|
1332 |
-
return "URL cannot be left empty."
|
1333 |
-
if model =='':
|
1334 |
-
return "You need to name your model. For example: My-Model"
|
1335 |
-
url = url.strip()
|
1336 |
-
zip_dirs = ["zips", "unzips"]
|
1337 |
-
for directory in zip_dirs:
|
1338 |
-
if os.path.exists(directory):
|
1339 |
-
shutil.rmtree(directory)
|
1340 |
-
os.makedirs("zips", exist_ok=True)
|
1341 |
-
os.makedirs("unzips", exist_ok=True)
|
1342 |
-
zipfile = model + '.zip'
|
1343 |
-
zipfile_path = './zips/' + zipfile
|
1344 |
-
try:
|
1345 |
-
if "drive.google.com" in url:
|
1346 |
-
subprocess.run(["gdown", url, "--fuzzy", "-O", zipfile_path])
|
1347 |
-
elif "mega.nz" in url:
|
1348 |
-
m = Mega()
|
1349 |
-
m.download_url(url, './zips')
|
1350 |
-
else:
|
1351 |
-
subprocess.run(["wget", url, "-O", zipfile_path])
|
1352 |
-
for filename in os.listdir("./zips"):
|
1353 |
-
if filename.endswith(".zip"):
|
1354 |
-
zipfile_path = os.path.join("./zips/",filename)
|
1355 |
-
shutil.unpack_archive(zipfile_path, "./unzips", 'zip')
|
1356 |
-
else:
|
1357 |
-
return "No zipfile found."
|
1358 |
-
for root, dirs, files in os.walk('./unzips'):
|
1359 |
-
for file in files:
|
1360 |
-
file_path = os.path.join(root, file)
|
1361 |
-
if file.endswith(".index"):
|
1362 |
-
os.mkdir(f'./logs/{model}')
|
1363 |
-
shutil.copy2(file_path,f'./logs/{model}')
|
1364 |
-
elif "G_" not in file and "D_" not in file and file.endswith(".pth"):
|
1365 |
-
shutil.copy(file_path,f'./weights/{model}.pth')
|
1366 |
-
shutil.rmtree("zips")
|
1367 |
-
shutil.rmtree("unzips")
|
1368 |
-
return "Model downloaded, you can go back to the inference page!"
|
1369 |
-
except:
|
1370 |
-
return "ERROR - The download failed. Check if the link is valid."
|
1371 |
-
def success_message(face):
|
1372 |
-
return f'{face.name} has been uploaded.', 'None'
|
1373 |
-
def mouth(size, face, voice, faces):
|
1374 |
-
if size == 'Half':
|
1375 |
-
size = 2
|
1376 |
-
else:
|
1377 |
-
size = 1
|
1378 |
-
if faces == 'None':
|
1379 |
-
character = face.name
|
1380 |
-
else:
|
1381 |
-
if faces == 'Ben Shapiro':
|
1382 |
-
character = '/content/wav2lip-HD/inputs/ben-shapiro-10.mp4'
|
1383 |
-
elif faces == 'Andrew Tate':
|
1384 |
-
character = '/content/wav2lip-HD/inputs/tate-7.mp4'
|
1385 |
-
command = "python inference.py " \
|
1386 |
-
"--checkpoint_path checkpoints/wav2lip.pth " \
|
1387 |
-
f"--face {character} " \
|
1388 |
-
f"--audio {voice} " \
|
1389 |
-
"--pads 0 20 0 0 " \
|
1390 |
-
"--outfile /content/wav2lip-HD/outputs/result.mp4 " \
|
1391 |
-
"--fps 24 " \
|
1392 |
-
f"--resize_factor {size}"
|
1393 |
-
process = subprocess.Popen(command, shell=True, cwd='/content/wav2lip-HD/Wav2Lip-master')
|
1394 |
-
stdout, stderr = process.communicate()
|
1395 |
-
return '/content/wav2lip-HD/outputs/result.mp4', 'Animation completed.'
|
1396 |
-
eleven_voices = ['Adam','Antoni','Josh','Arnold','Sam','Bella','Rachel','Domi','Elli']
|
1397 |
-
eleven_voices_ids=['pNInz6obpgDQGcFmaJgB','ErXwobaYiN019PkySvjV','TxGEqnHWrfWFTfGW9XjX','VR6AewLTigWG4xSOukaG','yoZ06aMxZJJ28mfd3POQ','EXAVITQu4vr4xnSDxMaL','21m00Tcm4TlvDq8ikWAM','AZnzlk1XvdvUeBnXmlld','MF3mGyEYCl7XYWbV9V6O']
|
1398 |
-
chosen_voice = dict(zip(eleven_voices, eleven_voices_ids))
|
1399 |
-
|
1400 |
-
def stoptraining(mim):
|
1401 |
-
if int(mim) == 1:
|
1402 |
-
try:
|
1403 |
-
CSVutil('csvdb/stop.csv', 'w+', 'stop', 'True')
|
1404 |
-
os.kill(PID, signal.SIGTERM)
|
1405 |
-
except Exception as e:
|
1406 |
-
print(f"Couldn't click due to {e}")
|
1407 |
-
return (
|
1408 |
-
{"visible": False, "__type__": "update"},
|
1409 |
-
{"visible": True, "__type__": "update"},
|
1410 |
-
)
|
1411 |
-
|
1412 |
-
|
1413 |
-
def elevenTTS(xiapi, text, id, lang):
|
1414 |
-
if xiapi!= '' and id !='':
|
1415 |
-
choice = chosen_voice[id]
|
1416 |
-
CHUNK_SIZE = 1024
|
1417 |
-
url = f"https://api.elevenlabs.io/v1/text-to-speech/{choice}"
|
1418 |
-
headers = {
|
1419 |
-
"Accept": "audio/mpeg",
|
1420 |
-
"Content-Type": "application/json",
|
1421 |
-
"xi-api-key": xiapi
|
1422 |
-
}
|
1423 |
-
if lang == 'en':
|
1424 |
-
data = {
|
1425 |
-
"text": text,
|
1426 |
-
"model_id": "eleven_monolingual_v1",
|
1427 |
-
"voice_settings": {
|
1428 |
-
"stability": 0.5,
|
1429 |
-
"similarity_boost": 0.5
|
1430 |
-
}
|
1431 |
-
}
|
1432 |
-
else:
|
1433 |
-
data = {
|
1434 |
-
"text": text,
|
1435 |
-
"model_id": "eleven_multilingual_v1",
|
1436 |
-
"voice_settings": {
|
1437 |
-
"stability": 0.5,
|
1438 |
-
"similarity_boost": 0.5
|
1439 |
-
}
|
1440 |
-
}
|
1441 |
-
|
1442 |
-
response = requests.post(url, json=data, headers=headers)
|
1443 |
-
with open('./temp_eleven.mp3', 'wb') as f:
|
1444 |
-
for chunk in response.iter_content(chunk_size=CHUNK_SIZE):
|
1445 |
-
if chunk:
|
1446 |
-
f.write(chunk)
|
1447 |
-
aud_path = save_to_wav('./temp_eleven.mp3')
|
1448 |
-
return aud_path, aud_path
|
1449 |
-
else:
|
1450 |
-
tts = gTTS(text, lang=lang)
|
1451 |
-
tts.save('./temp_gTTS.mp3')
|
1452 |
-
aud_path = save_to_wav('./temp_gTTS.mp3')
|
1453 |
-
return aud_path, aud_path
|
1454 |
-
|
1455 |
-
def ilariaTTS(text, ttsvoice):
|
1456 |
-
vo=language_dict[ttsvoice]
|
1457 |
-
asyncio.run(edge_tts.Communicate(text, vo).save("./temp_ilaria.mp3"))
|
1458 |
-
aud_path = save_to_wav('./temp_ilaria.mp3')
|
1459 |
-
return aud_path, aud_path
|
1460 |
-
|
1461 |
-
def upload_to_dataset(files, dir):
|
1462 |
-
if dir == '':
|
1463 |
-
dir = './dataset'
|
1464 |
-
if not os.path.exists(dir):
|
1465 |
-
os.makedirs(dir)
|
1466 |
-
count = 0
|
1467 |
-
for file in files:
|
1468 |
-
path=file.name
|
1469 |
-
shutil.copy2(path,dir)
|
1470 |
-
count += 1
|
1471 |
-
return f' {count} files uploaded to {dir}.'
|
1472 |
-
|
1473 |
-
def zip_downloader(model):
|
1474 |
-
if not os.path.exists(f'./weights/{model}.pth'):
|
1475 |
-
return {"__type__": "update"}, f'Make sure the Voice Name is correct. I could not find {model}.pth'
|
1476 |
-
index_found = False
|
1477 |
-
for file in os.listdir(f'./logs/{model}'):
|
1478 |
-
if file.endswith('.index') and 'added' in file:
|
1479 |
-
log_file = file
|
1480 |
-
index_found = True
|
1481 |
-
if index_found:
|
1482 |
-
return [f'./weights/{model}.pth', f'./logs/{model}/{log_file}'], "Done"
|
1483 |
-
else:
|
1484 |
-
return f'./weights/{model}.pth', "Could not find Index file."
