Bingsu commited on
Commit
ddaad3d
1 Parent(s): 0acb082

Upload with huggingface_hub

Browse files
config.json ADDED
@@ -0,0 +1,91 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.1,
3
+ "apply_spec_augment": true,
4
+ "architectures": [
5
+ "SpeechT5ForTextToSpeech"
6
+ ],
7
+ "attention_dropout": 0.1,
8
+ "bos_token_id": 1,
9
+ "conv_bias": false,
10
+ "conv_dim": [
11
+ 512,
12
+ 512,
13
+ 512,
14
+ 512,
15
+ 512,
16
+ 512,
17
+ 512
18
+ ],
19
+ "conv_kernel": [
20
+ 10,
21
+ 3,
22
+ 3,
23
+ 3,
24
+ 3,
25
+ 2,
26
+ 2
27
+ ],
28
+ "conv_stride": [
29
+ 5,
30
+ 2,
31
+ 2,
32
+ 2,
33
+ 2,
34
+ 2,
35
+ 2
36
+ ],
37
+ "decoder_attention_heads": 4,
38
+ "decoder_ffn_dim": 1024,
39
+ "decoder_layerdrop": 0.1,
40
+ "decoder_layers": 6,
41
+ "decoder_start_token_id": 2,
42
+ "encoder_attention_heads": 4,
43
+ "encoder_ffn_dim": 1024,
44
+ "encoder_layerdrop": 0.1,
45
+ "encoder_layers": 12,
46
+ "encoder_max_relative_position": 160,
47
+ "eos_token_id": 2,
48
+ "feat_extract_activation": "gelu",
49
+ "feat_extract_norm": "group",
50
+ "feat_proj_dropout": 0.0,
51
+ "guided_attention_loss_num_heads": 2,
52
+ "guided_attention_loss_scale": 10.0,
53
+ "guided_attention_loss_sigma": 0.4,
54
+ "hidden_act": "gelu",
55
+ "hidden_dropout": 0.1,
56
+ "hidden_size": 256,
57
+ "initializer_range": 0.02,
58
+ "is_encoder_decoder": true,
59
+ "layer_norm_eps": 1e-05,
60
+ "mask_feature_length": 10,
61
+ "mask_feature_min_masks": 0,
62
+ "mask_feature_prob": 0.0,
63
+ "mask_time_length": 10,
64
+ "mask_time_min_masks": 2,
65
+ "mask_time_prob": 0.05,
66
+ "max_speech_positions": 1876,
67
+ "max_text_positions": 600,
68
+ "model_type": "speecht5",
69
+ "num_conv_pos_embedding_groups": 16,
70
+ "num_conv_pos_embeddings": 128,
71
+ "num_feat_extract_layers": 7,
72
+ "num_mel_bins": 80,
73
+ "num_speakers": 13,
74
+ "pad_token_id": 0,
75
+ "positional_dropout": 0.1,
76
+ "reduction_factor": 2,
77
+ "scale_embedding": false,
78
+ "speaker_embedding_dim": 128,
79
+ "speech_decoder_postnet_dropout": 0.5,
80
+ "speech_decoder_postnet_kernel": 5,
81
+ "speech_decoder_postnet_layers": 5,
82
+ "speech_decoder_postnet_units": 256,
83
+ "speech_decoder_prenet_dropout": 0.5,
84
+ "speech_decoder_prenet_layers": 2,
85
+ "speech_decoder_prenet_units": 256,
86
+ "torch_dtype": "float32",
87
+ "transformers_version": "4.29.0.dev0",
88
+ "use_cache": true,
89
+ "use_guided_attention_loss": true,
90
+ "vocab_size": 392
91
+ }
preprocessor_config.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": false,
3
+ "feature_extractor_type": "SpeechT5FeatureExtractor",
4
+ "feature_size": 1,
5
+ "fmax": 7600,
6
