hfc-en-1p8m/manifest.json
40.7 KB · 1453 lines · json Raw
1 {
2 "format": "roota.raw-fp16.v1",
3 "package_name": "hfc-en-1p8m",
4 "language": "en_US",
5 "voice": "en_US-hfc_female-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": 1.08,
11 "duration_policy": "round(exp(log_duration) * duration_length_scale)"
12 },
13 "weights_file": "weights.fp16.bin",
14 "weights_size_bytes": 3668760,
15 "weights_sha256": "ad1c06f41d9149fc9b0f689e36261dd3b89f96dd9e9433aa1b7dbe50dca49d90",
16 "total_parameters": 1834380,
17 "components": {
18 "duration": {
19 "checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-hfc_female-medium-retrain-20260712/a5-duration/duration-student.pt",
20 "checkpoint_sha256": "82061a1480a3f1b65769ce882ab801caeea18f1c4eeccb66971bde30a58b7a8c",
21 "config": {
22 "architecture": "duration_conv",
23 "vocab_size": 129,
24 "hidden": 64,
25 "depth": 3,
26 "kernel_size": 5,
27 "max_tokens": 265,
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 },
43 "parameters": 135940,
44 "tensors": [
45 {
46 "name": "blocks.0.net.0.bias",
47 "shape": [
48 64
49 ],
50 "dtype": "float16",
51 "offset_bytes": 0,
52 "nbytes": 128,
53 "sha256": "156ae4f947f7c600e2f4d35202ec9b4b9aa164e6cff621b93699a521cde7bb40"
54 },
55 {
56 "name": "blocks.0.net.0.weight",
57 "shape": [
58 64,
59 64,
60 5
61 ],
62 "dtype": "float16",
63 "offset_bytes": 128,
64 "nbytes": 40960,
65 "sha256": "3fce779d04be7afe860dc3dd75f50ee74e262abf902a0a292cd4c6a5cc43ef97"
66 },
67 {
68 "name": "blocks.0.net.2.bias",
69 "shape": [
70 64
71 ],
72 "dtype": "float16",
73 "offset_bytes": 41088,
74 "nbytes": 128,
75 "sha256": "55b50cf90d83c61ce37eb5244bc56ec1fc794e650e21077d4d837864b10d9616"
76 },
77 {
78 "name": "blocks.0.net.2.weight",
79 "shape": [
80 64,
81 64,
82 5
83 ],
84 "dtype": "float16",
85 "offset_bytes": 41216,
86 "nbytes": 40960,
87 "sha256": "4bb9a3af1a89b951a874ae4f1891e21d282a3ccf6c717978d11658eed0bb11dc"
88 },
89 {
90 "name": "blocks.0.scale",
91 "shape": [
92 1
93 ],
94 "dtype": "float16",
95 "offset_bytes": 82176,
96 "nbytes": 2,
97 "sha256": "17d4de60d262b95b0429aafc56a1e1a8fedc0d006f0712192ae8e5c60a13671c"
98 },
99 {
100 "name": "blocks.1.net.0.bias",
101 "shape": [
102 64
103 ],
104 "dtype": "float16",
105 "offset_bytes": 82178,
106 "nbytes": 128,
107 "sha256": "333a85fecb7ba9f1ee6f7c4fbf0ee8ca29be2415e428935a64546c5a6f4527e7"
108 },
109 {
110 "name": "blocks.1.net.0.weight",
111 "shape": [
112 64,
113 64,
114 5
115 ],
116 "dtype": "float16",
117 "offset_bytes": 82306,
118 "nbytes": 40960,
119 "sha256": "0dd8a1ce2976c83fb0dddbee91326bdb42409e20b830d90c32169699e042c34c"
120 },
121 {
122 "name": "blocks.1.net.2.bias",
123 "shape": [
124 64
125 ],
126 "dtype": "float16",
127 "offset_bytes": 123266,
128 "nbytes": 128,
129 "sha256": "a8c76a90b7e34d67900211583988c64b7db85dc407927301b1a9f352ca50fc23"
130 },
131 {
132 "name": "blocks.1.net.2.weight",
133 "shape": [
134 64,
135 64,
136 5
137 ],
138 "dtype": "float16",
139 "offset_bytes": 123394,
140 "nbytes": 40960,
141 "sha256": "6ed2aa5c92956699ae966b6e8cb270c8636f71901614ffa56622c0c0a4211cd4"
142 },
143 {
144 "name": "blocks.1.scale",
145 "shape": [
146 1
147 ],
148 "dtype": "float16",
149 "offset_bytes": 164354,
150 "nbytes": 2,
151 "sha256": "2396a1256ac4b1c6849c931ddb8018bdd984bb2383be21bb819a33b95d8d603f"
152 },
153 {
154 "name": "blocks.2.net.0.bias",
155 "shape": [
156 64
157 ],
158 "dtype": "float16",
159 "offset_bytes": 164356,
160 "nbytes": 128,
161 "sha256": "50bc15ebbf78fe75cd6ddcd8749e4e9c03b547d46bd703d98029d3a80d25b028"
162 },
163 {
164 "name": "blocks.2.net.0.weight",
165 "shape": [
166 64,
167 64,
168 5
169 ],
170 "dtype": "float16",
171 "offset_bytes": 164484,
172 "nbytes": 40960,
173 "sha256": "4c264c039eefac3bc066cc255286a4c85327d12712100f07819289deaa2c8428"
174 },
175 {
176 "name": "blocks.2.net.2.bias",
177 "shape": [
178 64
179 ],
180 "dtype": "float16",
181 "offset_bytes": 205444,
182 "nbytes": 128,
183 "sha256": "957c0503cf9401cf65e3dbc24be5c7d119c1c5dd83f8328f5796ed8e54de50da"
184 },
185 {
186 "name": "blocks.2.net.2.weight",
187 "shape": [
188 64,
189 64,
190 5
191 ],
192 "dtype": "float16",
193 "offset_bytes": 205572,
194 "nbytes": 40960,
195 "sha256": "fe3b4b83da3675cb2050cab61de94a22e46a07411450883f11f0e4849b8472ef"