|
1485 |
-
|
1486 |
-
with gr.Blocks(theme=gr.themes.Default(primary_hue="pink", secondary_hue="rose"), title="Ilaria RVC 💖") as app:
|
1487 |
-
with gr.Tabs():
|
1488 |
-
with gr.TabItem("Inference"):
|
1489 |
-
gr.HTML("<h1> Ilaria RVC 💖 </h1>")
|
1490 |
-
gr.HTML("<h10> You can find voice models on AI Hub: https://discord.gg/aihub </h10>")
|
1491 |
-
gr.HTML("<h4> Huggingface port by Ilaria of the Rejekt Easy GUI </h4>")
|
1492 |
-
|
1493 |
-
# Inference Preset Row
|
1494 |
-
# with gr.Row():
|
1495 |
-
# mangio_preset = gr.Dropdown(label="Inference Preset", choices=sorted(get_presets()))
|
1496 |
-
# mangio_preset_name_save = gr.Textbox(
|
1497 |
-
# label="Your preset name"
|
1498 |
-
# )
|
1499 |
-
# mangio_preset_save_btn = gr.Button('Save Preset', variant="primary")
|
1500 |
-
|
1501 |
-
# Other RVC stuff
|
1502 |
-
with gr.Row():
|
1503 |
-
sid0 = gr.Dropdown(label="1.Choose the model.", choices=sorted(names), value=check_for_name())
|
1504 |
-
refresh_button = gr.Button("Refresh", variant="primary")
|
1505 |
-
if check_for_name() != '':
|
1506 |
-
get_vc(sorted(names)[0])
|
1507 |
-
vc_transform0 = gr.Number(label="Pitch: 0 from man to man (or woman to woman); 12 from man to woman and -12 from woman to man.", value=0)
|
1508 |
-
#clean_button = gr.Button(i18n("卸载音色省显存"), variant="primary")
|
1509 |
-
spk_item = gr.Slider(
|
1510 |
-
minimum=0,
|
1511 |
-
maximum=2333,
|
1512 |
-
step=1,
|
1513 |
-
label=i18n("请选择说话人id"),
|
1514 |
-
value=0,
|
1515 |
-
visible=False,
|
1516 |
-
interactive=True,
|
1517 |
-
)
|
1518 |
-
#clean_button.click(fn=clean, inputs=[], outputs=[sid0])
|
1519 |
-
sid0.change(
|
1520 |
-
fn=get_vc,
|
1521 |
-
inputs=[sid0],
|
1522 |
-
outputs=[spk_item],
|
1523 |
-
)
|
1524 |
-
but0 = gr.Button("Convert", variant="primary")
|
1525 |
-
with gr.Row():
|
1526 |
-
with gr.Column():
|
1527 |
-
with gr.Row():
|
1528 |
-
dropbox = gr.File(label="Drag your audio file and click refresh.")
|
1529 |
-
with gr.Row():
|
1530 |
-
record_button=gr.Audio(label="Or you can use your microphone!", type="filepath")
|
1531 |
-
|
1532 |
-
with gr.Row():
|
1533 |
-
input_audio0 = gr.Dropdown(
|
1534 |
-
label="2.Choose the audio file.",
|
1535 |
-
value="./audios/Test_Audio.mp3",
|
1536 |
-
choices=audio_files
|
1537 |
-
)
|
1538 |
-
dropbox.upload(fn=save_to_wav2, inputs=[dropbox], outputs=[input_audio0])
|
1539 |
-
dropbox.upload(fn=change_choices2, inputs=[], outputs=[input_audio0])
|
1540 |
-
refresh_button2 = gr.Button("Refresh", variant="primary", size='sm')
|
1541 |
-
record_button.change(fn=save_to_wav, inputs=[record_button], outputs=[input_audio0])
|
1542 |
-
record_button.change(fn=change_choices2, inputs=[], outputs=[input_audio0])
|
1543 |
-
with gr.Row():
|
1544 |
-
with gr.Accordion('ElevenLabs / Google TTS', open=False):
|
1545 |
-
with gr.Column():
|
1546 |
-
lang = gr.Radio(label='Chinese & Japanese do not work with ElevenLabs currently.',choices=['en','it','es','fr','pt','zh-CN','de','hi','ja'], value='en')
|
1547 |
-
api_box = gr.Textbox(label="Enter your API Key for ElevenLabs, or leave empty to use GoogleTTS", value='')
|
1548 |
-
elevenid=gr.Dropdown(label="Voice:", choices=eleven_voices)
|
1549 |
-
with gr.Column():
|
1550 |
-
tfs = gr.Textbox(label="Input your Text", interactive=True, value="This is a test.")
|
1551 |
-
tts_button = gr.Button(value="Speak")
|
1552 |
-
tts_button.click(fn=elevenTTS, inputs=[api_box,tfs, elevenid, lang], outputs=[record_button, input_audio0])
|
1553 |
-
with gr.Row():
|
1554 |
-
with gr.Accordion('Wav2Lip', open=False, visible=False):
|
1555 |
-
with gr.Row():
|
1556 |
-
size = gr.Radio(label='Resolution:',choices=['Half','Full'])
|
1557 |
-
face = gr.UploadButton("Upload A Character",type='filepath')
|
1558 |
-
faces = gr.Dropdown(label="OR Choose one:", choices=['None','Ben Shapiro','Andrew Tate'])
|
1559 |
-
with gr.Row():
|
1560 |
-
preview = gr.Textbox(label="Status:",interactive=False)
|
1561 |
-
face.upload(fn=success_message,inputs=[face], outputs=[preview, faces])
|
1562 |
-
with gr.Row():
|
1563 |
-
animation = gr.Video()
|
1564 |
-
refresh_button2.click(fn=change_choices2, inputs=[], outputs=[input_audio0, animation])
|
1565 |
-
with gr.Row():
|
1566 |
-
animate_button = gr.Button('Animate')
|
1567 |
-
|
1568 |
-
with gr.Column():
|
1569 |
-
vc_output2 = gr.Audio(
|
1570 |
-
label="Final Result! (Click on the three dots to download the audio)",
|
1571 |
-
type='filepath',
|
1572 |
-
interactive=False,
|
1573 |
-
)
|
1574 |
-
|
1575 |
-
with gr.Accordion('IlariaTTS', open=True):
|
1576 |
-
with gr.Column():
|
1577 |
-
ilariaid=gr.Dropdown(label="Voice:", choices=ilariavoices, value="English-Jenny (Female)")
|
1578 |
-
ilariatext = gr.Textbox(label="Input your Text", interactive=True, value="This is a test.")
|
1579 |
-
ilariatts_button = gr.Button(value="Speak")
|
1580 |
-
ilariatts_button.click(fn=ilariaTTS, inputs=[ilariatext, ilariaid], outputs=[record_button, input_audio0])
|
1581 |
-
|
1582 |
-
#with gr.Column():
|
1583 |
-
with gr.Accordion("Index Settings", open=False):
|
1584 |
-
#with gr.Row():
|
1585 |
-
|
1586 |
-
file_index1 = gr.Dropdown(
|
1587 |
-
label="3. Choose the index file (in case it wasn't automatically found.)",
|
1588 |
-
choices=get_indexes(),
|
1589 |
-
value=get_index(),
|
1590 |
-
interactive=True,
|
1591 |
-
)
|
1592 |
-
sid0.change(fn=match_index, inputs=[sid0],outputs=[file_index1])
|
1593 |
-
refresh_button.click(
|
1594 |
-
fn=change_choices, inputs=[], outputs=[sid0, file_index1]
|
1595 |
-
)
|
1596 |
-
# file_big_npy1 = gr.Textbox(
|
1597 |
-
# label=i18n("特征文件路径"),
|
1598 |
-
# value="E:\\codes\py39\\vits_vc_gpu_train\\logs\\mi-test-1key\\total_fea.npy",
|
1599 |
-
# interactive=True,
|
1600 |
-
# )
|
1601 |
-
index_rate1 = gr.Slider(
|
1602 |
-
minimum=0,
|
1603 |
-
maximum=1,
|
1604 |
-
label=i18n("检索特征占比"),
|
1605 |
-
value=0.66,
|
1606 |
-
interactive=True,
|
1607 |
-
)
|
1608 |
-
|
1609 |
-
animate_button.click(fn=mouth, inputs=[size, face, vc_output2, faces], outputs=[animation, preview])
|
1610 |
-
|
1611 |
-
with gr.Accordion("Advanced Options", open=False):
|
1612 |
-
f0method0 = gr.Radio(
|
1613 |
-
label="Optional: Change the Pitch Extraction Algorithm. Extraction methods are sorted from 'worst quality' to 'best quality'. If you don't know what you're doing, leave rmvpe.",
|
1614 |
-
choices=["pm", "dio", "crepe-tiny", "mangio-crepe-tiny", "crepe", "harvest", "mangio-crepe", "rmvpe"], # Fork Feature. Add Crepe-Tiny
|
1615 |
-
value="rmvpe",
|
1616 |
-
interactive=True,
|
1617 |
-
)
|
1618 |
-
|
1619 |
-
crepe_hop_length = gr.Slider(
|
1620 |
-
minimum=1,
|
1621 |
-
maximum=512,
|
1622 |
-
step=1,
|
1623 |
-
label="Mangio-Crepe Hop Length. Higher numbers will reduce the chance of extreme pitch changes but lower numbers will increase accuracy. 64-192 is a good range to experiment with.",
|
1624 |
-
value=120,
|
1625 |
-
interactive=True,
|
1626 |
-
visible=False,
|
1627 |
-
)
|
1628 |
-
f0method0.change(fn=whethercrepeornah, inputs=[f0method0], outputs=[crepe_hop_length])
|
1629 |
-
filter_radius0 = gr.Slider(
|
1630 |
-
minimum=0,
|
1631 |
-
maximum=7,
|
1632 |
-
label=i18n(">=3则使用对harvest音高识别的结果使用中值滤波,数值为滤波半径,使用可以削弱哑音"),
|
1633 |
-
value=3,
|
1634 |
-
step=1,
|
1635 |
-
interactive=True,
|
1636 |
-
)
|
1637 |
-
resample_sr0 = gr.Slider(
|
1638 |
-
minimum=0,
|
1639 |
-
maximum=48000,
|
1640 |
-
label=i18n("后处理重采样至最终采样率,0为不进行重采样"),
|
1641 |
-
value=0,
|
1642 |
-
step=1,
|
1643 |
-
interactive=True,
|
1644 |
-
visible=False
|
1645 |
-
)
|
1646 |
-
rms_mix_rate0 = gr.Slider(
|
1647 |
-
minimum=0,
|
1648 |
-
maximum=1,
|
1649 |
-
label=i18n("输入源音量包络替换输出音量包络融合比例,越靠近1越使用输出包络"),
|
1650 |
-
value=0.21,
|
1651 |
-
interactive=True,
|
1652 |
-
)
|
1653 |