+ "fmin": 80,
7
+ "frame_signal_scale": 1.0,
8
+ "hop_length": 16,
9
+ "mel_floor": 1e-10,
10
+ "num_mel_bins": 80,
11
+ "padding_side": "right",
12
+ "padding_value": 0.0,
13
+ "processor_class": "SpeechT5Processor",
14
+ "reduction_factor": 2,
15
+ "return_attention_mask": true,
16
+ "sampling_rate": 16000,
17
+ "win_function": "hann_window",
18
+ "win_length": 64
19
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:319e4c805cc0303a45f68092fcf525587ca7dff3edde051dc2abb75008da0fa8
3
+ size 72288837
speaker_embedding.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b82296aacd6654b57bf4c7f0f1ea1080e7f4242af34543ec8af7628cb427f49
3
+ size 7497
special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": "<pad>",
5
+ "unk_token": "<unk>"
6
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "clean_up_tokenization_spaces": true,
4
+ "do_lower_case": false,
5
+ "do_phonemize": true,
6
+ "eos_token": "</s>",
7
+ "model_max_length": 1000000000000000019884624838656,
8
+ "pad_token": "<pad>",
9
+ "phone_delimiter_token": " ",
10
+ "phonemizer_backend": "espeak",
11
+ "phonemizer_lang": "ko",
12
+ "tokenizer_class": "Wav2Vec2PhonemeCTCTokenizer",
13
+ "tokenizer_file": null,
14
+ "unk_token": "<unk>",
15
+ "word_delimiter_token": "|"
16
+ }
vocab.json ADDED
@@ -0,0 +1,394 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "1": 184,
3
+ "</s>": 2,
4
+ "<pad>": 0,
5
+ "<s>": 1,
6
+ "<unk>": 3,
7
+ "??": 85,
8
+ "N": 280,
9
+ "S": 208,
10
+ "X": 326,
11
+ "a": 9,
12
+ "a.": 250,
13
+ "a.ː": 251,
14
+ "a1": 354,
15
+ "a2": 187,
16
+ "a4": 391,
17
+ "a5": 106,
18
+ "ai": 265,
19
+ "ai2": 231,
20
+ "ai5": 130,
21
+ "aiɜ": 181,
22
+ "au": 282,
23
+ "aɜ": 160,
24
+ "aɨ": 193,
25
+ "aɪ": 37,
26
+ "aɪə": 139,
27
+ "aɪɚ": 131,
28
+ "aʊ": 53,
29
+ "aː": 73,
30
+ "ã": 330,
31
+ "b": 26,
32
+ "bʰ": 296,
33
+ "bʲ": 178,
34
+ "bː": 239,
35
+ "c": 200,
36
+ "cʰ": 356,
37
+ "cʰcʰ": 371,
38
+ "cː": 334,
39
+ "d": 12,
40
+ "dZ": 216,
41
+ "d[": 152,
42
+ "d^": 300,
43
+ "dzː": 382,
44
+ "dʑ": 121,
45
+ "dʑʲ": 345,
46
+ "dʒ": 60,
47
+ "dʒʲ": 324,
48
+ "dʒː": 338,
49
+ "dʰ": 303,
50
+ "dʰː": 384,
51
+ "dʲ": 115,
52
+ "dʲʲ": 383,
53
+ "dː": 253,
54
+ "dˤ": 205,
55
+ "dˤdˤ": 337,
56
+ "e": 16,
57
+ "e:": 319,
58
+ "ee": 255,
59
+ "ei2": 198,
60
+ "ei5": 125,
61
+ "eiɜ": 212,
62
+ "eɑ": 317,
63
+ "eə": 269,
64
+ "eɪ": 44,
65
+ "eʊ": 163,
66
+ "eː": 51,
67
+ "ẽ": 360,
68
+ "ẽː": 332,
69
+ "e̞": 123,
70
+ "e̞e̞": 380,
71
+ "f": 23,
72
+ "fʲ": 230,
73
+ "h": 39,
74
+ "i": 10,
75
+ "i.": 270,
76
+ "i.1": 335,
77
+ "i.2": 218,
78
+ "i.4": 388,
79
+ "i.5": 99,
80
+ "i.ɜ": 162,
81
+ "i.ː": 261,
82