196 },
197 {
198 "name": "blocks.2.scale",
199 "shape": [
200 1
201 ],
202 "dtype": "float16",
203 "offset_bytes": 246532,
204 "nbytes": 2,
205 "sha256": "304c9bf8d28eb0d8c6e40bebae8ee1e04890d94c03c76fe18d08a25846f130ce"
206 },
207 {
208 "name": "embedding.weight",
209 "shape": [
210 129,
211 64
212 ],
213 "dtype": "float16",
214 "offset_bytes": 246534,
215 "nbytes": 16512,
216 "sha256": "d0bab86e3c87fbd92336f8231bc3c827f7da5ebee06f53495ed6ba14c858f7ad"
217 },
218 {
219 "name": "input_proj.bias",
220 "shape": [
221 64
222 ],
223 "dtype": "float16",
224 "offset_bytes": 263046,
225 "nbytes": 128,
226 "sha256": "3fe7102b45291f055c40886887866fa47785ac5985230002430dee231b792f8e"
227 },
228 {
229 "name": "input_proj.weight",
230 "shape": [
231 64,
232 67,
233 1
234 ],
235 "dtype": "float16",
236 "offset_bytes": 263174,
237 "nbytes": 8576,
238 "sha256": "945e2fa09e06e1ce4ba93308803090bfe665b12d31415274fc695ce1aa2e9e2c"
239 },
240 {
241 "name": "output.bias",
242 "shape": [
243 1
244 ],
245 "dtype": "float16",
246 "offset_bytes": 271750,
247 "nbytes": 2,
248 "sha256": "2215e8ac4e2b871c2a48189e79738c956c081e23ac2f2415bf77da199dfd920c"
249 },
250 {
251 "name": "output.weight",
252 "shape": [
253 1,
254 64,
255 1
256 ],
257 "dtype": "float16",
258 "offset_bytes": 271752,
259 "nbytes": 128,
260 "sha256": "6a5ea3ba28174473892bd0493111e54a59b959f7fb187b8fafea27fa40020096"
261 }
262 ]
263 },
264 "acoustic": {
265 "checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-hfc_female-medium-retrain-20260712/w3-wide192-joint/latent-student.pt",
266 "checkpoint_sha256": "633c3a25fea6775b8785495149d8823c0a2b583006100ec8f729cd1c9d8fb548",
267 "config": {
268 "architecture": "token_context",
269 "vocab_size": 145,
270 "hidden": 96,
271 "depth": 4,
272 "kernel_size": 5,
273 "out_channels": 192,
274 "token_depth": 3
275 },
276 "parameters": 698119,
277 "tensors": [
278 {
279 "name": "embedding.weight",
280 "shape": [
281 145,
282 96
283 ],
284 "dtype": "float16",
285 "offset_bytes": 271880,
286 "nbytes": 27840,
287 "sha256": "498c6cf6a03c0c3eabd742453d08104ec63c3aac901afc8b16614fc71d1e5bc3"
288 },
289 {
290 "name": "frame_blocks.0.net.0.bias",
291 "shape": [
292 96
293 ],
294 "dtype": "float16",
295 "offset_bytes": 299720,
296 "nbytes": 192,
297 "sha256": "13ec846a31f280e87d49dfb8b96d9a72b6e2adc8244bfd50889b30523bd58159"
298 },
299 {
300 "name": "frame_blocks.0.net.0.weight",
301 "shape": [
302 96,
303 96,
304 5
305 ],
306 "dtype": "float16",
307 "offset_bytes": 299912,
308 "nbytes": 92160,
309 "sha256": "9dfb3a46a268be61d0634451c1756b1c8cc7ee250440094947f9cbc0673c522d"
310 },
311 {
312 "name": "frame_blocks.0.net.2.bias",
313 "shape": [
314 96
315 ],
316 "dtype": "float16",
317 "offset_bytes": 392072,
318 "nbytes": 192,
319 "sha256": "1a162c6b059b3d4491d5396299c7d01537ffdd710a8e20c6c12345aad66a079c"
320 },
321 {
322 "name": "frame_blocks.0.net.2.weight",
323 "shape": [
324 96,
325 96,
326 5
327 ],
328 "dtype": "float16",
329 "offset_bytes": 392264,
330 "nbytes": 92160,
331 "sha256": "54d71ba11f83aaccc2fcb9d506301f0f96d6ce1f3b8a8ba74040cb39979433c3"
332 },
333 {
334 "name": "frame_blocks.0.scale",
335 "shape": [
336 1
337 ],
338 "dtype": "float16",
339 "offset_bytes": 484424,
340 "nbytes": 2,
341 "sha256": "6b71eb627a67eb847cba0f5858f6984e33dd8369bca8a6043c84cc78ed0310ba"
342 },
343 {
344 "name": "frame_blocks.1.net.0.bias",
345 "shape": [
346 96
347 ],
348 "dtype": "float16",
349 "offset_bytes": 484426,
350 "nbytes": 192,
351 "sha256": "5da57389aaf2f1283231927cc7d9e58fc4ce3b25378172d447d9aa8cdad430db"
352 },
353 {
354 "name": "frame_blocks.1.net.0.weight",
355 "shape": [
356 96,
357 96,
358 5
359 ],
360 "dtype": "float16",
361 "offset_bytes": 484618,
362 "nbytes": 92160,
363 "sha256": "add2db7212887407f4e1c6c1cf8e4b29400a9eab32d2ee111c7a9a61871b4226"
364 },
365 {
366 "name": "frame_blocks.1.net.2.bias",
367 "shape": [
368 96
369 ],
370 "dtype": "float16",
371 "offset_bytes": 576778,
372 "nbytes": 192,
373 "sha256": "b5101f14542a7463d3ba1f3c5a378db6f63912af820e24bb468b48611b309189"
374 },
375 {
376 "name": "frame_blocks.1.net.2.weight",
377 "shape": [
378 96,
379 96,
380 5
381 ],
382 "dtype": "float16",
383 "offset_bytes": 576970,
