ayymen commited on
Commit
273f148
1 Parent(s): 92d16d7

Initial commit

Browse files
Files changed (4) hide show
  1. checkpoint_390000.pth +3 -0
  2. config.json +347 -0
  3. language_ids.json +4 -0
  4. speakers.pth +3 -0
checkpoint_390000.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1315b8eb07c0b4702623f4894eb0b66b82d5ff6cf0751f6b71aa72a47c72ff19
3
+ size 1027864162
config.json ADDED
@@ -0,0 +1,347 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "output_path": "coqui_tts",
3
+ "logger_uri": null,
4
+ "run_name": "vits_multi",
5
+ "project_name": null,
6
+ "run_description": "\ud83d\udc38Coqui trainer run.",
7
+ "print_step": 25,
8
+ "plot_step": 100,
9
+ "model_param_stats": false,
10
+ "wandb_entity": null,
11
+ "dashboard_logger": "tensorboard",
12
+ "save_on_interrupt": true,
13
+ "log_model_step": 5000,
14
+ "save_step": 5000,
15
+ "save_n_checkpoints": 5,
16
+ "save_checkpoints": true,
17
+ "save_all_best": true,
18
+ "save_best_after": 0,
19
+ "target_loss": null,
20
+ "print_eval": true,
21
+ "test_delay_epochs": -1,
22
+ "run_eval": true,
23
+ "run_eval_steps": null,
24
+ "distributed_backend": "nccl",
25
+ "distributed_url": "tcp://localhost:54321",
26
+ "mixed_precision": true,
27
+ "precision": "fp16",
28
+ "epochs": 1000,
29
+ "batch_size": 16,
30
+ "eval_batch_size": 4,
31
+ "grad_clip": [
32
+ 1000.0,
33
+ 1000.0
34
+ ],
35
+ "scheduler_after_epoch": true,
36
+ "lr": 0.001,
37
+ "optimizer": "AdamW",
38
+ "optimizer_params": {
39
+ "betas": [
40
+ 0.8,
41
+ 0.99
42
+ ],
43
+ "eps": 1e-09,
44
+ "weight_decay": 0.01
45
+ },
46
+ "lr_scheduler": null,
47
+ "lr_scheduler_params": {},
48
+ "use_grad_scaler": false,
49
+ "allow_tf32": false,
50
+ "cudnn_enable": true,
51
+ "cudnn_deterministic": false,
52
+ "cudnn_benchmark": false,
53
+ "training_seed": 54321,
54
+ "model": "vits",
55
+ "num_loader_workers": 4,
56
+ "num_eval_loader_workers": 4,
57
+ "use_noise_augment": false,
58
+ "audio": {
59
+ "fft_size": 1024,
60
+ "sample_rate": 22050,
61
+ "win_length": 1024,
62
+ "hop_length": 256,
63
+ "num_mels": 80,
64
+ "mel_fmin": 0,
65
+ "mel_fmax": null
66
+ },
67
+ "use_phonemes": false,
68
+ "phonemizer": null,
69
+ "phoneme_language": null,
70
+ "compute_input_seq_cache": true,
71
+ "text_cleaner": "no_cleaners",
72
+ "enable_eos_bos_chars": false,
73
+ "test_sentences_file": "",
74
+ "phoneme_cache_path": null,
75
+ "characters": {
76
+ "characters_class": "TTS.tts.models.vits.VitsCharacters",
77
+ "vocab_dict": null,
78
+ "pad": "<PAD>",
79
+ "eos": "<EOS>",
80
+ "bos": "<BOS>",
81
+ "blank": "<BLNK>",
82