+ "i1": 306,
83
+ "i2": 135,
84
+ "i4": 339,
85
+ "i5": 82,
86
+ "i:": 389,
87
+ "ie": 272,
88
+ "iou1": 352,
89
+ "iou2": 199,
90
+ "iou4": 355,
91
+ "iou5": 215,
92
+ "iouɜ": 190,
93
+ "iɑ1": 366,
94
+ "iɑ2": 227,
95
+ "iɑ5": 129,
96
+ "iɑɜ": 223,
97
+ "iə": 81,
98
+ "iɛ1": 357,
99
+ "iɛ2": 194,
100
+ "iɛ4": 379,
101
+ "iɛ5": 107,
102
+ "iɛɜ": 146,
103
+ "iɜ": 105,
104
+ "iʊ": 290,
105
+ "iː": 30,
106
+ "iː1": 373,
107
+ "iːː": 364,
108
+ "ĩ": 314,
109
+ "i̪1": 281,
110
+ "i̪2": 247,
111
+ "i̪4": 310,
112
+ "i̪5": 157,
113
+ "i̪ɜ": 258,
114
+ "j": 24,
115
+ "ja": 110,
116
+ "ju": 141,
117
+ "k": 11,
118
+ "kh": 132,
119
+ "kʰ": 288,
120
+ "kʰː": 375,
121
+ "kʲ": 168,
122
+ "kː": 201,
123
+ "l": 8,
124
+ "lː": 386,
125
+ "l̩": 243,
126
+ "m": 13,
127
+ "mʲ": 119,
128
+ "n": 4,
129
+ "nʲ": 86,
130
+ "nʲʲ": 309,
131
+ "n̩": 169,
132
+ "o": 19,
133
+ "o1": 263,
134
+ "o2": 164,
135
+ "o4": 275,
136
+ "o5": 126,
137
+ "o:": 277,
138
+ "oe": 204,
139
+ "oe:": 305,
140
+ "onɡ2": 210,
141
+ "onɡ5": 127,
142
+ "onɡɜ": 192,
143
+ "ou1": 378,
144
+ "ou2": 228,
145
+ "ou5": 147,
146
+ "ouɜ": 248,
147
+ "oɜ": 144,
148
+ "oɪ": 114,
149
+ "oʊ": 49,
150
+ "oː": 55,
151
+ "oːɹ": 77,
152
+ "õ": 292,
153
+ "o̞": 197,
154
+ "o̞o̞": 376,
155
+ "p": 18,
156
+ "pf": 225,
157
+ "ph": 171,
158
+ "pʰ": 344,
159
+ "pʲ": 167,
160
+ "pː": 145,
161
+ "q": 101,
162
+ "qː": 285,
163
+ "r": 31,
164
+ "r.": 304,
165
+ "rʲ": 96,
166
+ "r̝": 232,
167
+ "r̝̊": 217,
168
+ "r̩": 254,
169
+ "s": 5,
170
+ "s.": 83,
171
+ "s^": 220,
172
+ "sx": 291,
173
+ "sʲ": 102,
174
+ "s̪": 180,
175
+ "s̪ː": 301,
176
+ "t": 6,
177
+ "tS": 219,
178
+ "t[": 138,
179
+ "t^": 262,
180
+ "t^ː": 331,
181
+ "th": 112,
182
+ "ts": 54,
183
+ "ts.": 95,
184
+ "ts.h": 118,
185
+ "tsh": 154,
186
+ "tsʲ": 312,
187
+ "tsː": 320,
188
+ "tɕ": 84,
189
+ "tɕh": 117,
190
+ "tɕʲ": 329,
191
+ "tʃ": 66,
192
+ "tʃʰ": 289,
193
+ "tʃʲ": 124,
194
+ "tʃː": 276,
195
+ "tʰ": 294,
196
+ "tʲ": 92,
197
+ "tː": 150,
198
+ "t̪": 189,
199
+ "t̪ː": 316,
200
+ "u": 34,
201
+ "u\"": 211,
202
+ "u.": 284,
203
+ "u.ː": 297,
204
+ "u1": 279,
205
+ "u2": 143,
206
+ "u4": 318,
207
+ "u5": 103,
208
+ "u:": 260,
209
+ "ua1": 347,
210
+ "ua2": 273,
211
+ "ua4": 348,
212
+ "ua5": 236,
213
+ "uai5": 295,
214
+ "uaiɜ": 328,
215
+ "uaɜ": 249,
216
+ "uei2": 311,
217
+ "uei5": 158,
218
+ "ueiɜ": 266,
219
+ "ui": 351,
220
+ "uo": 293,
221
+ "uo1": 387,
222
+ "uo2": 241,
223
+ "uo5": 151,
224
+ "uoɜ": 153,
225
+ "uə2": 298,
226
+ "uə5": 224,
227
+ "uəɜ": 233,
228
+ "uɜ": 149,
229
+ "uɨ": 173,
230
+ "uɪ": 237,
231
+ "uː": 46,
232
+ "ũ": 191,
233
+ "v": 25,
234
+ "vʲ": 122,
235
