it-serena-0p51m/manifest.json
40.5 KB · 1451 lines · json Raw
1 {
2 "format": "roota.raw-fp16.v1",
3 "package_name": "it_IT-flores-20260908",
4 "language": "it_IT",
5 "voice": "it_IT-serena-medium",
6 "created_from": "tools/export_roota_self_contained_package.py",
7 "sample_rate": 22050,
8 "hop_length": 256,
9 "inference": {
10 "duration_length_scale": 1.0,
11 "duration_policy": "round(exp(log_duration) * duration_length_scale)"
12 },
13 "weights_file": "weights.fp16.bin",
14 "weights_size_bytes": 1023382,
15 "weights_sha256": "265c349e124e5dd9b2ebc874b574b94bd78ba7a8b4af79e52e186d3d10e5574f",
16 "total_parameters": 511691,
17 "components": {
18 "duration": {
19 "checkpoint": "artifacts/voices/it_IT-tiny/duration/duration-student.pt",
20 "checkpoint_sha256": "6c5a4abb13744a345b1bc554f3dc3c471331afb0703c49c442813209ab067d35",
21 "config": {
22 "architecture": "duration_conv",
23 "vocab_size": 143,
24 "hidden": 32,
25 "depth": 3,
26 "kernel_size": 5,
27 "max_tokens": 1067,
28 "max_duration": 80,
29 "target_duration_scale": 1.0,
30 "pause_preserve_scale": 0.0,
31 "pause_preserve_window": 0,
32 "pause_token_ids": [
33 4,
34 8,
35 10,
36 11,
37 12,
38 13
39 ],
40 "long_preserve_threshold": 0.0,
41 "long_preserve_scale": 0.0,
42 "duration_vocab_size_forced": 0,
43 "duration_oov_id": 59
44 },
45 "parameters": 36676,
46 "tensors": [
47 {
48 "name": "blocks.0.net.0.bias",
49 "shape": [
50 32
51 ],
52 "dtype": "float16",
53 "offset_bytes": 0,
54 "nbytes": 64,
55 "sha256": "9ccca1ff60deaf06f4a0e50fba7d8dc6a4d1ddf2b40b9bbdd14e186c6f0fb51b"
56 },
57 {
58 "name": "blocks.0.net.0.weight",
59 "shape": [
60 32,
61 32,
62 5
63 ],
64 "dtype": "float16",
65 "offset_bytes": 64,
66 "nbytes": 10240,
67 "sha256": "03c6dd225f47c59d91474bcaef3743dd3d9f543fae4c7b03dfed6ef9ff6669be"
68 },
69 {
70 "name": "blocks.0.net.2.bias",
71 "shape": [
72 32
73 ],
74 "dtype": "float16",
75 "offset_bytes": 10304,
76 "nbytes": 64,
77 "sha256": "fcce689589d351d4d2f2797fb043cdd6321ca1cf1bc8ef77c12eb4d5dd9fc6c9"
78 },
79 {
80 "name": "blocks.0.net.2.weight",
81 "shape": [
82 32,
83 32,
84 5
85 ],
86 "dtype": "float16",
87 "offset_bytes": 10368,
88 "nbytes": 10240,
89 "sha256": "e6bcf58ee4fae59fa36f055a00dcd8836fb0ef36f6316365cbf09cf7d5aece0a"
90 },
91 {
92 "name": "blocks.0.scale",
93 "shape": [
94 1
95 ],
96 "dtype": "float16",
97 "offset_bytes": 20608,
98 "nbytes": 2,
99 "sha256": "0ed29ae2b9f4305d89a6f89d79fbf774b3dd476a78d7bafbb18e1a00fefc7b1e"
100 },
101 {
102 "name": "blocks.1.net.0.bias",
103 "shape": [
104 32
105 ],
106 "dtype": "float16",
107 "offset_bytes": 20610,
108 "nbytes": 64,
109 "sha256": "da1eef514f46f911fed85cac404309e9b6960e11bca36c8fffe8b2b99f8b975e"
110 },
111 {
112 "name": "blocks.1.net.0.weight",
113 "shape": [
114 32,
115 32,
116 5
117 ],
118 "dtype": "float16",
119 "offset_bytes": 20674,
120 "nbytes": 10240,
121 "sha256": "db63383a0102ebdd2256569a13e6deb4a68dd120efe02121718505ddd0b63a2d"
122 },
123 {
124 "name": "blocks.1.net.2.bias",
125 "shape": [
126 32
127 ],
128 "dtype": "float16",
129 "offset_bytes": 30914,
130 "nbytes": 64,
131 "sha256": "acb9da3d37dfea2573c368bddba21c4381118bd944072da5e379651319925915"
132 },
133 {
134 "name": "blocks.1.net.2.weight",
135 "shape": [
136 32,
137 32,
138 5
139 ],
140 "dtype": "float16",
141 "offset_bytes": 30978,
142 "nbytes": 10240,
143 "sha256": "e22f97f976d4a6db3f724a52d53041a2ef4bd1cd3eb46b70402a2fbda89840fb"
144 },
145 {
146 "name": "blocks.1.scale",
147 "shape": [
148 1
149 ],
150 "dtype": "float16",
151 "offset_bytes": 41218,
152 "nbytes": 2,
153 "sha256": "98bbad59c87d33178f5534418cadfca83c583a92844eea43c14ea434b6dfaee5"
154 },
155 {
156 "name": "blocks.2.net.0.bias",
157 "shape": [
158 32
159 ],
160 "dtype": "float16",
161 "offset_bytes": 41220,
162 "nbytes": 64,
163 "sha256": "bc8da640611e4fdf7a6973ee1c52f755bd947fcd0e15330229f46556ddc47a87"
164 },
165 {
166 "name": "blocks.2.net.0.weight",
167 "shape": [
168 32,
169 32,
170 5
171 ],
172 "dtype": "float16",
173 "offset_bytes": 41284,
174 "nbytes": 10240,
175 "sha256": "7b365df71bf05413cda2ba4358ecefb90b60bf654e0fd12ba237f51402e6e31f"
176 },
177 {
178 "name": "blocks.2.net.2.bias",
179 "shape": [
180 32
181 ],
182 "dtype": "float16",
183 "offset_bytes": 51524,
184 "nbytes": 64,