+ "characters": "\u2d30\u2d31\u2d33\u2d37\u2d39\u2d3b\u2d3c\u2d3d\u2d40\u2d43\u2d44\u2d45\u2d47\u2d49\u2d4a\u2d4d\u2d4e\u2d4f\u2d53\u2d54\u2d55\u2d56\u2d59\u2d5a\u2d5b\u2d5c\u2d5f\u2d61\u2d62\u2d63\u2d65\u2d6f",
83
+ "punctuations": " !,.:?",
84
+ "phonemes": null,
85
+ "is_unique": true,
86
+ "is_sorted": true
87
+ },
88
+ "add_blank": true,
89
+ "batch_group_size": 5,
90
+ "loss_masking": null,
91
+ "min_audio_len": 1,
92
+ "max_audio_len": Infinity,
93
+ "min_text_len": 1,
94
+ "max_text_len": Infinity,
95
+ "compute_f0": false,
96
+ "compute_energy": false,
97
+ "compute_linear_spec": true,
98
+ "precompute_num_workers": 0,
99
+ "start_by_longest": false,
100
+ "shuffle": false,
101
+ "drop_last": false,
102
+ "datasets": [
103
+ {
104
+ "formatter": "nemo",
105
+ "dataset_name": "",
106
+ "path": "shi_tls/cut_22kHz_offset/manifests/",
107
+ "meta_file_train": "tts_manifest.json",
108
+ "ignored_speakers": null,
109
+ "language": "shi",
110
+ "phonemizer": "",
111
+ "meta_file_val": "",
112
+ "meta_file_attn_mask": ""
113
+ },
114
+ {
115
+ "formatter": "nemo",
116
+ "dataset_name": "",
117
+ "path": "shi_tls_2/cut_22kHz_offset/manifests/",
118
+ "meta_file_train": "tts_manifest.json",
119
+ "ignored_speakers": null,
120
+ "language": "shi",
121
+ "phonemizer": "",
122
+ "meta_file_val": "",
123
+ "meta_file_attn_mask": ""
124
+ },
125
+ {
126
+ "formatter": "nemo",
127
+ "dataset_name": "",
128
+ "path": "rif_tls/cut_22kHz_offset/manifests/",
129
+ "meta_file_train": "tts_manifest.json",
130
+ "ignored_speakers": null,
131
+ "language": "rif",
132
+ "phonemizer": "",
133
+ "meta_file_val": "",
134
+ "meta_file_attn_mask": ""
135
+ }
136
+ ],
137
+ "test_sentences": [
138
+ [
139
+ "\u2d30\u2d63\u2d53\u2d4d. \u2d4e\u2d30\u2d4f\u2d63\u2d30\u2d3d\u2d49\u2d4f?",
140
+ "yan",
141
+ null,
142
+ "shi"
143
+ ],
144
+ [
145
+ "\u2d61\u2d30 \u2d5c\u2d30\u2d4e\u2d56\u2d30\u2d54\u2d5c \u2d4e\u2d30 \u2d37 \u2d53\u2d3d\u2d30\u2d4f \u2d5c\u2d59\u2d3d\u2d54\u2d5c?",
146
+ "yan",
147
+ null,
148
+ "shi"
149
+ ],
150
+ [
151
+ "\u2d33\u2d4f! \u2d30\u2d37 \u2d30\u2d3d \u2d49\u2d59\u2d59\u2d33\u2d4f \u2d55\u2d31\u2d31\u2d49 \u2d49\u2d5c\u2d5c\u2d53 \u2d3d.",
152
+ "yan",
153
+ null,
154
+ "shi"
155
+ ],
156
+ [
157
+ "\u2d30\u2d54\u2d54\u2d30\u2d61 \u2d4f \u2d4d\u2d40\u2d4e\u2d4e \u2d62\u2d53\u2d3d\u2d54 \u2d30\u2d56 \u2d49\u2d40\u2d37\u2d53\u2d4e\u2d4f \u2d4f\u2d4f\u2d56!",
158
+ "yan",
159
+ null,
160
+ "shi"
161
+ ],
162
+ [
163
+ "\u2d30\u2d63\u2d53\u2d4d. \u2d4e\u2d30\u2d4f\u2d63\u2d30\u2d3d\u2d49\u2d4f?",
164
+ "sin",
165
+ null,
166
+ "shi"
167
+ ],
168
+ [
169
