addy-hypr4 commited on
Commit
e3b0894
·
1 Parent(s): 26f766f

added new primary tts model - supertonic 3 ,kokoro and piper are deprecated

Browse files
.gitattributes CHANGED
@@ -50,3 +50,4 @@ tts/kokoro/*.tar.gz filter=lfs diff=lfs merge=lfs -text
50
  stt filter=lfs diff=lfs merge=lfs -text
51
  translit filter=lfs diff=lfs merge=lfs -text
52
  stt/nvidia-nemotron-3.5 filter=lfs diff=lfs merge=lfs -text
 
 
50
  stt filter=lfs diff=lfs merge=lfs -text
51
  translit filter=lfs diff=lfs merge=lfs -text
52
  stt/nvidia-nemotron-3.5 filter=lfs diff=lfs merge=lfs -text
53
+ tts/supertonic-3 filter=lfs diff=lfs merge=lfs -text
models_manifest.json CHANGED
@@ -1,11 +1,10 @@
1
  {
2
- "models_version": "1.0.2",
3
  "release_notes": [
4
- "Initial stable release of Vox speech and reasoning weights.",
5
- "Integrated Qwen3-ASR model and Kokoro TTS English synthesis.",
6
- "Integrated Nvidia Nemotron-3.5 ASR Engine (INT8 quantized) as the default primary speech recognition model."
7
  ],
8
- "total_size_bytes": 9811968972,
9
  "model_groups": [
10
  {
11
  "id": "ten_vad",
@@ -190,96 +189,58 @@
190
  ]
191
  },
192
  {
193
- "id": "kokoro_english_tts",
194
- "name": "Kokoro English TTS Engine",
195
  "category": "tts",
196
- "version": "1.0.0",
197
- "files": [
198
- {
199
- "id": "tts_kokoro_onnx",
200
- "path": "tts/kokoro/model.onnx",
201
- "size": 345555491,
202
- "sha256": "10ff414106a038ce7e9e0126c6461e4dc8a86efaa89dc91d2009d69fe635e339",
203
- "required": false
204
- },
205
- {
206
- "id": "tts_kokoro_voices",
207
- "path": "tts/kokoro/voices.bin",
208
- "size": 5755904,
209
- "sha256": "a372c67b056ef0b695c375d39b99630d23fb07ad4c8d87aa32a19a62fca523ad",
210
- "required": false
211
- },
212
- {
213
- "id": "tts_kokoro_tokens",
214
- "path": "tts/kokoro/tokens.txt",
215
- "size": 1078,
216
- "sha256": "4f31c71282d14af4e926cd12462078fe9d20d00c589e63fe2750a8f56d6d7f7b",
217
- "required": false
218
- },
219
- {
220
- "id": "tts_kokoro_espeak_ng_data",
221
- "path": "tts/kokoro/espeak-ng-data.tar.gz",
222
- "size": 8990857,
223
- "sha256": "7ddfb7247e98108baeab2a3f9c79c29247e58ae7d60d837b32e28999f555bf8d",
224
- "archive": "tar.gz",
225
- "required": false
226
- }
227
- ]
228
- },
229
- {
230
- "id": "piper_hindi_tts",
231
- "name": "Piper Hindi TTS Engine",
232
- "category": "tts",
233
- "version": "1.0.0",
234
  "files": [
235
  {
236
- "id": "tts_hi_pratham_onnx",
237
- "path": "tts/piper_hi/hi_IN-pratham-medium.onnx",
238
- "size": 63516050,
239
- "sha256": "169964b0871667f6793416d4b35e97357a68ba1ad01df8580c28048989ee7693",
240
  "required": false
241
  },
242
  {
243
- "id": "tts_hi_pratham_config",
244
- "path": "tts/piper_hi/hi_IN-pratham-medium.onnx.json",
245
- "size": 4970,
246
- "sha256": "b68edd2cd7950dd436314013b7cd12e9699e5a3f6fe5af5af94294cf6aa7b9fd",
247
  "required": false
248
  },
249
  {
250
- "id": "tts_hi_priyamvada_onnx",
251
- "path": "tts/piper_hi/hi_IN-priyamvada-medium.onnx",
252
- "size": 63145178,
253
- "sha256": "8871f3e07adb6ca490f8dbcd3956a8647c53c35b5d0a1c2a8d097b3bf721a31b",
254
  "required": false
255
  },
256
  {
257
- "id": "tts_hi_priyamvada_config",
258
- "path": "tts/piper_hi/hi_IN-priyamvada-medium.onnx.json",
259
- "size": 4973,
260
- "sha256": "5efc0ccf7529f3528996d46e0fac1f969f681d44a8e55bfa6236ff8841b5d52d",
261
  "required": false
262
  },
263
  {
264
- "id": "tts_hi_rohan_onnx",
265
- "path": "tts/piper_hi/hi_IN-rohan-medium.onnx",
266
- "size": 62950044,
267
- "sha256": "b65dc80fb34d9dcd1cf684cb297966a34983bbc93bb1696fe207f32b0b33a091",
268
  "required": false
269
  },
270
  {
271
- "id": "tts_hi_rohan_config",
272
- "path": "tts/piper_hi/hi_IN-rohan-medium.onnx.json",
273
- "size": 5041,
274
- "sha256": "07b9ae19bd0bac7fbbc99f7ee69c91245eb5470e926632c31fc0c50ba653c817",
275
  "required": false
276
  },
277
  {
278
- "id": "tts_piper_hi_espeak_ng_data",
279
- "path": "tts/piper_hi/espeak-ng-data.tar.gz",
280
- "size": 8990869,
281
- "sha256": "5d26294483bff947820b8ebf55fde437622c7c542eb104649e29626ab1763524",
282
- "archive": "tar.gz",
283
  "required": false
284
  }
285
  ]
 