185 "sha256": "b7ca04a5177c2b726b4f92638b2249fd558bdbd36cce734278d894048bcccfda"
186 },
187 {
188 "name": "blocks.2.net.2.weight",
189 "shape": [
190 32,
191 32,
192 5
193 ],
194 "dtype": "float16",
195 "offset_bytes": 51588,
196 "nbytes": 10240,
197 "sha256": "ceed35b60723cb045bcba5203f3635ead4f1897f101c8112576d25b40e2fb888"
198 },
199 {
200 "name": "blocks.2.scale",
201 "shape": [
202 1
203 ],
204 "dtype": "float16",
205 "offset_bytes": 61828,
206 "nbytes": 2,
207 "sha256": "781fd2cecb767046b0d7b646fb2e76e3139e177562491aaea92a8da9869d5232"
208 },
209 {
210 "name": "embedding.weight",
211 "shape": [
212 143,
213 32
214 ],
215 "dtype": "float16",
216 "offset_bytes": 61830,
217 "nbytes": 9152,
218 "sha256": "004b143398e18274b69f599194fa17fd04cd03bbf5c6777b2b2e689f1bb1ce83"
219 },
220 {
221 "name": "input_proj.bias",
222 "shape": [
223 32
224 ],
225 "dtype": "float16",
226 "offset_bytes": 70982,
227 "nbytes": 64,
228 "sha256": "6b2b1d5c671092f3d49d9bae2e6009884efe7a6e165e0f0333d6941a900eb9a6"
229 },
230 {
231 "name": "input_proj.weight",
232 "shape": [
233 32,
234 35,
235 1
236 ],
237 "dtype": "float16",
238 "offset_bytes": 71046,
239 "nbytes": 2240,
240 "sha256": "8f3218118917964da5bcdcae465ee3fd981090b8016f9730a47353d853197d0c"
241 },
242 {
243 "name": "output.bias",
244 "shape": [
245 1
246 ],
247 "dtype": "float16",
248 "offset_bytes": 73286,
249 "nbytes": 2,
250 "sha256": "85549cfa10aa3b0410a4745f0e478c1d1bf7e7541117cf342e2f3a561da7cf8a"
251 },
252 {
253 "name": "output.weight",
254 "shape": [
255 1,
256 32,
257 1
258 ],
259 "dtype": "float16",
260 "offset_bytes": 73288,
261 "nbytes": 64,
262 "sha256": "dd689478d14ec2f91a0ae51e6fa08efeeb932e5e0146be77d289785c5fdfbab5"
263 }
264 ]
265 },
266 "acoustic": {
267 "checkpoint": "artifacts/voices/it_IT-tiny/joint/latent-student.pt",
268 "checkpoint_sha256": "bb1bee41ea4dc0a6b7bc584a0d4ea30ac844a1362d3282ae2ff0250e665a939f",
269 "config": {
270 "architecture": "token_context",
271 "vocab_size": 143,
272 "hidden": 48,
273 "depth": 3,
274 "kernel_size": 5,
275 "out_channels": 192,
276 "token_depth": 3
277 },
278 "parameters": 160038,
279 "tensors": [
280 {
281 "name": "embedding.weight",
282 "shape": [
283 143,
284 48
285 ],
286 "dtype": "float16",
287 "offset_bytes": 73352,
288 "nbytes": 13728,
289 "sha256": "ec363f5a8c289f6f68c32eff91c30eb0c43acc7fca67d84584cb69d1948bcb9f"
290 },
291 {
292 "name": "frame_blocks.0.net.0.bias",
293 "shape": [
294 48
295 ],
296 "dtype": "float16",
297 "offset_bytes": 87080,
298 "nbytes": 96,
299 "sha256": "7074286f80d68750dd978a50e6e0b2d59fc1327c6a871016bcae6fe740e1021d"
300 },
301 {
302 "name": "frame_blocks.0.net.0.weight",
303 "shape": [
304 48,
305 48,
306 5
307 ],
308 "dtype": "float16",
309 "offset_bytes": 87176,
310 "nbytes": 23040,
311 "sha256": "d66e6a53565838cc150705976db59cae1a4da0c097f925abb5dcffc316eaad3d"
312 },
313 {
314 "name": "frame_blocks.0.net.2.bias",
315 "shape": [
316 48
317 ],
318 "dtype": "float16",
319 "offset_bytes": 110216,
320 "nbytes": 96,
321 "sha256": "1e8a411742f0400b9ab47ddbec84681747c88386cc6015992c1c3e332a0438aa"
322 },
323 {
324 "name": "frame_blocks.0.net.2.weight",
325 "shape": [
326 48,
327 48,
328 5
329 ],
330 "dtype": "float16",
331 "offset_bytes": 110312,
332 "nbytes": 23040,
333 "sha256": "15e5eabbebd2bc7f5dbb90e948d8a098c448e7cfc41a8506dad30013311fc368"
334 },
335 {
336 "name": "frame_blocks.0.scale",
337 "shape": [
338 1
339 ],
340 "dtype": "float16",
341 "offset_bytes": 133352,
342 "nbytes": 2,
343 "sha256": "d8ae8e0e8c60472ac2b46a20b669a2e6264e48017bb67dfdf41f8e0c2d177969"
344 },
345 {
346 "name": "frame_blocks.1.net.0.bias",
347 "shape": [
348 48
349 ],
350 "dtype": "float16",
351 "offset_bytes": 133354,
352 "nbytes": 96,
353 "sha256": "99492b8c08cee722376a7dd6d7232d4e259a008f88f419bf643b41c116b84bbb"
354 },
355 {
356 "name": "frame_blocks.1.net.0.weight",
357 "shape": [
358 48,
359 48,
360 5
361 ],
362 "dtype": "float16",
363 "offset_bytes": 133450,
364 "nbytes": 23040,
365 "sha256": "501489492283f3e02fcbd0c977bc0724ae6d9fcf668d014d749e2d6e2b9a2203"
366 },
367 {
368 "name": "frame_blocks.1.net.2.bias",
369 "shape": [
370 48
371 ],
372 "dtype": "float16",
373 "offset_bytes": 156490,
374 "nbytes": 96,