384 "nbytes": 92160,
385 "sha256": "9dd835d98aea7843eabcb32bceac992cc8ffd79aed09363b8847b8e50d21499e"
386 },
387 {
388 "name": "frame_blocks.1.scale",
389 "shape": [
390 1
391 ],
392 "dtype": "float16",
393 "offset_bytes": 669130,
394 "nbytes": 2,
395 "sha256": "988b32939f11f3207d5a75d1ebf047fb2a5a8e1995dc9af208b4fb1e9d21883d"
396 },
397 {
398 "name": "frame_blocks.2.net.0.bias",
399 "shape": [
400 96
401 ],
402 "dtype": "float16",
403 "offset_bytes": 669132,
404 "nbytes": 192,
405 "sha256": "8893cbca06439ab15a26093f5d190cbddc941893f510ee80a08c86cb99632631"
406 },
407 {
408 "name": "frame_blocks.2.net.0.weight",
409 "shape": [
410 96,
411 96,
412 5
413 ],
414 "dtype": "float16",
415 "offset_bytes": 669324,
416 "nbytes": 92160,
417 "sha256": "524c6505509e2867728671f825cad107c3db4742abfdb06cbb2b8fc661713553"
418 },
419 {
420 "name": "frame_blocks.2.net.2.bias",
421 "shape": [
422 96
423 ],
424 "dtype": "float16",
425 "offset_bytes": 761484,
426 "nbytes": 192,
427 "sha256": "d9c4eb8e4a2ab225a2d9d8a952f0a56dafce932b6d40196d8a1f5e5324953c00"
428 },
429 {
430 "name": "frame_blocks.2.net.2.weight",
431 "shape": [
432 96,
433 96,
434 5
435 ],
436 "dtype": "float16",
437 "offset_bytes": 761676,
438 "nbytes": 92160,
439 "sha256": "f23389471965a9a877dc87121d3c0e85c531b6b8f6df3b446a72ec5d05739428"
440 },
441 {
442 "name": "frame_blocks.2.scale",
443 "shape": [
444 1
445 ],
446 "dtype": "float16",
447 "offset_bytes": 853836,
448 "nbytes": 2,
449 "sha256": "86397704eb20f96180a94f21e0dd31591d66d436ebd9eda1611b0fbac2ce9ffc"
450 },
451 {
452 "name": "frame_blocks.3.net.0.bias",
453 "shape": [
454 96
455 ],
456 "dtype": "float16",
457 "offset_bytes": 853838,
458 "nbytes": 192,
459 "sha256": "5c1f652e21253517dd8e69ba0958f282175f2fcdc71e8ade999db17e03a2bf52"
460 },
461 {
462 "name": "frame_blocks.3.net.0.weight",
463 "shape": [
464 96,
465 96,
466 5
467 ],
468 "dtype": "float16",
469 "offset_bytes": 854030,
470 "nbytes": 92160,
471 "sha256": "f4454cd007aea6d3cdb059aed03b80dee39cab07d58012f7a10ea6ac63599771"
472 },
473 {
474 "name": "frame_blocks.3.net.2.bias",
475 "shape": [
476 96
477 ],
478 "dtype": "float16",
479 "offset_bytes": 946190,
480 "nbytes": 192,
481 "sha256": "c110555008958bd32c075132ea585f8cf94a2706e260e88e34b95241caff4f48"
482 },
483 {
484 "name": "frame_blocks.3.net.2.weight",
485 "shape": [
486 96,
487 96,
488 5
489 ],
490 "dtype": "float16",
491 "offset_bytes": 946382,
492 "nbytes": 92160,
493 "sha256": "b69561c5190e9d0e36092eb6d1ddc89235e93dd03e0ceefa977b39fc12342ef0"
494 },
495 {
496 "name": "frame_blocks.3.scale",
497 "shape": [
498 1
499 ],
500 "dtype": "float16",
501 "offset_bytes": 1038542,
502 "nbytes": 2,
503 "sha256": "da20960492e0576b01f49346960390edd002a638c9a973a1e76fc9c85a978066"
504 },
505 {
506 "name": "frame_input_proj.bias",
507 "shape": [
508 96
509 ],
510 "dtype": "float16",
511 "offset_bytes": 1038544,
512 "nbytes": 192,
513 "sha256": "a20c1404d0b055a2eae2b23b93c3c2cdc9f8108228ac039937b8c882a947f882"
514 },
515 {
516 "name": "frame_input_proj.weight",
517 "shape": [
518 96,
519 99,
520 1
521 ],
522 "dtype": "float16",
523 "offset_bytes": 1038736,
524 "nbytes": 19008,
525 "sha256": "cecb12f543316873f7c4574d16e580d3ab33ad4ccdc5dfa07326d7c48a09316c"
526 },
527 {
528 "name": "output.bias",
529 "shape": [
530 192
531 ],
532 "dtype": "float16",
533 "offset_bytes": 1057744,
534 "nbytes": 384,
535 "sha256": "96e515945f47aa84e04a206b0379c5f3a8f89ed512b08a2f6b19e5bcc5d0cda2"
536 },
537 {
538 "name": "output.weight",
539 "shape": [
540 192,
541 96,
542 1
543 ],
544 "dtype": "float16",
545 "offset_bytes": 1058128,
546 "nbytes": 36864,
547 "sha256": "b61a369d84f86c3ba0f386ec41fa05f81c206dc44def1230bd6347ac32e5d657"
548 },
549 {
550 "name": "token_blocks.0.net.0.bias",
551 "shape": [
552 96
553 ],
554 "dtype": "float16",
555 "offset_bytes": 1094992,
556 "nbytes": 192,
557 "sha256": "d02f3b597d03fbff758c41a96770bd77e7131bf1685e0cec34ce0b1685c0a320"
558 },
559 {
560 "name": "token_blocks.0.net.0.weight",
561 "shape": [
562 96,
563 96,
564 5
565 ],
566 "dtype": "float16",
567 "offset_bytes": 1095184,
568 "nbytes": 92160,
569 "sha256": "de967072e107722ae7558360928f5c0a776bb9be8f4056a13fbb63d9688d4a30"
570 },
571 {
