amy-en-1p1m/manifest.json
40.7 KB · 1453 lines · json Raw
1 {
2 "format": "roota.raw-fp16.v1",
3 "package_name": "amy-en-1p1m",
4 "language": "en_US",
5 "voice": "en_US-amy-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": 2169944,
15 "weights_sha256": "b45240e7c24ac4ca4a1572a7ccc971dcca1836dcfde9aeb205d54d5515a07610",
16 "total_parameters": 1084972,
17 "components": {
18 "duration": {
19 "checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-amy-medium-retrain-20260712/t-dur-h32/duration-student.pt",
20 "checkpoint_sha256": "345172ee86f602a77870703f488b23e71d1b358b24843a3e15d344ede511683e",
21 "config": {
22 "architecture": "duration_conv",
23 "vocab_size": 129,
24 "hidden": 32,
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": 36228,
44 "tensors": [
45 {
46 "name": "blocks.0.net.0.bias",
47 "shape": [
48 32
49 ],
50 "dtype": "float16",
51 "offset_bytes": 0,
52 "nbytes": 64,
53 "sha256": "3484279e5a71d853b4db8cc362a6d89f8273f676e8eec7513a9e9366fd1868be"
54 },
55 {
56 "name": "blocks.0.net.0.weight",
57 "shape": [
58 32,
59 32,
60 5
61 ],
62 "dtype": "float16",
63 "offset_bytes": 64,
64 "nbytes": 10240,
65 "sha256": "5191ae4966eba6fa2599cc83851bd4a4304310def4db93ed9513ec98e06f2ea2"
66 },
67 {
68 "name": "blocks.0.net.2.bias",
69 "shape": [
70 32
71 ],
72 "dtype": "float16",
73 "offset_bytes": 10304,
74 "nbytes": 64,
75 "sha256": "70a9872b7813cdb2c52f1d0c5cf10b91a55c642deece1ed57c3405f206aeb9c5"
76 },
77 {
78 "name": "blocks.0.net.2.weight",
79 "shape": [
80 32,
81 32,
82 5
83 ],
84 "dtype": "float16",
85 "offset_bytes": 10368,
86 "nbytes": 10240,
87 "sha256": "9cb40d457d2f1d45632282771c17af811275b3c088bd0af6367b8a78de60e687"
88 },
89 {
90 "name": "blocks.0.scale",
91 "shape": [
92 1
93 ],
94 "dtype": "float16",
95 "offset_bytes": 20608,
96 "nbytes": 2,
97 "sha256": "e4a8bf0e8b9e6298d9e1ba62dd0a38395f5b14750804d585e7ba069ee2084e2e"
98 },
99 {
100 "name": "blocks.1.net.0.bias",
101 "shape": [
102 32
103 ],
104 "dtype": "float16",
105 "offset_bytes": 20610,
106 "nbytes": 64,
107 "sha256": "a860191116846fcc57157e4876225c9431665d96db02f33736fde4e1edf61d3a"
108 },
109 {
110 "name": "blocks.1.net.0.weight",
111 "shape": [
112 32,
113 32,
114 5
115 ],
116 "dtype": "float16",
117 "offset_bytes": 20674,
118 "nbytes": 10240,
119 "sha256": "43edcdb124672461b38d67811e8cd6b51de5a5ed3640eadf9bb054569d695d68"
120 },
121 {
122 "name": "blocks.1.net.2.bias",
123 "shape": [
124 32
125 ],
126 "dtype": "float16",
127 "offset_bytes": 30914,
128 "nbytes": 64,
129 "sha256": "76d12ee13fcb0866fccc6a7d383c7a9a3fff5d9a6fe551107ddcc1c09576ae13"
130 },
131 {
132 "name": "blocks.1.net.2.weight",
133 "shape": [
134 32,
135 32,
136 5
137 ],
138 "dtype": "float16",
139 "offset_bytes": 30978,
140 "nbytes": 10240,
141 "sha256": "88e8407002f0936991df0796b773bfe9d47cc99a2a602abaf21f78a5bc6162b9"
142 },
143 {
144 "name": "blocks.1.scale",
145 "shape": [
146 1
147 ],
148 "dtype": "float16",
149 "offset_bytes": 41218,
150 "nbytes": 2,
151 "sha256": "ee0afb030fe0274152779b8ca6788588440baf31aeb1fab592ee377f4211fa9b"
152 },
153 {
154 "name": "blocks.2.net.0.bias",
155 "shape": [