375 "sha256": "a07b4eacae93e4e8f9132c6483977bed942865fec56446e67242872cad73ca3c"
376 },
377 {
378 "name": "frame_blocks.1.net.2.weight",
379 "shape": [
380 48,
381 48,
382 5
383 ],
384 "dtype": "float16",
385 "offset_bytes": 156586,
386 "nbytes": 23040,
387 "sha256": "d8a3a4ab0a6e8b59bc9c3f8d1e5f675afea5ca7069a0b5faa38d2fb630d29c0c"
388 },
389 {
390 "name": "frame_blocks.1.scale",
391 "shape": [
392 1
393 ],
394 "dtype": "float16",
395 "offset_bytes": 179626,
396 "nbytes": 2,
397 "sha256": "c351cefdbeee2b2feebc8fd2144739504a5c2687d9d6c49c72eae15c006b6a6d"
398 },
399 {
400 "name": "frame_blocks.2.net.0.bias",
401 "shape": [
402 48
403 ],
404 "dtype": "float16",
405 "offset_bytes": 179628,
406 "nbytes": 96,
407 "sha256": "4c27a4ea7243c7d43724da15982e61cb0c8534c6ae8f8bdcca5a0bf7bb83f08f"
408 },
409 {
410 "name": "frame_blocks.2.net.0.weight",
411 "shape": [
412 48,
413 48,
414 5
415 ],
416 "dtype": "float16",
417 "offset_bytes": 179724,
418 "nbytes": 23040,
419 "sha256": "37a818b22ebd8442fe275e581fdc93451cc0624415aed1accecc07e3efea3e10"
420 },
421 {
422 "name": "frame_blocks.2.net.2.bias",
423 "shape": [
424 48
425 ],
426 "dtype": "float16",
427 "offset_bytes": 202764,
428 "nbytes": 96,
429 "sha256": "fec00d35b3d0ac41f1551551ce8a642844a97bc34a185effa3de4d1de3a63b15"
430 },
431 {
432 "name": "frame_blocks.2.net.2.weight",
433 "shape": [
434 48,
435 48,
436 5
437 ],
438 "dtype": "float16",
439 "offset_bytes": 202860,
440 "nbytes": 23040,
441 "sha256": "ab6ddb6ebb1776588868988d459ed947e3ec33e2ab4ffe9a3a41a613bb3247b4"
442 },
443 {
444 "name": "frame_blocks.2.scale",
445 "shape": [
446 1
447 ],
448 "dtype": "float16",
449 "offset_bytes": 225900,
450 "nbytes": 2,
451 "sha256": "e444e5de7fc2559b5019bfc802b98a4262cb30c2efa089f05c08cc0a692c7da8"
452 },
453 {
454 "name": "frame_input_proj.bias",
455 "shape": [
456 48
457 ],
458 "dtype": "float16",
459 "offset_bytes": 225902,
460 "nbytes": 96,
461 "sha256": "8a8cf59c5575df063050c18a3517c60cad66196bb94b6c0ddf82495b3bd6845c"
462 },
463 {
464 "name": "frame_input_proj.weight",
465 "shape": [
466 48,
467 51,
468 1
469 ],
470 "dtype": "float16",
471 "offset_bytes": 225998,
472 "nbytes": 4896,
473 "sha256": "7b23adaf08fd8fcdf76a1415e45fd536e0830c5a6dab298b51c56d9b676ed4cc"
474 },
475 {
476 "name": "output.bias",
477 "shape": [
478 192
479 ],
480 "dtype": "float16",
481 "offset_bytes": 230894,
482 "nbytes": 384,
483 "sha256": "9fe41b1a655d70e6f3daf03ee18d9dca4cee3c83d2a34e6d28472928af167966"
484 },
485 {
486 "name": "output.weight",
487 "shape": [
488 192,
489 48,
490 1
491 ],
492 "dtype": "float16",
493 "offset_bytes": 231278,
494 "nbytes": 18432,
495 "sha256": "d0ac63520d36a2bf031d9ac5da86aa79eacfd69956e39960fe8271b517d3e3d9"
496 },
497 {
498 "name": "token_blocks.0.net.0.bias",
499 "shape": [
500 48
501 ],
502 "dtype": "float16",
503 "offset_bytes": 249710,
504 "nbytes": 96,
505 "sha256": "cfc1d45166c27d6c9ef37d2588e5e65f3dc5e525c588c94ac882f1811087e91b"
506 },
507 {
508 "name": "token_blocks.0.net.0.weight",
509 "shape": [
510 48,
511 48,
512 5
513 ],
514 "dtype": "float16",
515 "offset_bytes": 249806,
516 "nbytes": 23040,
517 "sha256": "ae9d0ebc609b1b366554fac04bfa8cd0dde4e3a6fe9f292b7009c5acc1003e39"
518 },
519 {
520 "name": "token_blocks.0.net.2.bias",
521 "shape": [
522 48
523 ],
524 "dtype": "float16",
525 "offset_bytes": 272846,
526 "nbytes": 96,
527 "sha256": "8a68b0d2c2a0777e962be2698068ae991daa81d2c32d6a3fcd5ced2a11b4d814"
528 },
529 {
530 "name": "token_blocks.0.net.2.weight",
531 "shape": [
532 48,
533 48,
534 5
535 ],
536 "dtype": "float16",
537 "offset_bytes": 272942,
538 "nbytes": 23040,
539 "sha256": "ee87c8b54ef63cb995fe2a417cb79f921246b7f4f53f1b5dde5666fdc63e4fed"
540 },
541 {
542 "name": "token_blocks.0.scale",
543 "shape": [
544 1
545 ],
546 "dtype": "float16",
547 "offset_bytes": 295982,
548 "nbytes": 2,
549 "sha256": "f5ca38f748a1d6eaf726b8a42fb575c3c71f1864a8143301782de13da2d9202b"
550 },
551 {
552 "name": "token_blocks.1.net.0.bias",
553 "shape": [
554 48
555 ],
556 "dtype": "float16",
557 "offset_bytes": 295984,
558 "nbytes": 96,
559 "sha256": "9c34ae331cf95dbb49ec58946c913f757bf9de8012b75d1a7d3649979bc07a3b"
560 },
561 {
562 "name": "token_blocks.1.net.0.weight",
563 "shape": [
564 48,
565 48,
566 5
567 ],