-
protect0 = gr.Slider(
|
1654 |
-
minimum=0,
|
1655 |
-
maximum=0.5,
|
1656 |
-
label=i18n("保护清辅音和呼吸声,防止电音撕裂等artifact,拉满0.5不开启,调低加大保护力度但可能降低索引效果"),
|
1657 |
-
value=0.33,
|
1658 |
-
step=0.01,
|
1659 |
-
interactive=True,
|
1660 |
-
)
|
1661 |
-
formanting = gr.Checkbox(
|
1662 |
-
value=bool(DoFormant),
|
1663 |
-
label="[EXPERIMENTAL] Formant shift inference audio",
|
1664 |
-
info="Used for male to female and vice-versa conversions",
|
1665 |
-
interactive=True,
|
1666 |
-
visible=True,
|
1667 |
-
)
|
1668 |
-
|
1669 |
-
formant_preset = gr.Dropdown(
|
1670 |
-
value='',
|
1671 |
-
choices=get_fshift_presets(),
|
1672 |
-
label="browse presets for formanting",
|
1673 |
-
visible=bool(DoFormant),
|
1674 |
-
)
|
1675 |
-
formant_refresh_button = gr.Button(
|
1676 |
-
value='\U0001f504',
|
1677 |
-
visible=bool(DoFormant),
|
1678 |
-
variant='primary',
|
1679 |
-
)
|
1680 |
-
#formant_refresh_button = ToolButton( elem_id='1')
|
1681 |
-
#create_refresh_button(formant_preset, lambda: {"choices": formant_preset}, "refresh_list_shiftpresets")
|
1682 |
-
|
1683 |
-
qfrency = gr.Slider(
|
1684 |
-
value=Quefrency,
|
1685 |
-
info="Default value is 1.0",
|
1686 |
-
label="Quefrency for formant shifting",
|
1687 |
-
minimum=0.0,
|
1688 |
-
maximum=16.0,
|
1689 |
-
step=0.1,
|
1690 |
-
visible=bool(DoFormant),
|
1691 |
-
interactive=True,
|
1692 |
-
)
|
1693 |
-
tmbre = gr.Slider(
|
1694 |
-
value=Timbre,
|
1695 |
-
info="Default value is 1.0",
|
1696 |
-
label="Timbre for formant shifting",
|
1697 |
-
minimum=0.0,
|
1698 |
-
maximum=16.0,
|
1699 |
-
step=0.1,
|
1700 |
-
visible=bool(DoFormant),
|
1701 |
-
interactive=True,
|
1702 |
-
)
|
1703 |
-
|
1704 |
-
formant_preset.change(fn=preset_apply, inputs=[formant_preset, qfrency, tmbre], outputs=[qfrency, tmbre])
|
1705 |
-
frmntbut = gr.Button("Apply", variant="primary", visible=bool(DoFormant))
|
1706 |
-
formanting.change(fn=formant_enabled,inputs=[formanting,qfrency,tmbre,frmntbut,formant_preset,formant_refresh_button],outputs=[formanting,qfrency,tmbre,frmntbut,formant_preset,formant_refresh_button])
|
1707 |
-
frmntbut.click(fn=formant_apply,inputs=[qfrency, tmbre], outputs=[qfrency, tmbre])
|
1708 |
-
formant_refresh_button.click(fn=update_fshift_presets,inputs=[formant_preset, qfrency, tmbre],outputs=[formant_preset, qfrency, tmbre])
|
1709 |
-
|
1710 |
-
with gr.Row():
|
1711 |
-
vc_output1 = gr.Textbox("")
|
1712 |
-
f0_file = gr.File(label=i18n("F0曲线文件, 可选, 一行一个音高, 代替默认F0及升降调"), visible=False)
|
1713 |
-
|
1714 |
-
but0.click(
|
1715 |
-
vc_single,
|
1716 |
-
[
|
1717 |
-
spk_item,
|
1718 |
-
input_audio0,
|
1719 |
-
vc_transform0,
|
1720 |
-
f0_file,
|
1721 |
-
f0method0,
|
1722 |
-
file_index1,
|
1723 |
-
# file_index2,
|
1724 |
-
# file_big_npy1,
|
1725 |
-
index_rate1,
|
1726 |
-
filter_radius0,
|
1727 |
-
resample_sr0,
|
1728 |
-
rms_mix_rate0,
|
1729 |
-
protect0,
|
1730 |
-
crepe_hop_length
|
1731 |
-
],
|
1732 |
-
[vc_output1, vc_output2],
|
1733 |
-
)
|
1734 |
-
|
1735 |
-
with gr.Accordion("Batch Conversion",open=False, visible=False):
|
1736 |
-
with gr.Row():
|
1737 |
-
with gr.Column():
|
1738 |
-
vc_transform1 = gr.Number(
|
1739 |
-
label=i18n("变调(整数, 半音数量, 升八度12降八度-12)"), value=0
|
1740 |
-
)
|
1741 |
-
opt_input = gr.Textbox(label=i18n("指定输出文件夹"), value="opt")
|
1742 |
-
f0method1 = gr.Radio(
|
1743 |
-
label=i18n(
|
1744 |
-
"选择音高提取算法,输入歌声可用pm提速,harvest低音好但巨慢无比,crepe效果好但吃GPU"
|
1745 |
-
),
|
1746 |
-
choices=["pm", "harvest", "crepe", "rmvpe"],
|
1747 |
-
value="rmvpe",
|
1748 |
-
interactive=True,
|
1749 |
-
)
|
1750 |
-
filter_radius1 = gr.Slider(
|
1751 |
-
minimum=0,
|
1752 |
-
maximum=7,
|
1753 |
-
label=i18n(">=3则使用对harvest音高识别的结果使用中值滤波,数值为滤波半径,使用可以削弱哑音"),
|
1754 |
-
value=3,
|
1755 |
-
step=1,
|
1756 |
-
interactive=True,
|
1757 |
-
)
|
1758 |
-
with gr.Column():
|
1759 |
-
file_index3 = gr.Textbox(
|
1760 |
-
label=i18n("特征检索库文件路径,为空则使用下拉的选择结果"),
|
1761 |
-
value="",
|
1762 |
-
interactive=True,
|
1763 |
-
)
|
1764 |
-
file_index4 = gr.Dropdown(
|
1765 |
-
label=i18n("自动检测index路径,下拉式选择(dropdown)"),
|
1766 |
-
choices=sorted(index_paths),
|
1767 |
-
interactive=True,
|
1768 |
-
)
|
1769 |
-
refresh_button.click(
|
1770 |
-
fn=lambda: change_choices()[1],
|
1771 |
-
inputs=[],
|
1772 |
-
outputs=file_index4,
|
1773 |
-
)
|
1774 |
-
# file_big_npy2 = gr.Textbox(
|
1775 |
-
# label=i18n("特征文件路径"),
|
1776 |
-
# value="E:\\codes\\py39\\vits_vc_gpu_train\\logs\\mi-test-1key\\total_fea.npy",
|
1777 |
-
# interactive=True,
|
1778 |
-
# )
|
1779 |
-
index_rate2 = gr.Slider(
|
1780 |
-
minimum=0,
|
1781 |
-
maximum=1,
|
1782 |
-
label=i18n("检索特征占比"),
|
1783 |
-
value=1,
|
1784 |
-
interactive=True,
|
1785 |
-
)
|
1786 |
-
with gr.Column():
|
1787 |
-
resample_sr1 = gr.Slider(
|
1788 |
-
minimum=0,
|
1789 |
-
maximum=48000,
|
1790 |
-
label=i18n("后处理重采样至最终采样率,0为不进行重采样"),
|
1791 |
-
value=0,
|
1792 |
-
step=1,
|
1793 |
-
interactive=True,
|
1794 |
-
)
|
1795 |
-
rms_mix_rate1 = gr.Slider(
|
1796 |
-
minimum=0,
|
1797 |
-
maximum=1,
|
1798 |
-
label=i18n("输入源音量包络替换输出音量包络融合比例,越靠近1越使用输出包络"),
|
1799 |
-
value=1,
|
1800 |
-
interactive=True,
|
1801 |
-
)
|
1802 |
-
protect1 = gr.Slider(
|
1803 |
-
minimum=0,
|
1804 |
-
maximum=0.5,
|
1805 |
-
label=i18n(
|
1806 |
-
"保护清辅音和呼吸声,防止电音撕裂等artifact,拉满0.5不开启,调低加大保护力度但可能降低索引效果"
|
1807 |
-
),
|
1808 |
-
value=0.33,
|
1809 |
-
step=0.01,
|
1810 |
-
interactive=True,
|
1811 |
-
)
|
1812 |
-
with gr.Column():
|
1813 |
-
dir_input = gr.Textbox(
|
1814 |
-
label=i18n("输入待处理音频文件夹路径(去文件管理器地址栏拷就行了)"),
|
1815 |
-
value="E:\codes\py39\\test-20230416b\\todo-songs",
|
1816 |
-
)
|
1817 |
-
inputs = gr.File(
|
1818 |
-
file_count="multiple", label=i18n("也可批量输入音频文件, 二选一, 优先读文件夹")
|
1819 |
-
)
|
1820 |
-
with gr.Row():
|
1821 |
-
format1 = gr.Radio(
|
1822 |
-
label=i18n("导出文件格式"),
|
1823 |
-
choices=["wav", "flac", "mp3", "m4a"],
|
1824 |
-
value="flac",
|
1825 |
-
interactive=True,
|
1826 |
-
)
|
1827 |
-
but1 = gr.Button(i18n("转换"), variant="primary")
|
1828 |
-
vc_output3 = gr.Textbox(label=i18n("输出信息"))
|
1829 |
-
but1.click(
|
1830 |
-
vc_multi,
|
1831 |
-
[
|
1832 |
-
spk_item,
|
1833 |
-
dir_input,
|
1834 |
-
opt_input,
|
1835 |
-
inputs,
|
1836 |
-
vc_transform1,
|
1837 |
-
f0method1,
|
1838 |
-
file_index3,
|
1839 |
-
file_index4,
|
1840 |
-
# file_big_npy2,
|
1841 |
-
index_rate2,
|
1842 |
-
filter_radius1,
|
1843 |
-
resample_sr1,
|
1844 |
-
rms_mix_rate1,
|
1845 |
-
protect1,
|
1846 |
-
format1,
|
1847 |
-
crepe_hop_length,
|
1848 |
-
],
|
1849 |
-
[vc_output3],
|
1850 |
-
)
|
1851 |
-
but1.click(fn=lambda: easy_uploader.clear())
|
1852 |
-
with gr.TabItem("Download Voice Models"):
|
1853 |
-
with gr.Row():
|
1854 |
-
url=gr.Textbox(label="Huggingface Link:")
|
1855 |
-
with gr.Row():
|
1856 |
-
model = gr.Textbox(label="Name of the model (without spaces):")
|
1857 |
-
download_button=gr.Button("Download")
|
1858 |
-
with gr.Row():
|
1859 |
-
status_bar=gr.Textbox(label="Download Status")
|
1860 |
-
download_button.click(fn=download_from_url, inputs=[url, model], outputs=[status_bar])
|
1861 |
-
with gr.Row():
|
1862 |
-
gr.Markdown(
|
1863 |
-
"""
|
1864 |
-
Made with 💖 by Ilaria | Support her on [Ko-Fi](https://ko-fi.com/ilariaowo)
|
1865 |
-
"""
|
1866 |
-
)
|
1867 |
-
|
1868 |
-
def has_two_files_in_pretrained_folder():
|
1869 |
-
pretrained_folder = "./pretrained/"
|
1870 |
-
if not os.path.exists(pretrained_folder):
|
1871 |
-
return False
|
1872 |
-
|
1873 |
-
files_in_folder = os.listdir(pretrained_folder)
|
1874 |
-
num_files = len(files_in_folder)
|
1875 |
-
return num_files >= 2
|
1876 |
-
print(
|
1877 |
-
"=" * 50,
|
1878 |
-
"Disabling Training, as ZeroGPU only supports a running time of 120 seconds.",
|
1879 |
-
"Please use a local machine, or colab for training.",