572 "name": "token_blocks.0.net.2.bias",
573 "shape": [
574 96
575 ],
576 "dtype": "float16",
577 "offset_bytes": 1187344,
578 "nbytes": 192,
579 "sha256": "72b894e3de66fd4aade47a5f10bc53e4dbbd58a630fbc51a88553a6868d39922"
580 },
581 {
582 "name": "token_blocks.0.net.2.weight",
583 "shape": [
584 96,
585 96,
586 5
587 ],
588 "dtype": "float16",
589 "offset_bytes": 1187536,
590 "nbytes": 92160,
591 "sha256": "9f79904d9e811a5a43430d8c28b619fac7b3c62f620078a6980807e5ad123b01"
592 },
593 {
594 "name": "token_blocks.0.scale",
595 "shape": [
596 1
597 ],
598 "dtype": "float16",
599 "offset_bytes": 1279696,
600 "nbytes": 2,
601 "sha256": "2d7832dbab25091b94e11371bd6df3928e928163f3fcd51c72d3a1fdd4921442"
602 },
603 {
604 "name": "token_blocks.1.net.0.bias",
605 "shape": [
606 96
607 ],
608 "dtype": "float16",
609 "offset_bytes": 1279698,
610 "nbytes": 192,
611 "sha256": "a24fff539814f1ea45c77ceece1f74045a75927d8c565ae651e2f31050c3bc75"
612 },
613 {
614 "name": "token_blocks.1.net.0.weight",
615 "shape": [
616 96,
617 96,
618 5
619 ],
620 "dtype": "float16",
621 "offset_bytes": 1279890,
622 "nbytes": 92160,
623 "sha256": "eb457d3b332c1e728b05ea31091fddabc0779a4996282b5e5c434f7b7aa58c50"
624 },
625 {
626 "name": "token_blocks.1.net.2.bias",
627 "shape": [
628 96
629 ],
630 "dtype": "float16",
631 "offset_bytes": 1372050,
632 "nbytes": 192,
633 "sha256": "82e63caa326a959d3741a8405d1a1322b02447e117b6e98c2cc0e2a5412a97c3"
634 },
635 {
636 "name": "token_blocks.1.net.2.weight",
637 "shape": [
638 96,
639 96,
640 5
641 ],
642 "dtype": "float16",
643 "offset_bytes": 1372242,
644 "nbytes": 92160,
645 "sha256": "7020aa7d85f8d25057b015cebab677eb9270f5703dd5e3c636150d315418b2bf"
646 },
647 {
648 "name": "token_blocks.1.scale",
649 "shape": [
650 1
651 ],
652 "dtype": "float16",
653 "offset_bytes": 1464402,
654 "nbytes": 2,
655 "sha256": "f4adfdde386715720739db11602aea96b12e83b0bc32fc07d2d186bdecc2ff02"
656 },
657 {
658 "name": "token_blocks.2.net.0.bias",
659 "shape": [
660 96
661 ],
662 "dtype": "float16",
663 "offset_bytes": 1464404,
664 "nbytes": 192,
665 "sha256": "d44fc2a9312eea07a2227242eac4f6cb7f940c7b49ef397e745444ec89f847c9"
666 },
667 {
668 "name": "token_blocks.2.net.0.weight",
669 "shape": [
670 96,
671 96,
672 5
673 ],
674 "dtype": "float16",
675 "offset_bytes": 1464596,
676 "nbytes": 92160,
677 "sha256": "cbda2ed2f367df942c8071f147fb1494ac2cdf1a654c44c7ac66bfd0ac76a5b9"
678 },
679 {
680 "name": "token_blocks.2.net.2.bias",
681 "shape": [
682 96
683 ],
684 "dtype": "float16",
685 "offset_bytes": 1556756,
686 "nbytes": 192,
687 "sha256": "60ba825e2b661e980c72552bda745192210e0afc28d5fd1b1a7ea6e7303f7bcb"
688 },
689 {
690 "name": "token_blocks.2.net.2.weight",
691 "shape": [
692 96,
693 96,
694 5
695 ],
696 "dtype": "float16",
697 "offset_bytes": 1556948,
698 "nbytes": 92160,
699 "sha256": "8cef228f4ede0c78596c1ce177937eee8a21392999125d0f9624a501195972df"
700 },
701 {
702 "name": "token_blocks.2.scale",
703 "shape": [
704 1
705 ],
706 "dtype": "float16",
707 "offset_bytes": 1649108,
708 "nbytes": 2,
709 "sha256": "37e3eceb17a3c09b85067328f61fda68041c314e00932f5ef88a2c46b1dcf1d5"
710 },
711 {
712 "name": "token_input_proj.bias",
713 "shape": [
714 96
715 ],
716 "dtype": "float16",
717 "offset_bytes": 1649110,
718 "nbytes": 192,
719 "sha256": "95e39460d908b474e2b317fd3a47c60a24bf93df433bac892db0a775781bd3ae"
720 },
721 {
722 "name": "token_input_proj.weight",
723 "shape": [
724 96,
725 98,
726 1
727 ],
728 "dtype": "float16",
729 "offset_bytes": 1649302,
730 "nbytes": 18816,
731 "sha256": "8c6d4d0916e0fa58fecc055ce3b2ad67118d7c4d8451778849a0d2ea1c712a81"
732 }
733 ]
734 },
735 "decoder": {
736 "checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-hfc_female-medium-retrain-20260712/w3-wide192-joint/decoder-student.pt",
737 "checkpoint_sha256": "226cc7fc8a4b08d7f903037ee7dc9028c8482991a9e709581446d2c08f0ad455",
738 "config": {
739 "in_channels": 192,
740 "source_in_channels": 192,
741 "channels": [
742 192,
743 96,
744 48,
745 24
746 ],
747 "res_layers": 1,
748 "variant": "piperlite",
749 "lrc_code_dim": 40,
750 "lrc_encoder_hidden": 64,
751 "rank_ratio": 0.5,
752 "activation": "leaky_relu",
753 "stage_affine": false,
754 "factorized_pre_rank": 0,