568 "dtype": "float16",
569 "offset_bytes": 296080,
570 "nbytes": 23040,
571 "sha256": "5f7cd28ce9208a3e11492dbc3167aa35ac5329f1a6d35b6f7027504b54133b66"
572 },
573 {
574 "name": "token_blocks.1.net.2.bias",
575 "shape": [
576 48
577 ],
578 "dtype": "float16",
579 "offset_bytes": 319120,
580 "nbytes": 96,
581 "sha256": "b087e35d2d00f9e80a3515b2fa4da9c69d990b58f249374bdddaf3b93fcc0f60"
582 },
583 {
584 "name": "token_blocks.1.net.2.weight",
585 "shape": [
586 48,
587 48,
588 5
589 ],
590 "dtype": "float16",
591 "offset_bytes": 319216,
592 "nbytes": 23040,
593 "sha256": "6aafbde19e9c3b92291446f10d789bc15a8dd05c35619a512d3ea879f5e8704c"
594 },
595 {
596 "name": "token_blocks.1.scale",
597 "shape": [
598 1
599 ],
600 "dtype": "float16",
601 "offset_bytes": 342256,
602 "nbytes": 2,
603 "sha256": "75191fa37fc5d2fca5ff0066d8af5455c7f2dae28d924a4261a352546c2594f9"
604 },
605 {
606 "name": "token_blocks.2.net.0.bias",
607 "shape": [
608 48
609 ],
610 "dtype": "float16",
611 "offset_bytes": 342258,
612 "nbytes": 96,
613 "sha256": "e6051a4f8b322602af2f114ca830d84d908c5ee510d0abdf9e1adf4b33ca0205"
614 },
615 {
616 "name": "token_blocks.2.net.0.weight",
617 "shape": [
618 48,
619 48,
620 5
621 ],
622 "dtype": "float16",
623 "offset_bytes": 342354,
624 "nbytes": 23040,
625 "sha256": "cd398f8ffdff939b520b3920ed86af9df24e1c145801024ea5a9b33129cd3da3"
626 },
627 {
628 "name": "token_blocks.2.net.2.bias",
629 "shape": [
630 48
631 ],
632 "dtype": "float16",
633 "offset_bytes": 365394,
634 "nbytes": 96,
635 "sha256": "fdc47eeb8b3d383d4ede36e312c6a42e9d385a64ed8bf19d7d181a6b6b9eaf15"
636 },
637 {
638 "name": "token_blocks.2.net.2.weight",
639 "shape": [
640 48,
641 48,
642 5
643 ],
644 "dtype": "float16",
645 "offset_bytes": 365490,
646 "nbytes": 23040,
647 "sha256": "18e534b1273b847076b56a9b4adefd9cf73bc681685a38e42f026e37046c7005"
648 },
649 {
650 "name": "token_blocks.2.scale",
651 "shape": [
652 1
653 ],
654 "dtype": "float16",
655 "offset_bytes": 388530,
656 "nbytes": 2,
657 "sha256": "8022823b690de83b9026495d6a2a7819b6c8cd2373e03112c06f8f17959a9e43"
658 },
659 {
660 "name": "token_input_proj.bias",
661 "shape": [
662 48
663 ],
664 "dtype": "float16",
665 "offset_bytes": 388532,
666 "nbytes": 96,
667 "sha256": "e5241b7ea99899d4c987e70691f62466331c2311c8eb8022648f015d5f00d833"
668 },
669 {
670 "name": "token_input_proj.weight",
671 "shape": [
672 48,
673 50,
674 1
675 ],
676 "dtype": "float16",
677 "offset_bytes": 388628,
678 "nbytes": 4800,
679 "sha256": "eef76ac2e1c0eef0b5b1f549067a4cce01109337b6304836c1809deb17a3d059"
680 }
681 ]
682 },
683 "decoder": {
684 "checkpoint": "artifacts/voices/it_IT-tiny/joint/decoder-student.pt",
685 "checkpoint_sha256": "a7b68517d29e19cd0caf2c152bbb3781771d0412e735a7efeaaffcf4f3ff8c5b",
686 "config": {
687 "in_channels": 192,
688 "source_in_channels": 192,
689 "channels": [
690 96,
691 48,
692 24,
693 12
694 ],
695 "res_layers": 1,
696 "variant": "piperlite",
697 "lrc_code_dim": 40,
698 "lrc_encoder_hidden": 64,
699 "rank_ratio": 0.5,
700 "activation": "leaky_relu",
701 "stage_affine": false,
702 "factorized_pre_rank": 0,
703 "piper_res_factor_rank_ratio": 0.0,
704 "res_bank_scale_mode": "kept",
705 "stage0_branches": [
706 0,
707 1,
708 2
709 ],
710 "stage1_branches": [
711 0,
712 1,
713 2
714 ],
715 "stage2_branches": [
716 0,
717 1,
718 2
719 ],
720 "stage3_branches": [
721 0,
722 1,
723 2
724 ],
725 "fsd_dim": 72,
726 "fsd_blocks": 5,
727 "fsd_film_rank": 12,
728 "fsd_head_rank": 48,
729 "wavehax_channels": 32,
730 "wavehax_blocks": 8,
731 "wavehax_mult_channels": 2,
732 "wavehax_kernel_freq": 7,
733 "wavehax_kernel_time": 7,
734 "wavehax_f0_channel": 80,
735 "wavehax_voiced_channel": 81,
736 "wavehax_sample_rate": 24000,
737 "wavehax_prior_power": 0.1,
738 "wavehax_prior_noise": 0.01,
739 "whx_channels": 32,
740 "whx_hidden": 64,
741 "whx_blocks": 8,
742 "whx_n_fft": 512,
743 "whx_f0_channel": 80,
744 "whx_voiced_channel": 81,
745 "whx_sample_rate": 24000,
746 "hift_width": 224,
747 "hift_gen_channels": 192,
748 "hift_style_dim": 64,
749 "hift_body_blocks": 3,
750 "hift_f0_channel": 80,
751 "hift_voiced_channel": 81,
752 "hift_sample_rate": 24000,
753 "stage_projection_bottlenecks": [],
754 "teacher_init_checkpoint": null,