156 32
157 ],
158 "dtype": "float16",
159 "offset_bytes": 41220,
160 "nbytes": 64,
161 "sha256": "1c7f0d36bea88c69e025b858212bd24c9f8424844497f68b4e823b2c721279d1"
162 },
163 {
164 "name": "blocks.2.net.0.weight",
165 "shape": [
166 32,
167 32,
168 5
169 ],
170 "dtype": "float16",
171 "offset_bytes": 41284,
172 "nbytes": 10240,
173 "sha256": "df1308b1af15ea5202068bba6cc3fb1673efccc7a82db01f9f649d5b4421c73c"
174 },
175 {
176 "name": "blocks.2.net.2.bias",
177 "shape": [
178 32
179 ],
180 "dtype": "float16",
181 "offset_bytes": 51524,
182 "nbytes": 64,
183 "sha256": "e256535ae6d06fc42a2a27f1aa793a0cd06497ad7b4aa25d789bc0d7b2ed34c5"
184 },
185 {
186 "name": "blocks.2.net.2.weight",
187 "shape": [
188 32,
189 32,
190 5
191 ],
192 "dtype": "float16",
193 "offset_bytes": 51588,
194 "nbytes": 10240,
195 "sha256": "84a7cf07dc7ca206e06bf6e130e1cbce88b7f57394f583348236bec602fb6506"
196 },
197 {
198 "name": "blocks.2.scale",
199 "shape": [
200 1
201 ],
202 "dtype": "float16",
203 "offset_bytes": 61828,
204 "nbytes": 2,
205 "sha256": "2c1b08f8b6d560b5cc346ce88975feebca160f6a5171480d322f724d73b3143d"
206 },
207 {
208 "name": "embedding.weight",
209 "shape": [
210 129,
211 32
212 ],
213 "dtype": "float16",
214 "offset_bytes": 61830,
215 "nbytes": 8256,
216 "sha256": "290edfdc8f3188798bde9c09527d7e2fd5e2cc12fe7d0122bb6e05797fa285b9"
217 },
218 {
219 "name": "input_proj.bias",
220 "shape": [
221 32
222 ],
223 "dtype": "float16",
224 "offset_bytes": 70086,
225 "nbytes": 64,
226 "sha256": "1bf00c244ec0b729ce2309d3c42e6c8ef5b10b2095dacd5a686ce1768332f785"
227 },
228 {
229 "name": "input_proj.weight",
230 "shape": [
231 32,
232 35,
233 1
234 ],
235 "dtype": "float16",
236 "offset_bytes": 70150,
237 "nbytes": 2240,
238 "sha256": "8052379e33b046dc8d04aaafd0cb8720287d80242b123d4d76d554dc5e4a95ac"
239 },
240 {
241 "name": "output.bias",
242 "shape": [
243 1
244 ],
245 "dtype": "float16",
246 "offset_bytes": 72390,
247 "nbytes": 2,
248 "sha256": "669160490670712db4dbcb6708054ff07a781c894ebf70be91b98ca49a746237"
249 },
250 {
251 "name": "output.weight",
252 "shape": [
253 1,
254 32,
255 1
256 ],
257 "dtype": "float16",
258 "offset_bytes": 72392,
259 "nbytes": 64,
260 "sha256": "c25fc15433e7a89b81cecaec5a1c11a8a04366bbc60282b3959d16747ba8e981"
261 }
262 ]
263 },
264 "acoustic": {
265 "checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-amy-medium-retrain-20260712/e3-lite731-ac64-joint/latent-student.pt",
266 "checkpoint_sha256": "fafc2be3f57fc1694913ff4ba101f511372715d220ec3e25d446645503f04765",
267 "config": {
268 "architecture": "token_context",
269 "vocab_size": 145,
270 "hidden": 64,
271 "depth": 4,
272 "kernel_size": 5,
273 "out_channels": 192,
274 "token_depth": 3
275 },
276 "parameters": 318023,
277 "tensors": [
278 {
279 "name": "embedding.weight",
280 "shape": [
281 145,
282 64
283 ],
284 "dtype": "float16",
285 "offset_bytes": 72456,
286 "nbytes": 18560,
287 "sha256": "119a6c41dbabb51162cb6bf4828ac29e69232fd121f4b203aa6edb3c59c28572"
288 },
289 {
290 "name": "frame_blocks.0.net.0.bias",
291 "shape": [
292 64
293 ],
294 "dtype": "float16",
295 "offset_bytes": 91016,