755 "piper_res_factor_rank_ratio": 0.0,
756 "res_bank_scale_mode": "kept",
757 "stage0_branches": [
758 0,
759 1,
760 2
761 ],
762 "stage1_branches": [
763 0,
764 1,
765 2
766 ],
767 "stage2_branches": [
768 0,
769 1,
770 2
771 ],
772 "stage3_branches": [
773 0,
774 1,
775 2
776 ],
777 "fsd_dim": 72,
778 "fsd_blocks": 5,
779 "fsd_film_rank": 12,
780 "fsd_head_rank": 48,
781 "stage_projection_bottlenecks": [],
782 "teacher_init_checkpoint": null,
783 "teacher_init_method": "first",
784 "teacher_init": null,
785 "init_decoder_checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-hfc_female-medium-retrain-20260712/w1-wide192-40k/decoder-student.pt",
786 "decoder_init": {
787 "checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-hfc_female-medium-retrain-20260712/w1-wide192-40k/decoder-student.pt",
788 "decoder_parameters": 1000321,
789 "matched_config": {
790 "in_channels": 192,
791 "channels": [
792 192,
793 96,
794 48,
795 24
796 ],
797 "res_layers": 1,
798 "variant": "piperlite",
799 "rank_ratio": 0.5,
800 "activation": "leaky_relu",
801 "stage_affine": false,
802 "factorized_pre_rank": 0,
803 "piper_res_factor_rank_ratio": 0.0,
804 "res_bank_scale_mode": "kept",
805 "stage0_branches": [
806 0,
807 1,
808 2
809 ],
810 "stage1_branches": [
811 0,
812 1,
813 2
814 ],
815 "stage2_branches": [
816 0,
817 1,
818 2
819 ],
820 "stage3_branches": [
821 0,
822 1,
823 2
824 ],
825 "fsd_dim": 72,
826 "fsd_blocks": 5,
827 "fsd_film_rank": 12,
828 "fsd_head_rank": 48,
829 "stage_projection_bottlenecks": [],
830 "post_filter_channels": 0,
831 "post_filter_layers": 0,
832 "post_filter_kernel": 9,
833 "post_filter_scale": 0.25,
834 "pre_tanh_repair_channels": 0,
835 "pre_tanh_repair_layers": 0,
836 "pre_tanh_repair_kernel": 7,
837 "pre_tanh_repair_scale": 0.15,
838 "istft_n_fft": 512
839 },
840 "allowed_post_filter_mismatches": [],
841 "allowed_pre_tanh_repair_mismatches": [],
842 "allowed_activation_mismatches": [],
843 "missing_keys": [],
844 "unexpected_keys": []
845 },
846 "post_filter_channels": 0,
847 "post_filter_layers": 0,
848 "post_filter_kernel": 9,
849 "post_filter_scale": 0.25,
850 "pre_tanh_repair_channels": 0,
851 "pre_tanh_repair_layers": 0,
852 "pre_tanh_repair_kernel": 7,
853 "pre_tanh_repair_scale": 0.15,
854 "istft_n_fft": 512,
855 "ap_amplitude_weight": 0.0,
856 "ap_phase_weight": 0.0,
857 "ap_complex_weight": 0.0,
858 "spectral_head_init": "default",
859 "spectral_head_init_scale": 0.001,
860 "ap_amp_init_bias": -6.0,
861 "ap_phase_real_init_bias": 1.0,
862 "spectral_head_init_summary": null,
863 "feature_exact_weight": 0.0,
864 "feature_exact_keys": [
865 "pre",
866 "up0"
867 ],
868 "feature_exact_teacher_init": null,
869 "signature_pack_dir": "artifacts/sub10m-search/root-a-piper-vits/en_US-hfc_female-medium-retrain-20260712/a26-signatures-train512",
870 "exact_feature_pack_dir": null,
871 "signature_hint_weight": 0.05,
872 "signature_temporal_weight": 0.4,
873 "signature_phase_weight": 0.0,
874 "signature_phase_bins": 0,
875 "signature_keys": [
876 "stage0_mix",
877 "stage1_mix",
878 "stage2_mix",
879 "pre_tanh",
880 "audio"
881 ],
882 "bottleneck_code_checkpoint": null,
883 "bottleneck_code_weight": 0.0,
884 "bottleneck_code_keys": [],
885 "bottleneck_code_training_only_parameters": 0,
886 "echo_tail_weight": 0.0,
887 "echo_tail_min_ms": 8.0,
888 "echo_tail_max_ms": 40.0,
889 "echo_tail_lags": 9,
890 "echo_tail_margin": 0.02,
891 "paired_acoustic_residual_weight": 0.0,
892 "paired_acoustic_residual_max_scale": 0.25,
893 "lrc_pred_code_mix_prob": 0.0,
894 "lrc_pred_code_checkpoint": null,
895 "lrc_pred_code_residual_prob": 0.0,
896 "lrc_pred_code_residual_max_scale": 0.25,
897 "lrc_pred_code_cache": null
898 },
899 "parameters": 1000321,
900 "tensors": [
901 {
902 "name": "post.bias",
903 "shape": [
904 1
905 ],
906 "dtype": "float16",
907 "offset_bytes": 1668118,
908 "nbytes": 2,
909 "sha256": "86b0fd7a0f97f4f4c0be18e2275f04c5c0e1516b4f4a4e9481dbf4371b3fe2d5"
910 },
911 {
912 "name": "post.weight",
913 "shape": [
914 1,
915 24,
916 7
917 ],
918 "dtype": "float16",
919 "offset_bytes": 1668120,
920 "nbytes": 336,
921 "sha256": "a111b84c83e3b9b248a28ebad035fc7c31b4d4287d1f90c4cc1216ae89e41ab8"
922 },
923 {
924 "name": "pre.bias",
925 "shape": [