755 "teacher_init_method": "first",
756 "teacher_init": null,
757 "init_decoder_checkpoint": "artifacts/voices/it_IT-tiny/decoder/decoder-student.pt",
758 "decoder_init": {
759 "checkpoint": "artifacts/voices/it_IT-tiny/decoder/decoder-student.pt",
760 "decoder_parameters": 314977,
761 "matched_config": {
762 "in_channels": 192,
763 "channels": [
764 96,
765 48,
766 24,
767 12
768 ],
769 "res_layers": 1,
770 "variant": "piperlite",
771 "rank_ratio": 0.5,
772 "activation": "leaky_relu",
773 "stage_affine": false,
774 "factorized_pre_rank": 0,
775 "piper_res_factor_rank_ratio": 0.0,
776 "res_bank_scale_mode": "kept",
777 "stage0_branches": [
778 0,
779 1,
780 2
781 ],
782 "stage1_branches": [
783 0,
784 1,
785 2
786 ],
787 "stage2_branches": [
788 0,
789 1,
790 2
791 ],
792 "stage3_branches": [
793 0,
794 1,
795 2
796 ],
797 "fsd_dim": 72,
798 "fsd_blocks": 5,
799 "fsd_film_rank": 12,
800 "fsd_head_rank": 48,
801 "wavehax_channels": 32,
802 "wavehax_blocks": 8,
803 "wavehax_mult_channels": 2,
804 "wavehax_kernel_freq": 7,
805 "wavehax_kernel_time": 7,
806 "wavehax_f0_channel": 80,
807 "wavehax_voiced_channel": 81,
808 "wavehax_sample_rate": 24000,
809 "wavehax_prior_power": 0.1,
810 "wavehax_prior_noise": 0.01,
811 "whx_channels": 32,
812 "whx_hidden": 64,
813 "whx_blocks": 8,
814 "whx_n_fft": 512,
815 "whx_f0_channel": 80,
816 "whx_voiced_channel": 81,
817 "whx_sample_rate": 24000,
818 "hift_width": 224,
819 "hift_gen_channels": 192,
820 "hift_style_dim": 64,
821 "hift_body_blocks": 3,
822 "hift_f0_channel": 80,
823 "hift_voiced_channel": 81,
824 "hift_sample_rate": 24000,
825 "stage_projection_bottlenecks": [],
826 "post_filter_channels": 0,
827 "post_filter_layers": 0,
828 "post_filter_kernel": 9,
829 "post_filter_scale": 0.25,
830 "pre_tanh_repair_channels": 0,
831 "pre_tanh_repair_layers": 0,
832 "pre_tanh_repair_kernel": 7,
833 "pre_tanh_repair_scale": 0.15,
834 "istft_n_fft": 512,
835 "istft_tail_head_rank": 20
836 },
837 "allowed_post_filter_mismatches": [],
838 "allowed_pre_tanh_repair_mismatches": [],
839 "allowed_activation_mismatches": [],
840 "missing_keys": [],
841 "unexpected_keys": []
842 },
843 "post_filter_channels": 0,
844 "post_filter_layers": 0,
845 "post_filter_kernel": 9,
846 "post_filter_scale": 0.25,
847 "pre_tanh_repair_channels": 0,
848 "pre_tanh_repair_layers": 0,
849 "pre_tanh_repair_kernel": 7,
850 "pre_tanh_repair_scale": 0.15,
851 "istft_n_fft": 512,
852 "istft_tail_head_rank": 20,
853 "ap_amplitude_weight": 0.0,
854 "ap_phase_weight": 0.0,
855 "ap_complex_weight": 0.0,
856 "spectral_head_init": "default",
857 "spectral_head_init_scale": 0.001,
858 "ap_amp_init_bias": -6.0,
859 "ap_phase_real_init_bias": 1.0,
860 "spectral_head_init_summary": null,
861 "feature_exact_weight": 0.0,
862 "feature_exact_keys": [
863 "pre",
864 "up0"
865 ],
866 "feature_exact_teacher_init": null,
867 "signature_pack_dir": "artifacts/voices/it_IT-tiny/signatures",
868 "exact_feature_pack_dir": null,
869 "signature_hint_weight": 0.05,
870 "signature_temporal_weight": 0.4,
871 "signature_phase_weight": 0.0,
872 "signature_phase_bins": 0,
873 "signature_keys": [
874 "stage0_mix",
875 "stage1_mix",
876 "stage2_mix",
877 "pre_tanh",
878 "audio"
879 ],
880 "bottleneck_code_checkpoint": null,
881 "bottleneck_code_weight": 0.0,
882 "bottleneck_code_keys": [],
883 "bottleneck_code_training_only_parameters": 0,
884 "echo_tail_weight": 0.0,
885 "echo_tail_min_ms": 8.0,
886 "echo_tail_max_ms": 40.0,
887 "echo_tail_lags": 9,
888 "echo_tail_margin": 0.02,
889 "paired_acoustic_residual_weight": 0.0,
890 "paired_acoustic_residual_max_scale": 0.25,
891 "lrc_pred_code_mix_prob": 0.0,
892 "lrc_pred_code_checkpoint": null,
893 "lrc_pred_code_residual_prob": 0.0,
894 "lrc_pred_code_residual_max_scale": 0.25,
895 "lrc_pred_code_cache": null
896 },
897 "parameters": 314977,
898 "tensors": [
899 {
900 "name": "post.bias",
901 "shape": [
902 1
903 ],
904 "dtype": "float16",
905 "offset_bytes": 393428,
906 "nbytes": 2,
907 "sha256": "f0ab23c008eff582adbf0c728030be1a241d8a3bb6bc66d8efc89ff417886c5b"
908 },
909 {
910 "name": "post.weight",
911 "shape": [
912 1,
913 12,
914 7
915 ],
916 "dtype": "float16",
917 "offset_bytes": 393430,
918 "nbytes": 168,