+ "\u2d61\u2d30 \u2d5c\u2d30\u2d4e\u2d56\u2d30\u2d54\u2d5c \u2d4e\u2d30 \u2d37 \u2d53\u2d3d\u2d30\u2d4f \u2d5c\u2d59\u2d3d\u2d54\u2d5c?",
170
+ "sin",
171
+ null,
172
+ "shi"
173
+ ],
174
+ [
175
+ "\u2d33\u2d4f! \u2d30\u2d37 \u2d30\u2d3d \u2d49\u2d59\u2d59\u2d33\u2d4f \u2d55\u2d31\u2d31\u2d49 \u2d49\u2d5c\u2d5c\u2d53 \u2d3d.",
176
+ "sin",
177
+ null,
178
+ "shi"
179
+ ],
180
+ [
181
+ "\u2d30\u2d54\u2d54\u2d30\u2d61 \u2d4f \u2d4d\u2d40\u2d4e\u2d4e \u2d62\u2d53\u2d3d\u2d54 \u2d30\u2d56 \u2d49\u2d40\u2d37\u2d53\u2d4e\u2d4f \u2d4f\u2d4f\u2d56!",
182
+ "sin",
183
+ null,
184
+ "shi"
185
+ ],
186
+ [
187
+ "\u2d30\u2d63\u2d53\u2d4d. \u2d4e\u2d30\u2d4e\u2d5b \u2d5c\u2d37\u2d4a\u2d49\u2d5c?",
188
+ "idj",
189
+ null,
190
+ "rif"
191
+ ],
192
+ [
193
+ "\u2d30\u2d47\u2d4e\u2d4e\u2d53\u2d4e \u2d49\u2d47\u2d4f\u2d3b\u2d4f \u2d53\u2d54 \u2d5c \u2d5c\u2d5c\u2d49\u2d37\u2d3c\u2d4f \u2d49\u2d63\u2d30\u2d4f.",
194
+ "idj",
195
+ null,
196
+ "rif"
197
+ ],
198
+ [
199
+ "\u2d47\u2d47\u2d49\u2d4e \u2d45 \u2d5c\u2d30\u2d37\u2d37\u2d30\u2d54\u2d5c \u2d4f\u2d4f\u2d5b!",
200
+ "idj",
201
+ null,
202
+ "rif"
203
+ ],
204
+ [
205
+ "\u2d5c\u2d3b\u2d5c\u2d5c\u2d4f \u2d30\u2d33 \u2d61\u2d53\u2d5b\u2d5b\u2d4f, \u2d5c\u2d5c\u2d54\u2d53\u2d4f \u2d45 \u2d53\u2d4e\u2d3d\u2d59\u2d30.",
206
+ "idj",
207
+ null,
208
+ "rif"
209
+ ]
210
+ ],
211
+ "eval_split_max_size": null,
212
+ "eval_split_size": 0.01,
213
+ "use_speaker_weighted_sampler": false,
214
+ "speaker_weighted_sampler_alpha": 1.0,
215
+ "use_language_weighted_sampler": false,
216
+ "language_weighted_sampler_alpha": 1.0,
217
+ "use_length_weighted_sampler": false,
218
+ "length_weighted_sampler_alpha": 1.0,
219
+ "model_args": {
220
+ "num_chars": 40,
221
+ "out_channels": 513,
222
+ "spec_segment_size": 32,
223
+ "hidden_channels": 192,
224
+ "hidden_channels_ffn_text_encoder": 768,
225
+ "num_heads_text_encoder": 2,
226
+ "num_layers_text_encoder": 6,
227
+ "kernel_size_text_encoder": 3,
228
+ "dropout_p_text_encoder": 0.1,
229
+ "dropout_p_duration_predictor": 0.5,
230
+ "kernel_size_posterior_encoder": 5,
231
+ "dilation_rate_posterior_encoder": 1,
232
+ "num_layers_posterior_encoder": 16,
233
+ "kernel_size_flow": 5,
234
+ "dilation_rate_flow": 1,
235
+ "num_layers_flow": 4,
236
+ "resblock_type_decoder": "1",
237
+ "resblock_kernel_sizes_decoder": [
238
+ 3,
239
+ 7,
240
+ 11
241
+ ],
242
+ "resblock_dilation_sizes_decoder": [
243
+ [
244
+ 1,
245
+ 3,
246
+ 5
247
+ ],
248
+ [
249
+ 1,
250
+ 3,
251
+ 5
252
+ ],
253
+ [
254
+ 1,
255
+ 3,
256
+ 5
257
+ ]
258
+ ],
259
+ "upsample_rates_decoder": [
260
+ 8,
261
+ 8,
262
+ 2,
263
+ 2
264
+ ],
265
+ "upsample_initial_channel_decoder": 512,