926 192
927 ],
928 "dtype": "float16",
929 "offset_bytes": 1668456,
930 "nbytes": 384,
931 "sha256": "385908de4bef9e637d36edc9e44e1e6e0628bb2f095d4975abdb0c4fc059175e"
932 },
933 {
934 "name": "pre.weight",
935 "shape": [
936 192,
937 192,
938 7
939 ],
940 "dtype": "float16",
941 "offset_bytes": 1668840,
942 "nbytes": 516096,
943 "sha256": "992f4aedbcc48d178f218cd57ab2d33a1276d55a022fa1afcfb18fa7fd5d8dbb"
944 },
945 {
946 "name": "res0.0.blocks.0.conv1.bias",
947 "shape": [
948 96
949 ],
950 "dtype": "float16",
951 "offset_bytes": 2184936,
952 "nbytes": 192,
953 "sha256": "29581507c3f207c63a993da50b3b3a14140b5e28f025c050aa9b6cf2537aed74"
954 },
955 {
956 "name": "res0.0.blocks.0.conv1.weight",
957 "shape": [
958 96,
959 96,
960 3
961 ],
962 "dtype": "float16",
963 "offset_bytes": 2185128,
964 "nbytes": 55296,
965 "sha256": "2154f19177e365620a964efe0e2eb364b2e898129c639fd43d679df5cd8f0b19"
966 },
967 {
968 "name": "res0.0.blocks.0.conv2.bias",
969 "shape": [
970 96
971 ],
972 "dtype": "float16",
973 "offset_bytes": 2240424,
974 "nbytes": 192,
975 "sha256": "9fe9af40808b35e7b2d20b482ba66318f7bb5e0c20df22c7edf768ad748aca44"
976 },
977 {
978 "name": "res0.0.blocks.0.conv2.weight",
979 "shape": [
980 96,
981 96,
982 3
983 ],
984 "dtype": "float16",
985 "offset_bytes": 2240616,
986 "nbytes": 55296,
987 "sha256": "2963fa0ba0fdade5572935d2c2430f6cf1b11c5c043509bd24ded39baf49494f"
988 },
989 {
990 "name": "res0.0.blocks.1.conv1.bias",
991 "shape": [
992 96
993 ],
994 "dtype": "float16",
995 "offset_bytes": 2295912,
996 "nbytes": 192,
997 "sha256": "5a3625ca72af8f82ade37e68b363cca45d9a7485e8bd9a8e223059082d2c7e6a"
998 },
999 {
1000 "name": "res0.0.blocks.1.conv1.weight",
1001 "shape": [
1002 96,
1003 96,
1004 5
1005 ],
1006 "dtype": "float16",
1007 "offset_bytes": 2296104,
1008 "nbytes": 92160,
1009 "sha256": "7531fe533103a3e385ffb833432a44815fbe14901a80fc93f1668bfdb0172fdf"
1010 },
1011 {
1012 "name": "res0.0.blocks.1.conv2.bias",
1013 "shape": [
1014 96
1015 ],
1016 "dtype": "float16",
1017 "offset_bytes": 2388264,
1018 "nbytes": 192,
1019 "sha256": "4af71f180f69c283ad217e2eb1af2d8e6666084f923465bea54bbca99629eb3a"
1020 },
1021 {
1022 "name": "res0.0.blocks.1.conv2.weight",
1023 "shape": [
1024 96,
1025 96,
1026 5
1027 ],
1028 "dtype": "float16",
1029 "offset_bytes": 2388456,
1030 "nbytes": 92160,
1031 "sha256": "289ee4ec3d76c09ad3ee7e8858c9dda324760ffd05ecfd9af6f7922ccb337b9e"
1032 },
1033 {
1034 "name": "res0.0.blocks.2.conv1.bias",
1035 "shape": [
1036 96
1037 ],
1038 "dtype": "float16",
1039 "offset_bytes": 2480616,
1040 "nbytes": 192,
1041 "sha256": "9538b527ed72f7875d99eaf0bff83e8eb6a09a9ffcf7b93048a183fa85d84042"
1042 },
1043 {
1044 "name": "res0.0.blocks.2.conv1.weight",
1045 "shape": [
1046 96,
1047 96,
1048 7
1049 ],
1050 "dtype": "float16",
1051 "offset_bytes": 2480808,
1052 "nbytes": 129024,
1053 "sha256": "b62471abb4411b5be5bbfe3e57917e45e8b03b62f508a5f16f6f10e23b35d5c9"
1054 },
1055 {
1056 "name": "res0.0.blocks.2.conv2.bias",
1057 "shape": [
1058 96
1059 ],
1060 "dtype": "float16",
1061 "offset_bytes": 2609832,
1062 "nbytes": 192,
1063 "sha256": "cb2af372d9dcbb2f2cfbaf98f971998af3d2a6c19b324cc2048378875227793b"
1064 },
1065 {
1066 "name": "res0.0.blocks.2.conv2.weight",
1067 "shape": [
1068 96,
1069 96,
1070 7
1071 ],
1072 "dtype": "float16",
1073 "offset_bytes": 2610024,
1074 "nbytes": 129024,
1075 "sha256": "3ab3e83ff640bc52957b04be4a5c84146fa06ab5c2c66eeb2a72619db92277de"
1076 },
1077 {
1078 "name": "res1.0.blocks.0.conv1.bias",
1079 "shape": [
1080 48
1081 ],
1082 "dtype": "float16",
1083 "offset_bytes": 2739048,
1084 "nbytes": 96,
1085 "sha256": "da6daf1c8e3700e8a2b564478cd0590ceb489846c2e5ee4ec6bd7cd6ceb9b3cc"
1086 },
1087 {
1088 "name": "res1.0.blocks.0.conv1.weight",
1089 "shape": [
1090 48,
1091 48,
1092 3
1093 ],
1094 "dtype": "float16",
1095 "offset_bytes": 2739144,
1096 "nbytes": 13824,
1097 "sha256": "d66f15a27f0f2c544aad084ec35f60e3b9ff81d7fc6b8fc262983f2164637090"
1098 },
1099 {
1100 "name": "res1.0.blocks.0.conv2.bias",
1101 "shape": [
1102 48
1103 ],
1104 "dtype": "float16",
1105 "offset_bytes": 2752968,
1106 "nbytes": 96,
1107 "sha256": "b7b771434e83905550fd49328c40af59a049fc7f643075cbcf45cb87c82e14e2"
1108 },