919 "sha256": "254091ae43d20532f1c717ec43162b90293652e5888dfc6c3df37f5b0f6240b2"
920 },
921 {
922 "name": "pre.bias",
923 "shape": [
924 96
925 ],
926 "dtype": "float16",
927 "offset_bytes": 393598,
928 "nbytes": 192,
929 "sha256": "1ae8d6a77df216899387cd88ebdad1a77ac6f92eca41cc114b66d6b1ae973edd"
930 },
931 {
932 "name": "pre.weight",
933 "shape": [
934 96,
935 192,
936 7
937 ],
938 "dtype": "float16",
939 "offset_bytes": 393790,
940 "nbytes": 258048,
941 "sha256": "c9a68556579410e2cabc5c1c49282a7fd9cbdd7967f1bb57f5a322faa104e04c"
942 },
943 {
944 "name": "res0.0.blocks.0.conv1.bias",
945 "shape": [
946 48
947 ],
948 "dtype": "float16",
949 "offset_bytes": 651838,
950 "nbytes": 96,
951 "sha256": "f97c28b28a5fe8fdfb89c5fd7568588bfd4ef12907fe8d2fb191f6979689d196"
952 },
953 {
954 "name": "res0.0.blocks.0.conv1.weight",
955 "shape": [
956 48,
957 48,
958 3
959 ],
960 "dtype": "float16",
961 "offset_bytes": 651934,
962 "nbytes": 13824,
963 "sha256": "441950e9b45f0ea5cc96ce1c9ed8694f4b51c0c495092819e0754971d97398f1"
964 },
965 {
966 "name": "res0.0.blocks.0.conv2.bias",
967 "shape": [
968 48
969 ],
970 "dtype": "float16",
971 "offset_bytes": 665758,
972 "nbytes": 96,
973 "sha256": "f66f32d8802f375c6534501fa6de5f2caaa41528b0129df9aa00215e0e1b2873"
974 },
975 {
976 "name": "res0.0.blocks.0.conv2.weight",
977 "shape": [
978 48,
979 48,
980 3
981 ],
982 "dtype": "float16",
983 "offset_bytes": 665854,
984 "nbytes": 13824,
985 "sha256": "e879d982175682994ebb19fc3d29891545e4d1e771ccede878e7ec920e18cb85"
986 },
987 {
988 "name": "res0.0.blocks.1.conv1.bias",
989 "shape": [
990 48
991 ],
992 "dtype": "float16",
993 "offset_bytes": 679678,
994 "nbytes": 96,
995 "sha256": "c67b923a39497dcf52dc1df1f21154b121f3f81b6fa3aff096f780a881fcad2d"
996 },
997 {
998 "name": "res0.0.blocks.1.conv1.weight",
999 "shape": [
1000 48,
1001 48,
1002 5
1003 ],
1004 "dtype": "float16",
1005 "offset_bytes": 679774,
1006 "nbytes": 23040,
1007 "sha256": "46279c384c521bcdd6aed20d76c16e530ec11b1bb5d18085e58b38ffb6ae09de"
1008 },
1009 {
1010 "name": "res0.0.blocks.1.conv2.bias",
1011 "shape": [
1012 48
1013 ],
1014 "dtype": "float16",
1015 "offset_bytes": 702814,
1016 "nbytes": 96,
1017 "sha256": "49964e64d3417b0a41628c81e8fe1b8873f565c51c634595e380e21895ef5681"
1018 },
1019 {
1020 "name": "res0.0.blocks.1.conv2.weight",
1021 "shape": [
1022 48,
1023 48,
1024 5
1025 ],
1026 "dtype": "float16",
1027 "offset_bytes": 702910,
1028 "nbytes": 23040,
1029 "sha256": "5d11b3825f95639179585e719733cb328142f3f6e58ed2d533ff94395651a1fe"
1030 },
1031 {
1032 "name": "res0.0.blocks.2.conv1.bias",
1033 "shape": [
1034 48
1035 ],
1036 "dtype": "float16",
1037 "offset_bytes": 725950,
1038 "nbytes": 96,
1039 "sha256": "6baa24baaa4b7a73c1adb26b4b0a4c5775c6264e1a71365552e16b2c6d9d2e1a"
1040 },
1041 {
1042 "name": "res0.0.blocks.2.conv1.weight",
1043 "shape": [
1044 48,
1045 48,
1046 7
1047 ],
1048 "dtype": "float16",
1049 "offset_bytes": 726046,
1050 "nbytes": 32256,
1051 "sha256": "97db3802cf251e6978fe543f6982cd767969eef6adb6e082492ea4000a4d6513"
1052 },
1053 {
1054 "name": "res0.0.blocks.2.conv2.bias",
1055 "shape": [
1056 48
1057 ],
1058 "dtype": "float16",
1059 "offset_bytes": 758302,
1060 "nbytes": 96,
1061 "sha256": "1b63cd29fc9326eefbff2803f282fe8cce0e8c45eea538bd1ba9e8f6efccf60b"
1062 },
1063 {
1064 "name": "res0.0.blocks.2.conv2.weight",
1065 "shape": [
1066 48,
1067 48,
1068 7
1069 ],
1070 "dtype": "float16",
1071 "offset_bytes": 758398,
1072 "nbytes": 32256,
1073 "sha256": "b0ea82903821c13d77bcf1e039351f7da475361b4804dd5dd0fc529cdc6dd9a9"
1074 },
1075 {
1076 "name": "res1.0.blocks.0.conv1.bias",
1077 "shape": [
1078 24
1079 ],
1080 "dtype": "float16",
1081 "offset_bytes": 790654,
1082 "nbytes": 48,
1083 "sha256": "97faba6e16ff8977f63f8cfd9c774cf335fb52a0b6979363eb40a134916b6a07"
1084 },
1085 {
1086 "name": "res1.0.blocks.0.conv1.weight",
1087 "shape": [
1088 24,
1089 24,
1090 3
1091 ],
1092 "dtype": "float16",
1093 "offset_bytes": 790702,
1094 "nbytes": 3456,
1095 "sha256": "36e366f7982d0a8c95aad008e49eea79474bc9c80ef728acda23e709c0975f58"
1096 },
1097 {
1098 "name": "res1.0.blocks.0.conv2.bias",
1099 "shape": [
1100 24
1101 ],
1102 "dtype": "float16",
1103 "offset_bytes": 794158,
1104 "nbytes": 48,