1
  {
2
+ "models_version": "1.1.0",
3
  "release_notes": [
4
+ "Portable, single-instance TTS engine with 31 languages and flow-matching architecture.",
5
+ "Supertonic 3 replaces Kokoro (EN) + Piper (HI) as the sole TTS engine."
 
6
  ],
7
+ "total_size_bytes": 9398491074,
8
  "model_groups": [
9
  {
10
  "id": "ten_vad",
 
189
  ]
190
  },
191
  {
192
+ "id": "supertonic_tts",
193
+ "name": "Supertonic 3 TTS Engine",
194
  "category": "tts",
195
+ "version": "3.0.0",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
196
  "files": [
197
  {
198
+ "id": "tts_supertonic_duration_predictor",
199
+ "path": "tts/supertonic-3/duration_predictor.int8.onnx",
200
+ "size": 3700147,
201
+ "sha256": "c3eb91414d5ff8a7a239b7fe9e34e7e2bf8a8140d8375ffb14718b1c639325db",
202
  "required": false
203
  },
204
  {
205
+ "id": "tts_supertonic_text_encoder",
206
+ "path": "tts/supertonic-3/text_encoder.int8.onnx",
207
+ "size": 36416150,
208
+ "sha256": "c7befd5ea8c3119769e8a6c1486c4edc6a3bc8365c67621c881bbb774b9902ff",
209
  "required": false
210
  },
211
  {
212
+ "id": "tts_supertonic_vector_estimator",
213
+ "path": "tts/supertonic-3/vector_estimator.int8.onnx",
214
+ "size": 78400833,
215
+ "sha256": "20cd86fa5c6effedfda0e7cffe5b0569ca401c440a0c3a1d72bf39286c0db3fd",
216
  "required": false
217
  },
218
  {
219
+ "id": "tts_supertonic_vocoder",
220
+ "path": "tts/supertonic-3/vocoder.int8.onnx",
221
+ "size": 25991073,
222
+ "sha256": "e923d60f53f95eb1ce235f1dc33ec56d9c057823c96fa6f8acf98f32b0da6152",
223
  "required": false
224
  },
225
  {
226
+ "id": "tts_supertonic_config",
227
+ "path": "tts/supertonic-3/tts.json",
228
+ "size": 8253,
229
+ "sha256": "42078d3aef1cd43ab43021f3c54f47d2d75ceb4e75f627f118890128b06a0d09",
230
  "required": false
231
  },
232
  {
233
+ "id": "tts_supertonic_indexer",
234
+ "path": "tts/supertonic-3/unicode_indexer.bin",
235
+ "size": 262144,
236
+ "sha256": "8402ca48e5189a8950138580b0fff64db6f072f24ac07cd54ba8b2fbb9883b30",
237
  "required": false
238
  },
239
  {
240
+ "id": "tts_supertonic_voice_bin",
241
+ "path": "tts/supertonic-3/voice.bin",
242
+ "size": 517168,
243
+ "sha256": "67d5209b0ee8ce6c74105ffbe12fe6a7628aea3b4ba2fcb308a4a67938a93ce8",
 