1109 {
1110 "name": "res1.0.blocks.0.conv2.weight",
1111 "shape": [
1112 48,
1113 48,
1114 3
1115 ],
1116 "dtype": "float16",
1117 "offset_bytes": 2753064,
1118 "nbytes": 13824,
1119 "sha256": "9aab48abe428fa40261952c15ef073522dfaf7c445750b412893f524ed0ef59c"
1120 },
1121 {
1122 "name": "res1.0.blocks.1.conv1.bias",
1123 "shape": [
1124 48
1125 ],
1126 "dtype": "float16",
1127 "offset_bytes": 2766888,
1128 "nbytes": 96,
1129 "sha256": "712607a469a53e3b4483b1d2e891fb662f33c04d8072ee9d3b2dbea8d482b28f"
1130 },
1131 {
1132 "name": "res1.0.blocks.1.conv1.weight",
1133 "shape": [
1134 48,
1135 48,
1136 5
1137 ],
1138 "dtype": "float16",
1139 "offset_bytes": 2766984,
1140 "nbytes": 23040,
1141 "sha256": "106eee5785076981dde779b0d6c92fcbe6acb05875d84ab8fc73bbd30f3e16cb"
1142 },
1143 {
1144 "name": "res1.0.blocks.1.conv2.bias",
1145 "shape": [
1146 48
1147 ],
1148 "dtype": "float16",
1149 "offset_bytes": 2790024,
1150 "nbytes": 96,
1151 "sha256": "41af691b3f4291eb2d8cb7b5c5d30309db71ac6e9646d1594b5c530b9c7bc860"
1152 },
1153 {
1154 "name": "res1.0.blocks.1.conv2.weight",
1155 "shape": [
1156 48,
1157 48,
1158 5
1159 ],
1160 "dtype": "float16",
1161 "offset_bytes": 2790120,
1162 "nbytes": 23040,
1163 "sha256": "c1848b3051fc6ad2a9e2b48e2e4f0d55a2b0e5e9544771505e3fc78061fd779d"
1164 },
1165 {
1166 "name": "res1.0.blocks.2.conv1.bias",
1167 "shape": [
1168 48
1169 ],
1170 "dtype": "float16",
1171 "offset_bytes": 2813160,
1172 "nbytes": 96,
1173 "sha256": "065e29ea3ee80d8dbc220297aedc7bdf2371ccfa7a882b0dae9508b5adc88fbf"
1174 },
1175 {
1176 "name": "res1.0.blocks.2.conv1.weight",
1177 "shape": [
1178 48,
1179 48,
1180 7
1181 ],
1182 "dtype": "float16",
1183 "offset_bytes": 2813256,
1184 "nbytes": 32256,
1185 "sha256": "6ed865bcf41e5644f4e19ce83c27ce0884de10010dc9a8568904e0e2a5bb5c42"
1186 },
1187 {
1188 "name": "res1.0.blocks.2.conv2.bias",
1189 "shape": [
1190 48
1191 ],
1192 "dtype": "float16",
1193 "offset_bytes": 2845512,
1194 "nbytes": 96,
1195 "sha256": "82d7da51aae6411e1b5da7ef549de03c53de4f000aa2a277aff7be67f87f0a16"
1196 },
1197 {
1198 "name": "res1.0.blocks.2.conv2.weight",
1199 "shape": [
1200 48,
1201 48,
1202 7
1203 ],
1204 "dtype": "float16",
1205 "offset_bytes": 2845608,
1206 "nbytes": 32256,
1207 "sha256": "a55131323378b135014f8030416813dad0289e50a3bb5646098ef7a84b2a6cbd"
1208 },
1209 {
1210 "name": "res2.0.blocks.0.conv1.bias",
1211 "shape": [
1212 24
1213 ],
1214 "dtype": "float16",
1215 "offset_bytes": 2877864,
1216 "nbytes": 48,
1217 "sha256": "ae971616091a4354fe889801f212165d79e70054417d86306a8c4cff7382fd89"
1218 },
1219 {
1220 "name": "res2.0.blocks.0.conv1.weight",
1221 "shape": [
1222 24,
1223 24,
1224 3
1225 ],
1226 "dtype": "float16",
1227 "offset_bytes": 2877912,
1228 "nbytes": 3456,
1229 "sha256": "6a6d9517f84794c3f286ccda722110ff37913a234f5d9680f5bb73d36d355d98"
1230 },
1231 {
1232 "name": "res2.0.blocks.0.conv2.bias",
1233 "shape": [
1234 24
1235 ],
1236 "dtype": "float16",
1237 "offset_bytes": 2881368,
1238 "nbytes": 48,
1239 "sha256": "a73a72d1db5b3894da2b5baab40a0de4fab07e0bbb495292ae45b8f47fe7ac77"
1240 },
1241 {
1242 "name": "res2.0.blocks.0.conv2.weight",
1243 "shape": [
1244 24,
1245 24,
1246 3
1247 ],
1248 "dtype": "float16",
1249 "offset_bytes": 2881416,
1250 "nbytes": 3456,
1251 "sha256": "19feadb645230801dc1cc6a592ff2c1919c4616e03a6871bc27987fc240bc518"
1252 },
1253 {
1254 "name": "res2.0.blocks.1.conv1.bias",
1255 "shape": [
1256 24
1257 ],
1258 "dtype": "float16",
1259 "offset_bytes": 2884872,
1260 "nbytes": 48,
1261 "sha256": "ab31af7c25c3391fb27df76ba211e873d670cca7bb56b01b24c581590ede5580"
1262 },
1263 {
1264 "name": "res2.0.blocks.1.conv1.weight",
1265 "shape": [
1266 24,
1267 24,
1268 5
1269 ],
1270 "dtype": "float16",
1271 "offset_bytes": 2884920,
1272 "nbytes": 5760,
1273 "sha256": "911c331de319e354ec4eb43a09d9795a1feff97a9affac1737112a10faebc9da"
1274 },
1275 {
1276 "name": "res2.0.blocks.1.conv2.bias",
1277 "shape": [
1278 24
1279 ],
1280 "dtype": "float16",
1281 "offset_bytes": 2890680,
1282 "nbytes": 48,
1283 "sha256": "9a7b3e535ec8b3afa6a01ea2fd98ac584bcf3ac2a24d82b0829ec4ccb8b4f800"
1284 },
1285 {
1286 "name": "res2.0.blocks.1.conv2.weight",
1287 "shape": [
1288 24,
1289 24,
1290 5
1291 ],