|
1880 |
-
"=" * 50,
|
1881 |
-
)
|
1882 |
-
|
1883 |
-
app.launch(share=False, quiet=False, max_threads=1022)
|
1884 |
-
#endpain
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import requests
|
3 |
+
import random
|
4 |
+
import os
|
5 |
+
import zipfile # built in module for unzipping files (thank god)
|
6 |
+
import librosa
|
7 |
+
import time
|
8 |
+
from infer_rvc_python import BaseLoader
|
9 |
+
from pydub import AudioSegment
|
10 |
+
from tts_voice import tts_order_voice
|
11 |
+
import edge_tts
|
12 |
+
import tempfile
|
13 |
+
import anyio
|
14 |
+
from audio_separator.separator import Separator
|
15 |
+
|
16 |
+
|
17 |
+
language_dict = tts_order_voice
|
18 |
+
|
19 |
+
# ilaria tts implementation :rofl:
|
20 |
+
async def text_to_speech_edge(text, language_code):
|
21 |
+
voice = language_dict[language_code]
|
22 |
+
communicate = edge_tts.Communicate(text, voice)
|
23 |
+
with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_file:
|
24 |
+
tmp_path = tmp_file.name
|
25 |
+
|
26 |
+
await communicate.save(tmp_path)
|
27 |
+
|
28 |
+
return tmp_path
|
29 |
+
|
30 |
+
# fucking dogshit toggle
|
31 |
+
try:
|
32 |
+
import spaces
|
33 |
+
spaces_status = True
|
34 |
+
except ImportError:
|
35 |
+
spaces_status = False
|
36 |
+
|
37 |
+
separator = Separator()
|
38 |
+
converter = BaseLoader(only_cpu=False, hubert_path=None, rmvpe_path=None) # <- yeah so like this handles rvc
|
39 |
+
|
40 |
+
global pth_file
|
41 |
+
global index_file
|
42 |
+
|
43 |
+
pth_file = "model.pth"
|
44 |
+
index_file = "model.index"
|
45 |
+
|
46 |
+
#CONFIGS
|
47 |
+
TEMP_DIR = "temp"
|
48 |
+
MODEL_PREFIX = "model"
|
49 |
+
PITCH_ALGO_OPT = [
|
50 |
+
"pm",
|
51 |
+
"harvest",
|
52 |
+
"crepe",
|
53 |
+
"rmvpe",
|
54 |
+
"rmvpe+",
|
55 |
+
]
|
56 |
+
UVR_5_MODELS = [
|
57 |
+
{"model_name": "BS-Roformer-Viperx-1297", "checkpoint": "model_bs_roformer_ep_317_sdr_12.9755.ckpt"},
|
58 |
+
{"model_name": "MDX23C-InstVoc HQ 2", "checkpoint": "MDX23C-8KFFT-InstVoc_HQ_2.ckpt"},
|
59 |
+
{"model_name": "Kim Vocal 2", "checkpoint": "Kim_Vocal_2.onnx"},
|
60 |
+
{"model_name": "5_HP-Karaoke", "checkpoint": "5_HP-Karaoke-UVR.pth"},
|
61 |
+
{"model_name": "UVR-DeNoise by FoxJoy", "checkpoint": "UVR-DeNoise.pth"},
|
62 |
+
{"model_name": "UVR-DeEcho-DeReverb by FoxJoy", "checkpoint": "UVR-DeEcho-DeReverb.pth"},
|
63 |
+
]
|
64 |
+
|
65 |
+
os.makedirs(TEMP_DIR, exist_ok=True)
|
66 |
+
|
67 |
+
def unzip_file(file):
|
68 |
+
filename = os.path.basename(file).split(".")[0] # converts "model.zip" to "model" so we can do things
|
69 |
+
with zipfile.ZipFile(file, 'r') as zip_ref:
|
70 |
+
zip_ref.extractall(os.path.join(TEMP_DIR, filename)) # might not be very ram efficient...
|
71 |
+
return True
|
72 |
+
|
73 |
+
|
74 |
+
def progress_bar(total, current): # best progress bar ever trust me sunglasses emoji 😎
|
75 |
+
return "[" + "=" * int(current / total * 20) + ">" + " " * (20 - int(current / total * 20)) + "] " + str(int(current / total * 100)) + "%"
|
76 |
+
|
77 |
+
def download_from_url(url, filename=None):
|
78 |
+
if "/blob/" in url:
|
79 |
+
url = url.replace("/blob/", "/resolve/") # made it delik proof 😎
|
80 |
+
if "huggingface" not in url:
|
81 |
+
return ["The URL must be from huggingface", "Failed", "Failed"]
|
82 |
+
if filename is None:
|
83 |
+
filename = os.path.join(TEMP_DIR, MODEL_PREFIX + str(random.randint(1, 1000)) + ".zip")
|
84 |
+
response = requests.get(url)
|
85 |
+
total = int(response.headers.get('content-length', 0)) # bytes to download (length of the file)
|
86 |
+
if total > 500000000:
|
87 |
+
|
88 |
+
return ["The file is too large. You can only download files up to 500 MB in size.", "Failed", "Failed"]
|
89 |
+
current = 0
|
90 |
+
with open(filename, "wb") as f:
|
91 |
+
for data in response.iter_content(chunk_size=4096): # download in chunks of 4096 bytes (4kb - helps with memory usage and speed)
|
92 |
+
f.write(data)
|
93 |
+
current += len(data)
|
94 |
+
print(progress_bar(total, current), end="\r") # \r is a carriage return, it moves the cursor to the start of the line so its like tqdm sunglasses emoji 😎
|
95 |
+
|
96 |
+
# unzip because the model is in a zip file lel
|
97 |
+
|
98 |
+
try:
|
99 |
+
unzip_file(filename)
|
100 |
+
except Exception as e:
|
101 |
+
return ["Failed to unzip the file", "Failed", "Failed"] # return early if it fails and like tell the user but its dogshit hahahahahahaha 😎 According to all known laws aviation, there is no way a bee should be able to fly.
|
102 |
+
unzipped_dir = os.path.join(TEMP_DIR, os.path.basename(filename).split(".")[0]) # just do what we did in unzip_file because we need the directory
|
103 |
+
pth_files = []
|
104 |
+
index_files = []
|
105 |
+
for root, dirs, files in os.walk(unzipped_dir): # could be done more efficiently because nobody stores models in subdirectories but like who cares (it's a futureproofing thing lel)
|
106 |
+
for file in files:
|
107 |
+
if file.endswith(".pth"):
|
108 |
+
pth_files.append(os.path.join(root, file))
|
109 |
+
elif file.endswith(".index"):
|
110 |
+
index_files.append(os.path.join(root, file))
|
111 |
+
|
112 |
+
print(pth_files, index_files) # debug print because im fucking stupid and i need to see what is going on
|
113 |
+
global pth_file
|
114 |
+
global index_file
|
115 |
+
pth_file = pth_files[0]
|
116 |
+
index_file = index_files[0]
|
117 |
+
|
118 |
+
pth_file_ui.value = pth_file
|
119 |
+
index_file_ui.value = index_file
|
120 |
+
print(pth_file_ui.value)
|
121 |
+
print(index_file_ui.value)
|
122 |
+
return ["Downloaded as " + filename, pth_files[0], index_files[0]]
|
123 |
+
|
124 |
+
def inference(audio, model_name):
|
125 |
+
output_data = inf_handler(audio, model_name)
|
126 |
+
vocals = output_data[0]
|
127 |
+
inst = output_data[1]
|
128 |
+
|
129 |
+
return vocals, inst
|
130 |
+
|
131 |
+
if spaces_status:
|
132 |
+
@spaces.GPU()
|
133 |
+
def convert_now(audio_files, random_tag, converter):
|
134 |
+
return converter(
|
135 |
+
audio_files,
|
136 |
+
random_tag,
|
137 |
+
overwrite=False,
|
138 |
+
parallel_workers=8
|
139 |
+
)
|
140 |
+
|
141 |
+
|
142 |
+
else:
|
143 |
+
def convert_now(audio_files, random_tag, converter):
|
144 |
+
return converter(
|
145 |
+
audio_files,
|
146 |
+
random_tag,
|
147 |
+
overwrite=False,
|
148 |
+
parallel_workers=8
|
149 |
+
)
|
150 |
+
|
151 |
+
def calculate_remaining_time(epochs, seconds_per_epoch):
|
152 |
+
total_seconds = epochs * seconds_per_epoch
|
153 |
+
|
154 |
+
hours = total_seconds // 3600
|
155 |
+
minutes = (total_seconds % 3600) // 60
|
156 |
+
seconds = total_seconds % 60
|
157 |
+
|
158 |
+
if hours == 0:
|
159 |
+
return f"{int(minutes)} minutes"
|
160 |
+
elif hours == 1:
|
161 |
+
return f"{int(hours)} hour and {int(minutes)} minutes"
|
162 |
+
else:
|
163 |
+
return f"{int(hours)} hours and {int(minutes)} minutes"
|
164 |
+
|
165 |
+
def inf_handler(audio, model_name): # its a shame that zerogpu just WONT cooperate with us
|
166 |
+
model_found = False
|
167 |
+
for model_info in UVR_5_MODELS:
|
168 |
+
if model_info["model_name"] == model_name:
|
169 |
+
separator.load_model(model_info["checkpoint"])
|
170 |
+
model_found = True
|
171 |
+
break
|
172 |
+
if not model_found:
|
173 |
+
separator.load_model()
|
174 |
+
output_files = separator.separate(audio)
|
175 |
+
vocals = output_files[0]
|
176 |
+
inst = output_files[1]
|
177 |
+
return vocals, inst
|
178 |
+
|
179 |
+
|
180 |
+
def run(
|
181 |
+
audio_files,
|
182 |
+
pitch_alg,
|
183 |
+
pitch_lvl,
|
184 |
+
index_inf,
|
185 |
+
r_m_f,
|
186 |
+
e_r,
|
187 |
+
c_b_p,
|
188 |
+
):
|
189 |
+
if not audio_files:
|
190 |
+
raise ValueError("The audio pls")
|
191 |
+
|
192 |
+
if isinstance(audio_files, str):