244
  "required": false
245
  }
246
  ]
tts/supertonic-3/duration_predictor.int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3eb91414d5ff8a7a239b7fe9e34e7e2bf8a8140d8375ffb14718b1c639325db
3
+ size 3700147
tts/supertonic-3/text_encoder.int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c7befd5ea8c3119769e8a6c1486c4edc6a3bc8365c67621c881bbb774b9902ff
3
+ size 36416150
tts/supertonic-3/tts.json ADDED
@@ -0,0 +1,311 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "tts_version": "v1.7.3",
3
+ "split": "opensource-multilingual",
4
+ "ttl": {
5
+ "latent_dim": 24,
6
+ "chunk_compress_factor": 6,
7
+ "batch_expander": {
8
+ "n_batch_expand": 6
9
+ },
10
+ "normalizer": {
11
+ "scale": 0.25
12
+ },
13
+ "text_encoder": {
14
+ "n_langs": 0,
15
+ "lang_emb_dim": 0,
16
+ "text_embedder": {
17
+ "char_emb_dim": 256
18
+ },
19
+ "convnext": {
20
+ "idim": 256,
21
+ "ksz": 5,
22
+ "intermediate_dim": 1024,
23
+ "num_layers": 6,
24
+ "dilation_lst": [
25
+ 1,
26
+ 1,
27
+ 2,
28
+ 2,
29
+ 4,
30
+ 4
31
+ ]
32
+ },
33
+ "attn_encoder": {
34
+ "hidden_channels": 256,
35
+ "filter_channels": 1024,
36
+ "n_heads": 4,
37
+ "n_layers": 4,
38
+ "p_dropout": 0.0
39
+ },
40
+ "proj_out": {
41
+ "idim": 256,
42
+ "odim": 256
43
+ }
44
+ },
45
+ "flow_matching": {
46
+ "sig_min": 1e-08
47
+ },
48
+ "style_encoder": {
49
+ "proj_in": {
50
+ "ldim": 24,
51
+ "chunk_compress_factor": 6,
52
+ "odim": 256
53
+ },
54
+ "convnext": {
55
+ "idim": 256,
56
+ "ksz": 5,
57
+ "intermediate_dim": 1024,
58
+ "num_layers": 6,
59
+ "dilation_lst": [
60
+ 1,
61
+ 1,
62
+ 1,
63
+ 1,
64
+ 1,
65
+ 1
66
+ ]
67
+ },
68
+ "style_token_layer": {
69
+ "input_dim": 256,
70
+ "n_style": 50,
71
+ "style_key_dim": 256,
72
+ "style_value_dim": 256,
73
+ "prototype_dim": 256,
74
+ "n_units": 256,
75
+ "n_heads": 2
76
+ }
77
+ },
78
+ "speech_prompted_text_encoder": {
79
+ "text_dim": 256,
80
+ "style_dim": 256,
81
+ "n_units": 256,
82
+ "n_heads": 2
83
+ },
84
+ "uncond_masker": {
85
+ "prob_both_uncond": 0.04,
86
+ "prob_text_uncond": 0.01,
87
+ "std": 0.1,
88
+ "text_dim": 256,
89
+ "n_style": 50,
90
+ "style_key_dim": 256,
91
+ "style_value_dim": 256
92
+ },
93
+ "vector_field": {
94
+ "n_langs": 0,
95
+ "lang_emb_dim": 0,
96
+ "proj_in": {
97
+ "ldim": 24,
98
+ "chunk_compress_factor": 6,
99
+ "odim": 512
100
+ },
101
+ "time_encoder": {
102
+ "time_dim": 64,
103
+ "hdim": 256
104
+ },
105
+ "main_blocks": {
106
+ "n_blocks": 4,
107
+ "time_cond_layer": {
108
+ "idim": 512,
109
+ "time_dim": 64
110
+ },
111
+ "style_cond_layer": {
112
+ "idim": 512,
113
+ "style_dim": 256
114
+ },
115
+ "text_cond_layer": {
116
+ "idim": 512,
117
+ "text_dim": 256,
118
+ "n_heads": 8,
119
+ "n_units": 512,
120
+ "use_residual": true,
121
+ "rotary_base": 10000,
122
+ "rotary_scale": 10
123
+ },
124
+ "convnext_0": {
125
+ "idim": 512,
126
+ "ksz": 5,
127
+ "intermediate_dim": 2048,
128
+ "num_layers": 4,
129
+ "dilation_lst": [
130
+ 1,
131
+ 2,
132
+ 4,
133
+ 8
134
+ ]
135
+ },
136
+ "convnext_1": {
137
+ "idim": 512,
138
+ "ksz": 5,
139
+ "intermediate_dim": 2048,
140
+ "num_layers": 1,
141
+ "dilation_lst": [
142
+ 1
143
+ ]
144
+ },
145
+ "convnext_2": {
146
+ "idim": 512,
147
+ "ksz": 5,
148
+ "intermediate_dim": 2048,
149
+ "num_layers": 1,
150
+ "dilation_lst": [
151
+ 1
152
+ ]
153
+ }
154
+ },
155
+ "last_convnext": {