266
+ "upsample_kernel_sizes_decoder": [
267
+ 16,
268
+ 16,
269
+ 4,
270
+ 4
271
+ ],
272
+ "periods_multi_period_discriminator": [
273
+ 2,
274
+ 3,
275
+ 5,
276
+ 7,
277
+ 11
278
+ ],
279
+ "use_sdp": false,
280
+ "noise_scale": 1.0,
281
+ "inference_noise_scale": 0.667,
282
+ "length_scale": 1.0,
283
+ "noise_scale_dp": 1.0,
284
+ "inference_noise_scale_dp": 1.0,
285
+ "max_inference_len": null,
286
+ "init_discriminator": true,
287
+ "use_spectral_norm_disriminator": false,
288
+ "use_speaker_embedding": true,
289
+ "num_speakers": 3,
290
+ "speakers_file": "speakers.pth",
291
+ "d_vector_file": null,
292
+ "speaker_embedding_channels": 256,
293
+ "use_d_vector_file": false,
294
+ "d_vector_dim": 0,
295
+ "detach_dp_input": true,
296
+ "use_language_embedding": true,
297
+ "embedded_language_dim": 4,
298
+ "num_languages": 2,
299
+ "language_ids_file": "language_ids.json",
300
+ "use_speaker_encoder_as_loss": false,
301
+ "speaker_encoder_config_path": "",
302
+ "speaker_encoder_model_path": "",
303
+ "condition_dp_on_speaker": true,
304
+ "freeze_encoder": false,
305
+ "freeze_DP": false,
306
+ "freeze_PE": false,
307
+ "freeze_flow_decoder": false,
308
+ "freeze_waveform_decoder": false,
309
+ "encoder_sample_rate": null,
310
+ "interpolate_z": true,
311
+ "reinit_DP": false,
312
+ "reinit_text_encoder": false
313
+ },
314
+ "lr_gen": 0.0002,
315
+ "lr_disc": 0.0002,
316
+ "lr_scheduler_gen": "ExponentialLR",
317
+ "lr_scheduler_gen_params": {
318
+ "gamma": 0.999875,
319
+ "last_epoch": -1
320
+ },
321
+ "lr_scheduler_disc": "ExponentialLR",
322
+ "lr_scheduler_disc_params": {
323
+ "gamma": 0.999875,
324
+ "last_epoch": -1
325
+ },
326
+ "kl_loss_alpha": 1.0,
327
+ "disc_loss_alpha": 1.0,
328
+ "gen_loss_alpha": 1.0,
329
+ "feat_loss_alpha": 1.0,
330
+ "mel_loss_alpha": 45.0,
331
+ "dur_loss_alpha": 1.0,
332
+ "speaker_encoder_loss_alpha": 1.0,
333
+ "return_wav": true,
334
+ "use_weighted_sampler": false,
335
+ "weighted_sampler_attrs": {},
336
+ "weighted_sampler_multipliers": {},
337
+ "r": 1,
338
+ "num_speakers": 0,
339
+ "use_speaker_embedding": true,
340
+ "speakers_file": "speakers.pth",
341
+ "speaker_embedding_channels": 256,
342
+ "language_ids_file": "language_ids.json",
343
+ "use_language_embedding": true,
344
+ "use_d_vector_file": false,
345
+ "d_vector_file": null,
346
+ "d_vector_dim": 0
347
+ }
language_ids.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "rif": 0,
3
+ "shi": 1
4
+ }
speakers.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a9df430489a8bf3eac98f38325dbdbd8d986fa731787724406062bacac5a471
3
+ size 864