|
193 |
+
audio_files = [audio_files]
|
194 |
+
|
195 |
+
try:
|
196 |
+
duration_base = librosa.get_duration(filename=audio_files[0])
|
197 |
+
print("Duration:", duration_base)
|
198 |
+
except Exception as e:
|
199 |
+
print(e)
|
200 |
+
|
201 |
+
random_tag = "USER_"+str(random.randint(10000000, 99999999))
|
202 |
+
|
203 |
+
file_m = pth_file_ui.value
|
204 |
+
file_index = index_file_ui.value
|
205 |
+
|
206 |
+
print("Random tag:", random_tag)
|
207 |
+
print("File model:", file_m)
|
208 |
+
print("Pitch algorithm:", pitch_alg)
|
209 |
+
print("Pitch level:", pitch_lvl)
|
210 |
+
print("File index:", file_index)
|
211 |
+
print("Index influence:", index_inf)
|
212 |
+
print("Respiration median filtering:", r_m_f)
|
213 |
+
print("Envelope ratio:", e_r)
|
214 |
+
|
215 |
+
converter.apply_conf(
|
216 |
+
tag=random_tag,
|
217 |
+
file_model=file_m,
|
218 |
+
pitch_algo=pitch_alg,
|
219 |
+
pitch_lvl=pitch_lvl,
|
220 |
+
file_index=file_index,
|
221 |
+
index_influence=index_inf,
|
222 |
+
respiration_median_filtering=r_m_f,
|
223 |
+
envelope_ratio=e_r,
|
224 |
+
consonant_breath_protection=c_b_p,
|
225 |
+
resample_sr=44100 if audio_files[0].endswith('.mp3') else 0,
|
226 |
+
)
|
227 |
+
time.sleep(0.1)
|
228 |
+
|
229 |
+
result = convert_now(audio_files, random_tag, converter)
|
230 |
+
print("Result:", result)
|
231 |
+
|
232 |
+
return result[0]
|
233 |
+
|
234 |
+
def upload_model(index_file, pth_file):
|
235 |
+
pth_file = pth_file.name
|
236 |
+
index_file = index_file.name
|
237 |
+
pth_file_ui.value = pth_file
|
238 |
+
index_file_ui.value = index_file
|
239 |
+
return "Uploaded!"
|
240 |
+
|
241 |
+
with gr.Blocks(theme="Ilaria RVC") as demo:
|
242 |
+
gr.Markdown("## Ilaria RVC 💖")
|
243 |
+
with gr.Tab("Inference"):
|
244 |
+
sound_gui = gr.Audio(value=None,type="filepath",autoplay=False,visible=True,)
|
245 |
+
pth_file_ui = gr.Textbox(label="Model pth file",value=pth_file,visible=False,interactive=False,)
|
246 |
+
index_file_ui = gr.Textbox(label="Index pth file",value=index_file,visible=False,interactive=False,)
|
247 |
+
|
248 |
+
with gr.Accordion("Settings", open=False):
|
249 |
+
pitch_algo_conf = gr.Dropdown(PITCH_ALGO_OPT,value=PITCH_ALGO_OPT[4],label="Pitch algorithm",visible=True,interactive=True,)
|
250 |
+
pitch_lvl_conf = gr.Slider(label="Pitch level (lower -> 'male' while higher -> 'female')",minimum=-24,maximum=24,step=1,value=0,visible=True,interactive=True,)
|
251 |
+
index_inf_conf = gr.Slider(minimum=0,maximum=1,label="Index influence -> How much accent is applied",value=0.75,)
|
252 |
+
respiration_filter_conf = gr.Slider(minimum=0,maximum=7,label="Respiration median filtering",value=3,step=1,interactive=True,)
|
253 |
+
envelope_ratio_conf = gr.Slider(minimum=0,maximum=1,label="Envelope ratio",value=0.25,interactive=True,)
|
254 |
+
consonant_protec_conf = gr.Slider(minimum=0,maximum=0.5,label="Consonant breath protection",value=0.5,interactive=True,)
|
255 |
+
|
256 |
+
button_conf = gr.Button("Convert",variant="primary",)
|
257 |
+
output_conf = gr.Audio(type="filepath",label="Output",)
|
258 |
+
|
259 |
+
button_conf.click(lambda :None, None, output_conf)
|
260 |
+
button_conf.click(
|
261 |
+
run,
|
262 |
+
inputs=[
|
263 |
+
sound_gui,
|
264 |
+
pitch_algo_conf,
|
265 |
+
pitch_lvl_conf,
|
266 |
+
index_inf_conf,
|
267 |
+
respiration_filter_conf,
|
268 |
+
envelope_ratio_conf,
|
269 |
+
consonant_protec_conf,
|
270 |
+
],
|
271 |
+
outputs=[output_conf],
|
272 |
+
)
|
273 |
+
|
274 |
+
with gr.Tab("Ilaria TTS"):
|
275 |
+
text_tts = gr.Textbox(label="Text", placeholder="Hello!", lines=3, interactive=True,)
|
276 |
+
dropdown_tts = gr.Dropdown(label="Language and Model",choices=list(language_dict.keys()),interactive=True, value=list(language_dict.keys())[0])
|
277 |
+
|
278 |
+
button_tts = gr.Button("Speak", variant="primary",)
|
279 |
+
|
280 |
+
output_tts = gr.Audio(type="filepath", label="Output",)
|
281 |
+
|
282 |
+
button_tts.click(text_to_speech_edge, inputs=[text_tts, dropdown_tts], outputs=[output_tts])
|
283 |
+
|
284 |
+
|
285 |
+
with gr.Tab("Model Loader (Download and Upload)"):
|
286 |
+
with gr.Accordion("Model Downloader", open=False):
|
287 |
+
gr.Markdown(
|
288 |
+
"Download the model from the following URL and upload it here. (Hugginface RVC model)"
|
289 |
+
)
|
290 |
+
model = gr.Textbox(lines=1, label="Model URL")
|
291 |
+
download_button = gr.Button("Download Model")
|
292 |
+
status = gr.Textbox(lines=1, label="Status", placeholder="Waiting....", interactive=False)
|
293 |
+
model_pth = gr.Textbox(lines=1, label="Model pth file", placeholder="Waiting....", interactive=False)
|
294 |
+
index_pth = gr.Textbox(lines=1, label="Index pth file", placeholder="Waiting....", interactive=False)
|
295 |
+
download_button.click(download_from_url, model, outputs=[status, model_pth, index_pth])
|
296 |
+
with gr.Accordion("Upload A Model", open=False):
|
297 |
+
index_file_upload = gr.File(label="Index File (.index)")
|
298 |
+
pth_file_upload = gr.File(label="Model File (.pth)")
|
299 |
+
upload_button = gr.Button("Upload Model")
|
300 |
+
upload_status = gr.Textbox(lines=1, label="Status", placeholder="Waiting....", interactive=False)
|
301 |
+
|
302 |
+
upload_button.click(upload_model, [index_file_upload, pth_file_upload], upload_status)
|
303 |
+
|
304 |
+
|
305 |
+
with gr.Tab("Vocal Separator (UVR)"):
|
306 |
+
gr.Markdown("Separate vocals and instruments from an audio file using UVR models. - This is only on CPU due to ZeroGPU being ZeroGPU :(")
|
307 |
+
uvr5_audio_file = gr.Audio(label="Audio File",type="filepath")
|
308 |
+
|
309 |
+
with gr.Row():
|
310 |
+
uvr5_model = gr.Dropdown(label="Model", choices=[model["model_name"] for model in UVR_5_MODELS])
|
311 |
+
uvr5_button = gr.Button("Separate Vocals", variant="primary",)
|
312 |
+
|
313 |
+
uvr5_output_voc = gr.Audio(type="filepath", label="Output 1",) # UVR models sometimes output it in a weird way where it's like the positions swap randomly, so let's just call them Outputs lol
|
314 |
+
uvr5_output_inst = gr.Audio(type="filepath", label="Output 2",)
|
315 |
+
|
316 |
+
uvr5_button.click(inference, [uvr5_audio_file, uvr5_model], [uvr5_output_voc, uvr5_output_inst])
|
317 |
+
|
318 |
+
with gr.Tab("Extra"):
|
319 |
+
with gr.Accordion("Training Time Calculator", open=False):
|
320 |
+
with gr.Column():
|
321 |
+
epochs_input = gr.Number(label="Number of Epochs")
|
322 |
+
seconds_input = gr.Number(label="Seconds per Epoch")
|
323 |
+
calculate_button = gr.Button("Calculate Time Remaining")
|
324 |
+
remaining_time_output = gr.Textbox(label="Remaining Time", interactive=False)
|
325 |
+
|
326 |
+
calculate_button.click(
|
327 |
+
fn=calculate_remaining_time,
|
328 |
+
inputs=[epochs_input, seconds_input],
|
329 |
+
outputs=[remaining_time_output]
|
330 |
+
)
|
331 |
+
|
332 |
+
with gr.Accordion("Model Fusion", open=False):
|
333 |
+
gr.Markdown(value="Fusion of two models to create a new model - coming soon! 😎")
|
334 |
+
|
335 |
+
with gr.Accordion("Model Quantization", open=False):
|
336 |
+
gr.Markdown(value="Quantization of a model to reduce its size - coming soon! 😎")
|
337 |
+
|
338 |
+
with gr.Accordion("Training Helper", open=False):
|
339 |
+
gr.Markdown(value="Help for training models - coming soon! 😎")
|
340 |
+
|
341 |
+
with gr.Tab("Credits"):
|
342 |
+
gr.Markdown(
|
343 |
+
"""
|
344 |
+
Ilaria RVC made by [Ilaria](https://huggingface.co/TheStinger) suport her on [ko-fi](https://ko-fi.com/ilariaowo)
|
345 |
+
|
346 |
+
The Inference code is made by [r3gm](https://huggingface.co/r3gm) (his module helped form this space 💖)
|
347 |
+
|
348 |
+
made with ❤️ by [mikus](https://github.com/cappuch) - i make this ui........