1292 "dtype": "float16",
1293 "offset_bytes": 2890728,
1294 "nbytes": 5760,
1295 "sha256": "48af87e63e88e747fe9d11cb334d606b400674134eacf3e872a4a3ca2516ccea"
1296 },
1297 {
1298 "name": "res2.0.blocks.2.conv1.bias",
1299 "shape": [
1300 24
1301 ],
1302 "dtype": "float16",
1303 "offset_bytes": 2896488,
1304 "nbytes": 48,
1305 "sha256": "95ab888d73280b318d55c9b3bc94df6a7b218be67d93472074dec7a881e8b67d"
1306 },
1307 {
1308 "name": "res2.0.blocks.2.conv1.weight",
1309 "shape": [
1310 24,
1311 24,
1312 7
1313 ],
1314 "dtype": "float16",
1315 "offset_bytes": 2896536,
1316 "nbytes": 8064,
1317 "sha256": "22df2c1b54462e7a67d46a1ecd8d88eb8b37d82c300aaf9ed504665e0d4d0575"
1318 },
1319 {
1320 "name": "res2.0.blocks.2.conv2.bias",
1321 "shape": [
1322 24
1323 ],
1324 "dtype": "float16",
1325 "offset_bytes": 2904600,
1326 "nbytes": 48,
1327 "sha256": "fa11b2a860c8dd4dc52ca47127ef18599fb08be7de21ac1c9d8a37fddca457ba"
1328 },
1329 {
1330 "name": "res2.0.blocks.2.conv2.weight",
1331 "shape": [
1332 24,
1333 24,
1334 7
1335 ],
1336 "dtype": "float16",
1337 "offset_bytes": 2904648,
1338 "nbytes": 8064,
1339 "sha256": "791e5317c51a22f8ca16feb3773bcf92d88208000834bf69462f2ec94614df16"
1340 },
1341 {
1342 "name": "up0.bias",
1343 "shape": [
1344 96
1345 ],
1346 "dtype": "float16",
1347 "offset_bytes": 2912712,
1348 "nbytes": 192,
1349 "sha256": "c2af0c25a62e0d9080b4202eb156a3471ac8a4651986c9050d129b4e27258ba3"
1350 },
1351 {
1352 "name": "up0.weight",
1353 "shape": [
1354 192,
1355 96,
1356 16
1357 ],
1358 "dtype": "float16",
1359 "offset_bytes": 2912904,
1360 "nbytes": 589824,
1361 "sha256": "a0f85af855586fb4e875877c05963b93b01c0b7d521cb559338117476a6dce8e"
1362 },
1363 {
1364 "name": "up1.bias",
1365 "shape": [
1366 48
1367 ],
1368 "dtype": "float16",
1369 "offset_bytes": 3502728,
1370 "nbytes": 96,
1371 "sha256": "acbd0b19621c4c9563d6f9337691c97f51fe5dfb74d27f648b31c7661f5b5c78"
1372 },
1373 {
1374 "name": "up1.weight",
1375 "shape": [
1376 96,
1377 48,
1378 16
1379 ],
1380 "dtype": "float16",
1381 "offset_bytes": 3502824,
1382 "nbytes": 147456,
1383 "sha256": "80bb13346ebf4d78738cbfc1eca4994cdfc7c231388c6a4add2a6b68a7f7ca34"
1384 },
1385 {
1386 "name": "up2.bias",
1387 "shape": [
1388 24
1389 ],
1390 "dtype": "float16",
1391 "offset_bytes": 3650280,
1392 "nbytes": 48,
1393 "sha256": "c1712ec69e68901f11f7c08c10324e000d4731eededd863adc9b6fe45fd69427"
1394 },
1395 {
1396 "name": "up2.weight",
1397 "shape": [
1398 48,
1399 24,
1400 8
1401 ],
1402 "dtype": "float16",
1403 "offset_bytes": 3650328,
1404 "nbytes": 18432,
1405 "sha256": "a2caf36212572c9bcbb698c753ee03b730d2151f415b706bdc403cf547165be5"
1406 }
1407 ]
1408 }
1409 },
1410 "runtime_ops": [
1411 "embedding_lookup",
1412 "conv1d_same",
1413 "conv1d_1x1",
1414 "conv_transpose1d",
1415 "silu",
1416 "leaky_relu",
1417 "tanh",
1418 "residual_add",
1419 "repeat_interleave",
1420 "masked_conv1d",
1421 "depthwise_conv1d",
1422 "low_rank_linear_or_conv1d"
1423 ],
1424 "runtime_boundary": {
1425 "neural_stack_self_contained": true,
1426 "arbitrary_text_self_contained": false,
1427 "accepted_runtime_input": "Piper phoneme ID sequence matching piper-phoneme-config.json",
1428 "output": "mono float waveform at manifest.sample_rate"
1429 },
1430 "frontend": {
1431 "status": "not fully self-contained for arbitrary text yet",
1432 "input_contract": "runtime may accept Piper phoneme ID sequences directly",
1433 "included_config": "piper-phoneme-config.json",
1434 "piper_phoneme_type": "espeak",
1435 "piper_espeak_voice": "en-us",
1436 "phoneme_id_map_entries": 159,
1437 "next_work": "embed or rewrite the Piper/eSpeak-compatible frontend, or keep a strict phoneme-ID API"
1438 },
1439 "size_budget": {
1440 "target_max_bytes": 4000000,
1441 "target_max_mib": 3.814697265625,
1442 "actual_model_asset_bytes": 3673793,
1443 "target_max_parameters": 1500000,
1444 "under_parameter_budget": false,
1445 "under_4mb_asset_budget": true
1446 },
1447 "limitations": [
1448 "This is a raw neural asset package, not a compiled standalone runtime binary.",
1449 "No PyTorch checkpoint, ONNX graph, or training state is included.",
1450 "Arbitrary text still requires a Piper/eSpeak-compatible frontend outside this package."
1451 ]
1452 }
1453