+ "w": 32,
236
+ "x": 62,
237
+ "xʲ": 377,
238
+ "y": 47,
239
+ "y1": 381,
240
+ "y2": 202,
241
+ "y5": 148,
242
+ "y:": 287,
243
+ "yi": 341,
244
+ "yiɜ": 302,
245
+ "yu2": 308,
246
+ "yu5": 240,
247
+ "yuɜ": 346,
248
+ "yæ2": 283,
249
+ "yæ5": 252,
250
+ "yæɜ": 186,
251
+ "yə2": 358,
252
+ "yə5": 222,
253
+ "yəɜ": 286,
254
+ "yɛ2": 333,
255
+ "yɛ5": 242,
256
+ "yɛ5ʲ": 390,
257
+ "yɛɜ": 206,
258
+ "yɜ": 177,
259
+ "yː": 79,
260
+ "z": 21,
261
+ "ä": 166,
262
+ "ää": 362,
263
+ "æ": 36,
264
+ "æi": 259,
265
+ "æiː": 349,
266
+ "æː": 267,
267
+ "ç": 64,
268
+ "ð": 22,
269
+ "ø": 76,
270
+ "øi": 369,
271
+ "øː": 108,
272
+ "ħ": 137,
273
+ "ŋ": 42,
274
+ "œ": 75,
275
+ "œː": 368,
276
+ "œ̃": 88,
277
+ "ũ": 374,
278
+ "ɐ": 20,
279
+ "ɐɐ": 350,
280
+ "ɐ̃": 165,
281
+ "ɐ̃ʊ̃": 183,
282
+ "ɑ": 59,
283
+ "ɑ1": 274,
284
+ "ɑ2": 179,
285
+ "ɑ4": 315,
286
+ "ɑ5": 113,
287
+ "ɑ:": 278,
288
+ "ɑu2": 172,
289
+ "ɑu5": 128,
290
+ "ɑuɜ": 196,
291
+ "ɑɜ": 155,
292
+ "ɑɨ": 161,
293
+ "ɑː": 41,
294
+ "ɑːɹ": 68,
295
+ "ɑ̃": 48,
296
+ "ɒ": 170,
297
+ "ɔ": 40,
298
+ "ɔø": 90,
299
+ "ɔɨ": 195,
300
+ "ɔɪ": 100,
301
+ "ɔː": 67,
302
+ "ɔːɹ": 71,
303
+ "ɔ̃": 56,
304
+ "ɕ": 74,
305
+ "ɕʲ": 342,
306
+ "ɖ": 203,
307
+ "ɖʰ": 359,
308
+ "ə": 7,
309
+ "ə1": 133,
310
+ "ə2": 142,
311
+ "ə4": 185,
312
+ "ə5": 116,
313
+ "əl": 61,
314
+ "ər1": 367,
315
+ "ər2": 229,
316
+ "ər4": 363,
317
+ "ər5": 257,
318
+ "ərɜ": 245,
319
+ "əɜ": 111,
320
+ "əɨ": 235,
321
+ "əɪ": 174,
322
+ "əʊ": 207,
323
+ "əː1": 372,
324
+ "ɚ": 43,
325
+ "ɛ": 14,
326
+ "ɛɪ": 268,
327
+ "ɛɹ": 94,
328
+ "ɛʊ": 336,
329
+ "ɛː": 87,
330
+ "ɛ̃": 69,
331
+ "ɜ": 58,
332
+ "ɜː": 63,
333
+ "ɟ": 159,
334
+ "ɟʰ": 343,
335
+ "ɟː": 307,
336
+ "ɡ": 35,
337
+ "ɡʰ": 322,
338
+ "ɡʲ": 209,
339
+ "ɡː": 340,
340
+ "ɣ": 57,
341
+ "ɨ": 93,
342
+ "ɨu": 226,
343
+ "ɨː": 188,
344
+ "ɪ": 17,
345
+ "ɪ^": 264,
346
+ "ɪu": 256,
347
+ "ɪuː": 370,
348
+ "ɪɹ": 97,
349
+ "ɪː": 365,
350
+ "ɫ": 175,
351
+ "ɬ": 176,
352
+ "ɭ": 89,
353
+ "ɭʲ": 98,
354
+ "ɯ": 120,
355
+ "ɯɯ": 323,
356
+ "ɯᵝ": 214,
357
+ "ɯᵝɯᵝ": 385,
358
+ "ɲ": 78,
359
+ "ɳ": 234,
360
+ "ɴ": 271,
361
+ "ɵ": 156,
362
+ "ɵː": 327,
363
+ "ɸ": 313,
364
+ "ɹ": 27,
365
+ "ɻ": 244,
366
+ "ɽ": 238,
367
+ "ɾ": 15,
368
+ "ʁ": 28,
369
+ "ʂ": 246,
370
+ "ʂʲ": 325,
371
+ "ʃ": 38,
372
+ "ʈ": 213,
373
+ "ʈʰ": 321,
374
+ "ʉ": 136,
375
+ "ʊ": 29,
376
+ "ʊə": 299,
377
+ "ʊɹ": 104,
378
+ "ʊː": 353,
379
+ "ʋ": 72,
380
+ "ʌ": 33,
381
+ "ʎ": 70,
382
+ "ʐ": 134,
383
+ "ʑ": 91,
384
+ "ʒ": 65,
385
+ "ʒʲ": 361,
386
+ "ʔ": 80,
387
+ "ʕ": 109,
388
+ "ʝ": 221,
389
+ "ʲ": 140,
390
+ "β": 45,
391
+ "θ": 52,
392
+ "χ": 182,
393
+ "ᵻ": 50
394
+ }