|
349 |
+
|
350 |
+
## In loving memory of JLabDX 🕊️
|
351 |
+
"""
|
352 |
+
)
|
353 |
+
|
354 |
+
demo.queue(api_open=False).launch(show_api=False) # idk ilaria if you want or dont want to
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
gitattributes
ADDED
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
*.7z filter=lfs diff=lfs merge=lfs -text
|
2 |
+
*.arrow filter=lfs diff=lfs merge=lfs -text
|
3 |
+
*.bin filter=lfs diff=lfs merge=lfs -text
|
4 |
+
*.bz2 filter=lfs diff=lfs merge=lfs -text
|
5 |
+
*.ckpt filter=lfs diff=lfs merge=lfs -text
|
6 |
+
*.ftz filter=lfs diff=lfs merge=lfs -text
|
7 |
+
*.gz filter=lfs diff=lfs merge=lfs -text
|
8 |
+
*.h5 filter=lfs diff=lfs merge=lfs -text
|
9 |
+
*.joblib filter=lfs diff=lfs merge=lfs -text
|
10 |
+
*.lfs.* filter=lfs diff=lfs merge=lfs -text
|
11 |
+
*.mlmodel filter=lfs diff=lfs merge=lfs -text
|
12 |
+
*.model filter=lfs diff=lfs merge=lfs -text
|
13 |
+
*.msgpack filter=lfs diff=lfs merge=lfs -text
|
14 |
+
*.npy filter=lfs diff=lfs merge=lfs -text
|
15 |
+
*.npz filter=lfs diff=lfs merge=lfs -text
|
16 |
+
*.onnx filter=lfs diff=lfs merge=lfs -text
|
17 |
+
*.ot filter=lfs diff=lfs merge=lfs -text
|
18 |
+
*.parquet filter=lfs diff=lfs merge=lfs -text
|
19 |
+
*.pb filter=lfs diff=lfs merge=lfs -text
|
20 |
+
*.pickle filter=lfs diff=lfs merge=lfs -text
|
21 |
+
*.pkl filter=lfs diff=lfs merge=lfs -text
|
22 |
+
*.pt filter=lfs diff=lfs merge=lfs -text
|
23 |
+
*.pth filter=lfs diff=lfs merge=lfs -text
|
24 |
+
*.rar filter=lfs diff=lfs merge=lfs -text
|
25 |
+
*.safetensors filter=lfs diff=lfs merge=lfs -text
|
26 |
+
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
27 |
+
*.tar.* filter=lfs diff=lfs merge=lfs -text
|
28 |
+
*.tar filter=lfs diff=lfs merge=lfs -text
|
29 |
+
*.tflite filter=lfs diff=lfs merge=lfs -text
|
30 |
+
*.tgz filter=lfs diff=lfs merge=lfs -text
|
31 |
+
*.wasm filter=lfs diff=lfs merge=lfs -text
|
32 |
+
*.xz filter=lfs diff=lfs merge=lfs -text
|
33 |
+
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
+
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
+
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
model.index filter=lfs diff=lfs merge=lfs -text
|
model.index
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:af434a9142b070f7091dcdbbf957b7a01bbc96294add99d186ef1e0d4b226eac
|
3 |
+
size 83987395
|
model.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:896fcee182ecdcea6645a691366ac50153bc63015f43c981da135a8cabe2f088
|
3 |
+
size 55028048
|
packages.txt
CHANGED
@@ -1,3 +1 @@
|
|
1 |
-
|
2 |
-
ffmpeg
|
3 |
-
aria2
|
|
|
1 |
+
ffmpeg
|
|
|
|
requirements.txt
CHANGED
@@ -1,23 +1,10 @@
|
|
1 |
-
|
2 |
-
|
3 |
-
edge-tts
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
gradio==4.36.1
|
12 |
-
ffmpeg-python
|
13 |
-
praat-parselmouth
|
14 |
-
pyworld
|
15 |
-
numpy==1.23.5
|
16 |
-
i18n
|
17 |
-
numba==0.56.4
|
18 |
-
librosa==0.9.2
|
19 |
-
mega.py
|
20 |
-
gdown @ git+https://github.com/IAHispano/gdown.git
|
21 |
-
onnxruntime
|
22 |
-
pyngrok==4.1.12
|
23 |
-
torch
|
|
|
1 |
+
torch==2.2.0
|
2 |
+
infer-rvc-python==1.1.0
|
3 |
+
edge-tts
|
4 |
+
pedalboard
|
5 |
+
noisereduce
|
6 |
+
numpy==1.23.5
|
7 |
+
audio-separator[gpu]
|
8 |
+
scipy
|
9 |
+
onnxruntime-gpu
|
10 |
+
samplerate
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
test.ogg
ADDED
Binary file (73.4 kB). View file
|
|
tts_voice.py
ADDED
@@ -0,0 +1,230 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
tts_order_voice = {'English-Jenny (Woman)': 'en-US-JennyNeural',
|
2 |
+
'English-Guy (Man)': 'en-US-GuyNeural',
|
3 |
+
'English-Ana (Woman)': 'en-US-AnaNeural',
|
4 |
+
'English-Aria (Woman)': 'en-US-AriaNeural',
|
5 |
+
'English-Christopher (Man)': 'en-US-ChristopherNeural',
|
6 |
+
'English-Eric (Man)': 'en-US-EricNeural',
|
7 |
+
'English-Michelle (Woman)': 'en-US-MichelleNeural',
|
8 |
+
'English-Roger (Man)': 'en-US-RogerNeural',
|
9 |
+
'Spanish (Mexican)-Dalia (Woman)': 'es-MX-DaliaNeural',
|
10 |
+
'Spanish (Mexican)-Jorge- (Man)': 'es-MX-JorgeNeural',
|
11 |
+
'Korean-Sun-Hi- (Woman)': 'ko-KR-SunHiNeural',
|
12 |
+
'Korean-InJoon- (Man)': 'ko-KR-InJoonNeural',
|
13 |
+
'Thai-Premwadee- (Woman)': 'th-TH-PremwadeeNeural',
|
14 |
+
'Thai-Niwat- (Man)': 'th-TH-NiwatNeural',
|
15 |
+
'Vietnamese-HoaiMy- (Woman)': 'vi-VN-HoaiMyNeural',
|
16 |
+
'Vietnamese-NamMinh- (Man)': 'vi-VN-NamMinhNeural',
|
17 |
+
'Japanese-Nanami- (Woman)': 'ja-JP-NanamiNeural',
|
18 |
+
'Japanese-Keita- (Man)': 'ja-JP-KeitaNeural',
|
19 |
+
'French-Denise- (Woman)': 'fr-FR-DeniseNeural',
|
20 |
+
'French-Eloise- (Woman)': 'fr-FR-EloiseNeural',
|
21 |
+
'French-Henri- (Man)': 'fr-FR-HenriNeural',
|
22 |
+
'Brazilian-Francisca- (Woman)': 'pt-BR-FranciscaNeural',
|
23 |
+
'Brazilian-Antonio- (Man)': 'pt-BR-AntonioNeural',
|
24 |
+
'Indonesian-Ardi- (Man)': 'id-ID-ArdiNeural',
|
25 |
+
'Indonesian-Gadis- (Woman)': 'id-ID-GadisNeural',
|
26 |
+
'Hebrew-Avri- (Man)': 'he-IL-AvriNeural',
|
27 |
+
'Hebrew-Hila- (Woman)': 'he-IL-HilaNeural',
|
28 |
+
'Italian-Isabella- (Woman)': 'it-IT-IsabellaNeural',
|
29 |
+
'Italian-Diego- (Man)': 'it-IT-DiegoNeural',
|
30 |
+
'Italian-Elsa- (Woman)': 'it-IT-ElsaNeural',
|
31 |
+
'Dutch-Colette- (Woman)': 'nl-NL-ColetteNeural',
|
32 |
+
'Dutch-Fenna- (Woman)': 'nl-NL-FennaNeural',
|
33 |
+
'Dutch-Maarten- (Man)': 'nl-NL-MaartenNeural',
|
34 |
+
'Malese-Osman- (Man)': 'ms-MY-OsmanNeural',
|
35 |
+
'Malese-Yasmin- (Woman)': 'ms-MY-YasminNeural',
|
36 |
+
'Norwegian-Pernille- (Woman)': 'nb-NO-PernilleNeural',
|
37 |
+
'Norwegian-Finn- (Man)': 'nb-NO-FinnNeural',
|
38 |
+
'Swedish-Sofie- (Woman)': 'sv-SE-SofieNeural',
|
39 |
+
'ArabicSwedish-Mattias- (Man)': 'sv-SE-MattiasNeural',
|
40 |
+
'Arabic-Hamed- (Man)': 'ar-SA-HamedNeural',
|
41 |
+
'Arabic-Zariyah- (Woman)': 'ar-SA-ZariyahNeural',
|
42 |
+
'Greek-Athina- (Woman)': 'el-GR-AthinaNeural',
|
43 |
+
'Greek-Nestoras- (Man)': 'el-GR-NestorasNeural',
|
44 |
+
'German-Katja- (Woman)': 'de-DE-KatjaNeural',
|
45 |
+
'German-Amala- (Woman)': 'de-DE-AmalaNeural',
|
46 |
+
'German-Conrad- (Man)': 'de-DE-ConradNeural',
|
47 |
+
'German-Killian- (Man)': 'de-DE-KillianNeural',