1105 "sha256": "1d145b9fb41c3a0c230912b5b8b2188102f6479185bbc60b4fd50f8d9d486742"
1106 },
1107 {
1108 "name": "res1.0.blocks.0.conv2.weight",
1109 "shape": [
1110 24,
1111 24,
1112 3
1113 ],
1114 "dtype": "float16",
1115 "offset_bytes": 794206,
1116 "nbytes": 3456,
1117 "sha256": "90efb4dff0ca0d1531f9d777dada052defe58026bad5a2ee79d35ee0a8e3bc14"
1118 },
1119 {
1120 "name": "res1.0.blocks.1.conv1.bias",
1121 "shape": [
1122 24
1123 ],
1124 "dtype": "float16",
1125 "offset_bytes": 797662,
1126 "nbytes": 48,
1127 "sha256": "26406536d46db67fe6dd2ad131c2878286b74ccfac52b5c35739666dd0ebec3a"
1128 },
1129 {
1130 "name": "res1.0.blocks.1.conv1.weight",
1131 "shape": [
1132 24,
1133 24,
1134 5
1135 ],
1136 "dtype": "float16",
1137 "offset_bytes": 797710,
1138 "nbytes": 5760,
1139 "sha256": "ee6551964bee226507141b181132afbcf65977fd5db824fada99013dda082d85"
1140 },
1141 {
1142 "name": "res1.0.blocks.1.conv2.bias",
1143 "shape": [
1144 24
1145 ],
1146 "dtype": "float16",
1147 "offset_bytes": 803470,
1148 "nbytes": 48,
1149 "sha256": "00ea80a5b8a3f688ec9588414a9c2c8fcf72836eee5dc6edc513ff76e340eeb7"
1150 },
1151 {
1152 "name": "res1.0.blocks.1.conv2.weight",
1153 "shape": [
1154 24,
1155 24,
1156 5
1157 ],
1158 "dtype": "float16",
1159 "offset_bytes": 803518,
1160 "nbytes": 5760,
1161 "sha256": "7bd42d59dc094b8b8ec873adef067aac727976ddfc0314cd89f6d02a2d7e53ba"
1162 },
1163 {
1164 "name": "res1.0.blocks.2.conv1.bias",
1165 "shape": [
1166 24
1167 ],
1168 "dtype": "float16",
1169 "offset_bytes": 809278,
1170 "nbytes": 48,
1171 "sha256": "7cf6ede16bd0fb67ea81372fca17f444bfd066c5964ba292c183558bcd262968"
1172 },
1173 {
1174 "name": "res1.0.blocks.2.conv1.weight",
1175 "shape": [
1176 24,
1177 24,
1178 7
1179 ],
1180 "dtype": "float16",
1181 "offset_bytes": 809326,
1182 "nbytes": 8064,
1183 "sha256": "e7af0131ea966c92315ea6af0bde55a5888fe52d65f6a19f2c1166e773de654d"
1184 },
1185 {
1186 "name": "res1.0.blocks.2.conv2.bias",
1187 "shape": [
1188 24
1189 ],
1190 "dtype": "float16",
1191 "offset_bytes": 817390,
1192 "nbytes": 48,
1193 "sha256": "0b7c54eba593f39e22e369733b732b5fa3e76268ac4662a55af5bb1c04e0086c"
1194 },
1195 {
1196 "name": "res1.0.blocks.2.conv2.weight",
1197 "shape": [
1198 24,
1199 24,
1200 7
1201 ],
1202 "dtype": "float16",
1203 "offset_bytes": 817438,
1204 "nbytes": 8064,
1205 "sha256": "3d0b0f12ccb97329bb08d37071ef253cbe73817debb877d08c403dac4d86db8b"
1206 },
1207 {
1208 "name": "res2.0.blocks.0.conv1.bias",
1209 "shape": [
1210 12
1211 ],
1212 "dtype": "float16",
1213 "offset_bytes": 825502,
1214 "nbytes": 24,
1215 "sha256": "ad2fc18fce307dcd625a25fdc364b958946188c7c79ab5c2e7fc6e92baff5470"
1216 },
1217 {
1218 "name": "res2.0.blocks.0.conv1.weight",
1219 "shape": [
1220 12,
1221 12,
1222 3
1223 ],
1224 "dtype": "float16",
1225 "offset_bytes": 825526,
1226 "nbytes": 864,
1227 "sha256": "85ec719ce0ec8ed567eedbae0da9b54d4277d142f27cc19358afb0f6874ecae6"
1228 },
1229 {
1230 "name": "res2.0.blocks.0.conv2.bias",
1231 "shape": [
1232 12
1233 ],
1234 "dtype": "float16",
1235 "offset_bytes": 826390,
1236 "nbytes": 24,
1237 "sha256": "3a54627566b583980893f646fc98133bda2a739b3611ba5a3c70d6906357d22f"
1238 },
1239 {
1240 "name": "res2.0.blocks.0.conv2.weight",
1241 "shape": [
1242 12,
1243 12,
1244 3
1245 ],
1246 "dtype": "float16",
1247 "offset_bytes": 826414,
1248 "nbytes": 864,
1249 "sha256": "05b182814bb61bbedc30cee46ad36047921ddf429faec05b667c34ef64efd0b4"
1250 },
1251 {
1252 "name": "res2.0.blocks.1.conv1.bias",
1253 "shape": [
1254 12
1255 ],
1256 "dtype": "float16",
1257 "offset_bytes": 827278,
1258 "nbytes": 24,
1259 "sha256": "84cd6af602983c2928aafe9be3b48fb04eb5240b91effcb183ada2b84f665d7b"
1260 },
1261 {
1262 "name": "res2.0.blocks.1.conv1.weight",
1263 "shape": [
1264 12,
1265 12,
1266 5
1267 ],
1268 "dtype": "float16",
1269 "offset_bytes": 827302,
1270 "nbytes": 1440,
1271 "sha256": "b731a1e4aa34257f5ec35e7ef20ea6906ba88cf91209d41eb601965979efd138"
1272 },
1273 {
1274 "name": "res2.0.blocks.1.conv2.bias",
1275 "shape": [
1276 12
1277 ],
1278 "dtype": "float16",
1279 "offset_bytes": 828742,
1280 "nbytes": 24,
1281 "sha256": "af3e20ea2db31d837449a2b9d5312cca97271e5b3450644ed98ff244e06c6017"
1282 },
1283 {
1284 "name": "res2.0.blocks.1.conv2.weight",
1285 "shape": [
1286 12,