156
+ "idim": 512,
157
+ "ksz": 5,
158
+ "intermediate_dim": 2048,
159
+ "num_layers": 4,
160
+ "dilation_lst": [
161
+ 1,
162
+ 1,
163
+ 1,
164
+ 1
165
+ ]
166
+ },
167
+ "proj_out": {
168
+ "idim": 512,
169
+ "chunk_compress_factor": 6,
170
+ "ldim": 24
171
+ }
172
+ }
173
+ },
174
+ "ae": {
175
+ "sample_rate": 44100,
176
+ "n_delay": 0,
177
+ "base_chunk_size": 512,
178
+ "chunk_compress_factor": 1,
179
+ "ldim": 24,
180
+ "encoder": {
181
+ "spec_processor": {
182
+ "n_fft": 2048,
183
+ "win_length": 2048,
184
+ "hop_length": 512,
185
+ "n_mels": 228,
186
+ "sample_rate": 44100,
187
+ "eps": 1e-05,
188
+ "norm_mean": 0.0,
189
+ "norm_std": 1.0
190
+ },
191
+ "ksz_init": 7,
192
+ "ksz": 7,
193
+ "num_layers": 10,
194
+ "dilation_lst": [
195
+ 1,
196
+ 1,
197
+ 1,
198
+ 1,
199
+ 1,
200
+ 1,
201
+ 1,
202
+ 1,
203
+ 1,
204
+ 1
205
+ ],
206
+ "intermediate_dim": 2048,
207
+ "idim": 1253,
208
+ "hdim": 512,
209
+ "odim": 24
210
+ },
211
+ "decoder": {
212
+ "ksz_init": 7,
213
+ "ksz": 7,
214
+ "num_layers": 10,
215
+ "dilation_lst": [
216
+ 1,
217
+ 2,
218
+ 4,
219
+ 1,
220
+ 2,
221
+ 4,
222
+ 1,
223
+ 1,
224
+ 1,
225
+ 1
226
+ ],
227
+ "intermediate_dim": 2048,
228
+ "idim": 24,
229
+ "hdim": 512,
230
+ "head": {
231
+ "idim": 512,
232
+ "hdim": 2048,
233
+ "odim": 512,
234
+ "ksz": 3
235
+ }
236
+ }
237
+ },
238
+ "dp": {
239
+ "latent_dim": 24,
240
+ "chunk_compress_factor": 6,
241
+ "normalizer": {
242
+ "scale": 1.0
243
+ },
244
+ "sentence_encoder": {
245
+ "char_emb_dim": 64,
246
+ "text_embedder": {
247
+ "char_emb_dim": 64
248
+ },
249
+ "convnext": {
250
+ "idim": 64,
251
+ "ksz": 5,
252
+ "intermediate_dim": 256,
253
+ "num_layers": 6,
254
+ "dilation_lst": [
255
+ 1,
256
+ 1,
257
+ 1,
258
+ 1,
259
+ 1,
260
+ 1
261
+ ]
262
+ },
263
+ "attn_encoder": {
264
+ "hidden_channels": 64,
265
+ "filter_channels": 256,
266
+ "n_heads": 2,
267
+ "n_layers": 2,
268
+ "p_dropout": 0.0
269
+ },
270
+ "proj_out": {
271
+ "idim": 64,
272
+ "odim": 64
273
+ }
274
+ },
275
+ "style_encoder": {
276
+ "proj_in": {
277
+ "ldim": 24,
278
+ "chunk_compress_factor": 6,
279
+ "odim": 64
280
+ },
281
+ "convnext": {
282
+ "idim": 64,
283
+ "ksz": 5,
284
+ "intermediate_dim": 256,
285
+ "num_layers": 4,
286
+ "dilation_lst": [
287
+ 1,
288
+ 1,
289
+ 1,
290
+ 1
291
+ ]
292
+ },
293
+ "style_token_layer": {
294
+ "input_dim": 64,
295
+ "n_style": 8,
296
+ "style_key_dim": 0,
297
+ "style_value_dim": 16,
298
+ "prototype_dim": 64,
299
+ "n_units": 64,
300
+ "n_heads": 2
301
+ }
302
+ },
303
+ "predictor": {
304
+ "sentence_dim": 64,
305
+ "n_style": 8,
306
+ "style_dim": 16,
307
+ "hdim": 128,
308
+ "n_layer": 2
309
+ }
310
+ }
311
+ }
tts/supertonic-3/unicode_indexer.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8402ca48e5189a8950138580b0fff64db6f072f24ac07cd54ba8b2fbb9883b30
3
+ size 262144
tts/supertonic-3/vector_estimator.int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:20cd86fa5c6effedfda0e7cffe5b0569ca401c440a0c3a1d72bf39286c0db3fd
3
+ size 78400833
tts/supertonic-3/vocoder.int8.onnx ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e923d60f53f95eb1ce235f1dc33ec56d9c057823c96fa6f8acf98f32b0da6152
3
+ size 25991073
tts/supertonic-3/voice.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67d5209b0ee8ce6c74105ffbe12fe6a7628aea3b4ba2fcb308a4a67938a93ce8
3
+ size 517168