|
48 |
+
'Afrikaans-Adri- (Woman)': 'af-ZA-AdriNeural',
|
49 |
+
'Afrikaans-Willem- (Man)': 'af-ZA-WillemNeural',
|
50 |
+
'Ethiopian-Ameha- (Man)': 'am-ET-AmehaNeural',
|
51 |
+
'Ethiopian-Mekdes- (Woman)': 'am-ET-MekdesNeural',
|
52 |
+
'Arabic (UAD)-Fatima- (Woman)': 'ar-AE-FatimaNeural',
|
53 |
+
'Arabic (UAD)-Hamdan- (Man)': 'ar-AE-HamdanNeural',
|
54 |
+
'Arabic (Bahrain)-Ali- (Man)': 'ar-BH-AliNeural',
|
55 |
+
'Arabic (Bahrain)-Laila- (Woman)': 'ar-BH-LailaNeural',
|
56 |
+
'Arabic (Algeria)-Ismael- (Man)': 'ar-DZ-IsmaelNeural',
|
57 |
+
'Arabic (Egypt)-Salma- (Woman)': 'ar-EG-SalmaNeural',
|
58 |
+
'Arabic (Egypt)-Shakir- (Man)': 'ar-EG-ShakirNeural',
|
59 |
+
'Arabic (Iraq)-Bassel- (Man)': 'ar-IQ-BasselNeural',
|
60 |
+
'Arabic (Iraq)-Rana- (Woman)': 'ar-IQ-RanaNeural',
|
61 |
+
'Arabic (Jordan)-Sana- (Woman)': 'ar-JO-SanaNeural',
|
62 |
+
'Arabic (Jordan)-Taim- (Man)': 'ar-JO-TaimNeural',
|
63 |
+
'Arabic (Kuwait)-Fahed- (Man)': 'ar-KW-FahedNeural',
|
64 |
+
'Arabic (Kuwait)-Noura- (Woman)': 'ar-KW-NouraNeural',
|
65 |
+
'Arabic (Lebanon)-Layla- (Woman)': 'ar-LB-LaylaNeural',
|
66 |
+
'Arabic (Lebanon)-Rami- (Man)': 'ar-LB-RamiNeural',
|
67 |
+
'Arabic (Libya)-Iman- (Woman)': 'ar-LY-ImanNeural',
|
68 |
+
'Arabic (Libya)-Omar- (Man)': 'ar-LY-OmarNeural',
|
69 |
+
'Arabic (Morocco)-Jamal- (Man)': 'ar-MA-JamalNeural',
|
70 |
+
'Arabic (Morocco)-Mouna- (Woman)': 'ar-MA-MounaNeural',
|
71 |
+
'Arabic (Oman)-Abdullah- (Man)': 'ar-OM-AbdullahNeural',
|
72 |
+
'Arabic (Oman)-Aysha- (Woman)': 'ar-OM-AyshaNeural',
|
73 |
+
'Arabic (Qatar)-Amal- (Woman)': 'ar-QA-AmalNeural',
|
74 |
+
'Arabic (Qatar)-Moaz- (Man)': 'ar-QA-MoazNeural',
|
75 |
+
'Arabic (Syrian Arab Republic)-Amany- (Woman)': 'ar-SY-AmanyNeural',
|
76 |
+
'Arabic (Syrian Arab Republic)-Laith- (Man)': 'ar-SY-LaithNeural',
|
77 |
+
'Arabic (Tunisia)-Hedi- (Man)': 'ar-TN-HediNeural',
|
78 |
+
'Arabic (Tunisia)-Reem- (Woman)': 'ar-TN-ReemNeural',
|
79 |
+
'Arabic (Yemen )-Maryam- (Woman)': 'ar-YE-MaryamNeural',
|
80 |
+
'Arabic (Yemen )-Saleh- (Man)': 'ar-YE-SalehNeural',
|
81 |
+
'Azerbaijani-Babek- (Man)': 'az-AZ-BabekNeural',
|
82 |
+
'Azerbaijani-Banu- (Woman)': 'az-AZ-BanuNeural',
|
83 |
+
'Bulgarian-Borislav- (Man)': 'bg-BG-BorislavNeural',
|
84 |
+
'Bulgarian-Kalina- (Woman)': 'bg-BG-KalinaNeural',
|
85 |
+
'Bengali (Bangladesh)-Nabanita- (Woman)': 'bn-BD-NabanitaNeural',
|
86 |
+
'Bengali (Bangladesh)-Pradeep- (Man)': 'bn-BD-PradeepNeural',
|
87 |
+
'Bengali (India)-Bashkar- (Man)': 'bn-IN-BashkarNeural',
|
88 |
+
'Bengali (India)-Tanishaa- (Woman)': 'bn-IN-TanishaaNeural',
|
89 |
+
'Bosniak (Bosnia and Herzegovina)-Goran- (Man)': 'bs-BA-GoranNeural',
|
90 |
+
'Bosniak (Bosnia and Herzegovina)-Vesna- (Woman)': 'bs-BA-VesnaNeural',
|
91 |
+
'Catalan (Spain)-Joana- (Woman)': 'ca-ES-JoanaNeural',
|
92 |
+
'Catalan (Spain)-Enric- (Man)': 'ca-ES-EnricNeural',
|
93 |
+
'Czech (Czech Republic)-Antonin- (Man)': 'cs-CZ-AntoninNeural',
|
94 |
+
'Czech (Czech Republic)-Vlasta- (Woman)': 'cs-CZ-VlastaNeural',
|
95 |
+
'Welsh (UK)-Aled- (Man)': 'cy-GB-AledNeural',
|
96 |
+
'Welsh (UK)-Nia- (Woman)': 'cy-GB-NiaNeural',
|
97 |
+
'Danish (Denmark)-Christel- (Woman)': 'da-DK-ChristelNeural',
|
98 |
+
'Danish (Denmark)-Jeppe- (Man)': 'da-DK-JeppeNeural',
|
99 |
+
'German (Austria)-Ingrid- (Woman)': 'de-AT-IngridNeural',
|
100 |
+
'German (Austria)-Jonas- (Man)': 'de-AT-JonasNeural',
|
101 |
+
'German (Switzerland)-Jan- (Man)': 'de-CH-JanNeural',
|
102 |
+
'German (Switzerland)-Leni- (Woman)': 'de-CH-LeniNeural',
|
103 |
+
'English (Australia)-Natasha- (Woman)': 'en-AU-NatashaNeural',
|
104 |
+
'English (Australia)-William- (Man)': 'en-AU-WilliamNeural',
|
105 |
+
'English (Canada)-Clara- (Woman)': 'en-CA-ClaraNeural',
|
106 |
+
'English (Canada)-Liam- (Man)': 'en-CA-LiamNeural',
|
107 |
+
'English (UK)-Libby- (Woman)': 'en-GB-LibbyNeural',
|
108 |
+
'English (UK)-Maisie- (Woman)': 'en-GB-MaisieNeural',
|
109 |
+
'English (UK)-Ryan- (Man)': 'en-GB-RyanNeural',
|
110 |
+
'English (UK)-Sonia- (Woman)': 'en-GB-SoniaNeural',
|
111 |
+
'English (UK)-Thomas- (Man)': 'en-GB-ThomasNeural',
|
112 |
+
'English (Hong Kong)-Sam- (Man)': 'en-HK-SamNeural',
|
113 |
+
'English (Hong Kong)-Yan- (Woman)': 'en-HK-YanNeural',
|
114 |
+
'English (Ireland)-Connor- (Man)': 'en-IE-ConnorNeural',
|
115 |
+
'English (Ireland)-Emily- (Woman)': 'en-IE-EmilyNeural',
|
116 |
+
'English (India)-Neerja- (Woman)': 'en-IN-NeerjaNeural',
|
117 |
+
'English (India)-Prabhat- (Man)': 'en-IN-PrabhatNeural',
|
118 |
+
'English (Kenya)-Asilia- (Woman)': 'en-KE-AsiliaNeural',
|
119 |
+
'English (Kenya)-Chilemba- (Man)': 'en-KE-ChilembaNeural',
|
120 |
+
'English (Nigeria)-Abeo- (Man)': 'en-NG-AbeoNeural',
|
121 |
+
'English (Nigeria)-Ezinne- (Woman)': 'en-NG-EzinneNeural',
|
122 |
+
'English (New Zealand)-Mitchell- (Man)': 'en-NZ-MitchellNeural',
|
123 |
+
'English (Philippines)-James- (Man)': 'en-PH-JamesNeural',
|
124 |
+
'English (Philippines)-Rosa- (Woman)': 'en-PH-RosaNeural',
|
125 |
+
'English (Singapore)-Luna- (Woman)': 'en-SG-LunaNeural',
|
126 |
+
'English (Singapore)-Wayne- (Man)': 'en-SG-WayneNeural',
|
127 |
+
'English (Tanzania)-Elimu- (Man)': 'en-TZ-ElimuNeural',
|
128 |
+
'English (Tanzania)-Imani- (Woman)': 'en-TZ-ImaniNeural',
|
129 |
+
'English (South Africa)-Leah- (Woman)': 'en-ZA-LeahNeural',
|
130 |
+
'English (South Africa)-Luke- (Man)': 'en-ZA-LukeNeural',
|
131 |
+
'Spanish (Argentina)-Elena- (Woman)': 'es-AR-ElenaNeural',
|
132 |
+
'Spanish (Argentina)-Tomas- (Man)': 'es-AR-TomasNeural',
|
133 |
+
'Spanish (Bolivia)-Marcelo- (Man)': 'es-BO-MarceloNeural',
|
134 |
+
'Spanish (Bolivia)-Sofia- (Woman)': 'es-BO-SofiaNeural',
|
135 |
+
'Spanish (Colombia)-Gonzalo- (Man)': 'es-CO-GonzaloNeural',
|
136 |
+
'Spanish (Colombia)-Salome- (Woman)': 'es-CO-SalomeNeural',
|
137 |
+
'Spanish (Costa Rica)-Juan- (Man)': 'es-CR-JuanNeural',
|
138 |
+
'Spanish (Costa Rica)-Maria- (Woman)': 'es-CR-MariaNeural',
|
139 |
+
'Spanish (Cuba)-Belkys- (Woman)': 'es-CU-BelkysNeural',
|
140 |
+
'Spanish (Dominican Republic)-Emilio- (Man)': 'es-DO-EmilioNeural',
|
141 |
+