1287 12,
1288 5
1289 ],
1290 "dtype": "float16",
1291 "offset_bytes": 828766,
1292 "nbytes": 1440,
1293 "sha256": "05a19a1c771a580799eccc75f13a3faf5157337bd393e12fd0960dce9c01959d"
1294 },
1295 {
1296 "name": "res2.0.blocks.2.conv1.bias",
1297 "shape": [
1298 12
1299 ],
1300 "dtype": "float16",
1301 "offset_bytes": 830206,
1302 "nbytes": 24,
1303 "sha256": "293d7666a44610f3ad44a6849df6055bf5f5ce6025d0c756aa5d3b0a17c0cbb5"
1304 },
1305 {
1306 "name": "res2.0.blocks.2.conv1.weight",
1307 "shape": [
1308 12,
1309 12,
1310 7
1311 ],
1312 "dtype": "float16",
1313 "offset_bytes": 830230,
1314 "nbytes": 2016,
1315 "sha256": "ef05e7cf10dff5f69b8a376dd3ebef5d5194102a803ea6ccd33e860e45388946"
1316 },
1317 {
1318 "name": "res2.0.blocks.2.conv2.bias",
1319 "shape": [
1320 12
1321 ],
1322 "dtype": "float16",
1323 "offset_bytes": 832246,
1324 "nbytes": 24,
1325 "sha256": "8586c0ac4ccbac1f0008bcf79df24b6c166501b22ed0f11afe6fc407f98f90bf"
1326 },
1327 {
1328 "name": "res2.0.blocks.2.conv2.weight",
1329 "shape": [
1330 12,
1331 12,
1332 7
1333 ],
1334 "dtype": "float16",
1335 "offset_bytes": 832270,
1336 "nbytes": 2016,
1337 "sha256": "d2cf2b1f3bb3eb97dcdebd6a5cda1dd579f6620dd6fd64bfdd82a166a6a9c7dc"
1338 },
1339 {
1340 "name": "up0.bias",
1341 "shape": [
1342 48
1343 ],
1344 "dtype": "float16",
1345 "offset_bytes": 834286,
1346 "nbytes": 96,
1347 "sha256": "0203d4d1e1ccdd5a324b1e1351b589d9ade1b92670adc9c3413a186784b4ad42"
1348 },
1349 {
1350 "name": "up0.weight",
1351 "shape": [
1352 96,
1353 48,
1354 16
1355 ],
1356 "dtype": "float16",
1357 "offset_bytes": 834382,
1358 "nbytes": 147456,
1359 "sha256": "088bf539c476674794aa97ed8681168eacc66a9f59cebc6612fc56dd22905e6e"
1360 },
1361 {
1362 "name": "up1.bias",
1363 "shape": [
1364 24
1365 ],
1366 "dtype": "float16",
1367 "offset_bytes": 981838,
1368 "nbytes": 48,
1369 "sha256": "3ca833f582cc17f7ba696f09b25138292aff3e7e6778bf8e7fd6066f091679a6"
1370 },
1371 {
1372 "name": "up1.weight",
1373 "shape": [
1374 48,
1375 24,
1376 16
1377 ],
1378 "dtype": "float16",
1379 "offset_bytes": 981886,
1380 "nbytes": 36864,
1381 "sha256": "6bce2ecb793def0756a8e218677ba185e0f3fabd80cce00a6cf321ef25ecba73"
1382 },
1383 {
1384 "name": "up2.bias",
1385 "shape": [
1386 12
1387 ],
1388 "dtype": "float16",
1389 "offset_bytes": 1018750,
1390 "nbytes": 24,
1391 "sha256": "58df0e2f1680dc96ec24dc0357ee609eda47960683d08291649b9b5d82369d6c"
1392 },
1393 {
1394 "name": "up2.weight",
1395 "shape": [
1396 24,
1397 12,
1398 8
1399 ],
1400 "dtype": "float16",
1401 "offset_bytes": 1018774,
1402 "nbytes": 4608,
1403 "sha256": "b37c097e632f7ca208c56876344381d348a742c597436834fae64d66e1e78d9d"
1404 }
1405 ]
1406 }
1407 },
1408 "runtime_ops": [
1409 "embedding_lookup",
1410 "conv1d_same",
1411 "conv1d_1x1",
1412 "conv_transpose1d",
1413 "silu",
1414 "leaky_relu",
1415 "tanh",
1416 "residual_add",
1417 "repeat_interleave",
1418 "masked_conv1d",
1419 "depthwise_conv1d",
1420 "low_rank_linear_or_conv1d"
1421 ],
1422 "runtime_boundary": {
1423 "neural_stack_self_contained": true,
1424 "arbitrary_text_self_contained": false,
1425 "accepted_runtime_input": "Piper phoneme ID sequence matching piper-phoneme-config.json",
1426 "output": "mono float waveform at manifest.sample_rate"
1427 },
1428 "frontend": {
1429 "status": "not fully self-contained for arbitrary text yet",
1430 "input_contract": "runtime may accept Piper phoneme ID sequences directly",
1431 "included_config": "piper-phoneme-config.json",
1432 "piper_phoneme_type": "espeak",
1433 "piper_espeak_voice": "it",
1434 "phoneme_id_map_entries": 157,
1435 "next_work": "embed or rewrite the Piper/eSpeak-compatible frontend, or keep a strict phoneme-ID API"
1436 },
1437 "size_budget": {
1438 "target_max_bytes": 4000000,
1439 "target_max_mib": 3.814697265625,
1440 "actual_model_asset_bytes": 1028341,
1441 "target_max_parameters": 1500000,
1442 "under_parameter_budget": true,
1443 "under_4mb_asset_budget": true
1444 },
1445 "limitations": [
1446 "This is a raw neural asset package, not a compiled standalone runtime binary.",
1447 "No PyTorch checkpoint, ONNX graph, or training state is included.",
1448 "Arbitrary text still requires a Piper/eSpeak-compatible frontend outside this package."
1449 ]
1450 }
1451