'Spanish (Dominican Republic)-Ramona- (Woman)': 'es-DO-RamonaNeural',
|
142 |
+
'Spanish (Ecuador)-Andrea- (Woman)': 'es-EC-AndreaNeural',
|
143 |
+
'Spanish (Ecuador)-Luis- (Man)': 'es-EC-LuisNeural',
|
144 |
+
'Spanish (Spain)-Alvaro- (Man)': 'es-ES-AlvaroNeural',
|
145 |
+
'Spanish (Spain)-Elvira- (Woman)': 'es-ES-ElviraNeural',
|
146 |
+
'Spanish (Equatorial Guinea)-Teresa- (Woman)': 'es-GQ-TeresaNeural',
|
147 |
+
'Spanish (Guatemala)-Andres- (Man)': 'es-GT-AndresNeural',
|
148 |
+
'Spanish (Guatemala)-Marta- (Woman)': 'es-GT-MartaNeural',
|
149 |
+
'Spanish (Honduras)-Carlos- (Man)': 'es-HN-CarlosNeural',
|
150 |
+
'Spanish (Honduras)-Karla- (Woman)': 'es-HN-KarlaNeural',
|
151 |
+
'Spanish (Nicaragua)-Federico- (Man)': 'es-NI-FedericoNeural',
|
152 |
+
'Spanish (Nicaragua)-Yolanda- (Woman)': 'es-NI-YolandaNeural',
|
153 |
+
'Spanish (Panama)-Margarita- (Woman)': 'es-PA-MargaritaNeural',
|
154 |
+
'Spanish (Panama)-Roberto- (Man)': 'es-PA-RobertoNeural',
|
155 |
+
'Spanish (Peru)-Alex- (Man)': 'es-PE-AlexNeural',
|
156 |
+
'Spanish (Peru)-Camila- (Woman)': 'es-PE-CamilaNeural',
|
157 |
+
'Spanish (Puerto Rico)-Karina- (Woman)': 'es-PR-KarinaNeural',
|
158 |
+
'Spanish (Puerto Rico)-Victor- (Man)': 'es-PR-VictorNeural',
|
159 |
+
'Spanish (Paraguay)-Mario- (Man)': 'es-PY-MarioNeural',
|
160 |
+
'Spanish (Paraguay)-Tania- (Woman)': 'es-PY-TaniaNeural',
|
161 |
+
'Spanish (El Salvador)-Lorena- (Woman)': 'es-SV-LorenaNeural',
|
162 |
+
'Spanish (El Salvador)-Rodrigo- (Man)': 'es-SV-RodrigoNeural',
|
163 |
+
'Spanish (United States)-Alonso- (Man)': 'es-US-AlonsoNeural',
|
164 |
+
'Spanish (United States)-Paloma- (Woman)': 'es-US-PalomaNeural',
|
165 |
+
'Spanish (Uruguay)-Mateo- (Man)': 'es-UY-MateoNeural',
|
166 |
+
'Spanish (Uruguay)-Valentina- (Woman)': 'es-UY-ValentinaNeural',
|
167 |
+
'Spanish (Venezuela)-Paola- (Woman)': 'es-VE-PaolaNeural',
|
168 |
+
'Spanish (Venezuela)-Sebastian- (Man)': 'es-VE-SebastianNeural',
|
169 |
+
'Estonian (Estonia)-Anu- (Woman)': 'et-EE-AnuNeural',
|
170 |
+
'Estonian (Estonia)-Kert- (Man)': 'et-EE-KertNeural',
|
171 |
+
'Persian (Iran)-Dilara- (Woman)': 'fa-IR-DilaraNeural',
|
172 |
+
'Persian (Iran)-Farid- (Man)': 'fa-IR-FaridNeural',
|
173 |
+
'Finnish (Finland)-Harri- (Man)': 'fi-FI-HarriNeural',
|
174 |
+
'Finnish (Finland)-Noora- (Woman)': 'fi-FI-NooraNeural',
|
175 |
+
'French (Belgium)-Charline- (Woman)': 'fr-BE-CharlineNeural',
|
176 |
+
'French (Belgium)-Gerard- (Man)': 'fr-BE-GerardNeural',
|
177 |
+
'French (Canada)-Sylvie- (Woman)': 'fr-CA-SylvieNeural',
|
178 |
+
'French (Canada)-Antoine- (Man)': 'fr-CA-AntoineNeural',
|
179 |
+
'French (Canada)-Jean- (Man)': 'fr-CA-JeanNeural',
|
180 |
+
'French (Switzerland)-Ariane- (Woman)': 'fr-CH-ArianeNeural',
|
181 |
+
'French (Switzerland)-Fabrice- (Man)': 'fr-CH-FabriceNeural',
|
182 |
+
'Irish (Ireland)-Colm- (Man)': 'ga-IE-ColmNeural',
|
183 |
+
'Irish (Ireland)-Orla- (Woman)': 'ga-IE-OrlaNeural',
|
184 |
+
'Galician (Spain)-Roi- (Man)': 'gl-ES-RoiNeural',
|
185 |
+
'Galician (Spain)-Sabela- (Woman)': 'gl-ES-SabelaNeural',
|
186 |
+
'Gujarati (India)-Dhwani- (Woman)': 'gu-IN-DhwaniNeural',
|
187 |
+
'Gujarati (India)-Niranjan- (Man)': 'gu-IN-NiranjanNeural',
|
188 |
+
'Hindi (India)-Madhur- (Man)': 'hi-IN-MadhurNeural',
|
189 |
+
'Hindi (India)-Swara- (Woman)': 'hi-IN-SwaraNeural',
|
190 |
+
'Croatian (Croatia)-Gabrijela- (Woman)': 'hr-HR-GabrijelaNeural',
|
191 |
+
'Croatian (Croatia)-Srecko- (Man)': 'hr-HR-SreckoNeural',
|
192 |
+
'Hungarian (Hungary)-Noemi- (Woman)': 'hu-HU-NoemiNeural',
|
193 |
+
'Hungarian (Hungary)-Tamas- (Man)': 'hu-HU-TamasNeural',
|
194 |
+
'Icelandic (Iceland)-Gudrun- (Woman)': 'is-IS-GudrunNeural',
|
195 |
+
'Icelandic (Iceland)-Gunnar- (Man)': 'is-IS-GunnarNeural',
|
196 |
+
'Javanese (Indonesia)-Dimas- (Man)': 'jv-ID-DimasNeural',
|
197 |
+
'Javanese (Indonesia)-Siti- (Woman)': 'jv-ID-SitiNeural',
|
198 |
+
'Georgian (Georgia)-Eka- (Woman)': 'ka-GE-EkaNeural',
|
199 |
+
'Georgian (Georgia)-Giorgi- (Man)': 'ka-GE-GiorgiNeural',
|
200 |
+
'Kazakh (Kazakhstan)-Aigul- (Woman)': 'kk-KZ-AigulNeural',
|
201 |
+
'Kazakh (Kazakhstan)-Daulet- (Man)': 'kk-KZ-DauletNeural',
|
202 |
+
'Khmer (Cambodia)-Piseth- (Man)': 'km-KH-PisethNeural',
|
203 |
+
'Khmer (Cambodia)-Sreymom- (Woman)': 'km-KH-SreymomNeural',
|
204 |
+
'Kannada (India)-Gagan- (Man)': 'kn-IN-GaganNeural',
|
205 |
+
'Kannada (India)-Sapna- (Woman)': 'kn-IN-SapnaNeural',
|
206 |
+
'Lao (Laos)-Chanthavong- (Man)': 'lo-LA-ChanthavongNeural',
|
207 |
+
'Lao (Laos)-Keomany- (Woman)': 'lo-LA-KeomanyNeural',
|
208 |
+
'Lithuanian (Lithuania)-Leonas- (Man)': 'lt-LT-LeonasNeural',
|
209 |
+
'Lithuanian (Lithuania)-Ona- (Woman)': 'lt-LT-OnaNeural',
|
210 |
+
'Latvian (Latvia)-Everita- (Woman)': 'lv-LV-EveritaNeural',
|
211 |
+
'Latvian (Latvia)-Nils- (Man)': 'lv-LV-NilsNeural',
|
212 |
+
'Macedonian (North Macedonia)-Aleksandar- (Man)': 'mk-MK-AleksandarNeural',
|
213 |
+
'Macedonian (North Macedonia)-Marija- (Woman)': 'mk-MK-MarijaNeural',
|
214 |
+
'Malayalam (India)-Midhun- (Man)': 'ml-IN-MidhunNeural',
|
215 |
+
'Malayalam (India)-Sobhana- (Woman)': 'ml-IN-SobhanaNeural',
|
216 |
+
'Mongolian (Mongolia)-Bataa- (Man)': 'mn-MN-BataaNeural',
|
217 |
+
'Mongolian (Mongolia)-Yesui- (Woman)': 'mn-MN-YesuiNeural',
|
218 |
+
'Marathi (India)-Aarohi- (Woman)': 'mr-IN-AarohiNeural',
|
219 |
+
'Marathi (India)-Manohar- (Man)': 'mr-IN-ManoharNeural',
|
220 |
+
'Maltese (Malta)-Grace- (Woman)': 'mt-MT-GraceNeural',
|
221 |
+
'Maltese (Malta)-Joseph- (Man)': 'mt-MT-JosephNeural',
|
222 |
+
'Burmese (Myanmar)-Nilar- (Woman)': 'my-MM-NilarNeural',
|
223 |
+
'Burmese (Myanmar)-Thiha- (Man)': 'my-MM-ThihaNeural',
|
224 |
+
'Nepali (Nepal)-Hemkala- (Woman)': 'ne-NP-HemkalaNeural',
|
225 |
+
'Nepali (Nepal)-Sagar- (Man)': 'ne-NP-SagarNeural',
|
226 |
+
'Dutch (Belgium)-Arnaud- (Man)': 'nl-BE-ArnaudNeural',
|
227 |
+
'Dutch (Belgium)-Dena- (Woman)': 'nl-BE-DenaNeural',
|
228 |
+
'Polish (Poland)-Marek- (Man)': 'pl-PL-MarekNeural',
|
229 |
+
'Polish (Poland)-Zofia- (Woman)': 'pl-PL-ZofiaNeural',
|
230 |
+
'Pashto (Afghanistan)-Gul Nawaz- (Man)': 'ps-AF-Gul',}
|