296 "nbytes": 128,
297 "sha256": "46fd2dac79e5626185751b7e2d78cdc577c2071eabb8130f13007dc8686d1269"
298 },
299 {
300 "name": "frame_blocks.0.net.0.weight",
301 "shape": [
302 64,
303 64,
304 5
305 ],
306 "dtype": "float16",
307 "offset_bytes": 91144,
308 "nbytes": 40960,
309 "sha256": "489e192bee4c690a87d4653b01a72cb4633a4d042b746aa1ea9a278e3e70dfa0"
310 },
311 {
312 "name": "frame_blocks.0.net.2.bias",
313 "shape": [
314 64
315 ],
316 "dtype": "float16",
317 "offset_bytes": 132104,
318 "nbytes": 128,
319 "sha256": "948187616eaa4ce055be35eebaa0e3cf89fb5976a3425101fa3b1289a03ee9fb"
320 },
321 {
322 "name": "frame_blocks.0.net.2.weight",
323 "shape": [
324 64,
325 64,
326 5
327 ],
328 "dtype": "float16",
329 "offset_bytes": 132232,
330 "nbytes": 40960,
331 "sha256": "856887d4e2a1250895d7aaa85acd29d739f237a418674666321cca66dde3242a"
332 },
333 {
334 "name": "frame_blocks.0.scale",
335 "shape": [
336 1
337 ],
338 "dtype": "float16",
339 "offset_bytes": 173192,
340 "nbytes": 2,
341 "sha256": "b65e97d0c3f7a87e8eefc9751b1eb8fc90fde5bb1c9cf51994ab0f457eccf16b"
342 },
343 {
344 "name": "frame_blocks.1.net.0.bias",
345 "shape": [
346 64
347 ],
348 "dtype": "float16",
349 "offset_bytes": 173194,
350 "nbytes": 128,
351 "sha256": "3557ba451b6fd693eb39d12692d4e80aecced6ad47efd53bb2d6acf86e65000c"
352 },
353 {
354 "name": "frame_blocks.1.net.0.weight",
355 "shape": [
356 64,
357 64,
358 5
359 ],
360 "dtype": "float16",
361 "offset_bytes": 173322,
362 "nbytes": 40960,
363 "sha256": "80544b191de8274289881dae779c330fa66d57a0ee25edaffd7010c2eb2e3cea"
364 },
365 {
366 "name": "frame_blocks.1.net.2.bias",
367 "shape": [
368 64
369 ],
370 "dtype": "float16",
371 "offset_bytes": 214282,
372 "nbytes": 128,
373 "sha256": "7a967791af67822b6def9e0194e61b9d14b068224a11837198f1a39d023c0537"
374 },
375 {
376 "name": "frame_blocks.1.net.2.weight",
377 "shape": [
378 64,
379 64,
380 5
381 ],
382 "dtype": "float16",
383 "offset_bytes": 214410,
384 "nbytes": 40960,
385 "sha256": "66fd49be76925d60c2dc65f366ca944849ae354ac779d59d4054c4852ebe73eb"
386 },
387 {
388 "name": "frame_blocks.1.scale",
389 "shape": [
390 1
391 ],
392 "dtype": "float16",
393 "offset_bytes": 255370,
394 "nbytes": 2,
395 "sha256": "392502be3a1794de84879ec25b1d0fef66802992a8727f2997ae8bcc0b4bea9d"
396 },
397 {
398 "name": "frame_blocks.2.net.0.bias",
399 "shape": [
400 64
401 ],
402 "dtype": "float16",
403 "offset_bytes": 255372,
404 "nbytes": 128,
405 "sha256": "b58c2701c6e2e1cf0999168347582858f076b1f2781662410acf41c1f5e967e3"
406 },
407 {
408 "name": "frame_blocks.2.net.0.weight",
409 "shape": [
410 64,
411 64,
412 5
413 ],
414 "dtype": "float16",
415 "offset_bytes": 255500,
416 "nbytes": 40960,
417 "sha256": "80c0701b590713f42363a0a17df56f1ec8b3ead92411afe97f3faa3e0c9b068b"
418 },
419 {
420 "name": "frame_blocks.2.net.2.bias",
421 "shape": [
422 64
423 ],
424 "dtype": "float16",
425 "offset_bytes": 296460,
426 "nbytes": 128,
427 "sha256": "e46cf5d207c18c3f08c38250c7eb507529a35435f52dd81e2f75f298efb1cc9b"
428 },
429 {
430 "name": "frame_blocks.2.net.2.weight",
431 "shape": [
432 64,
433 64,
434 5
435 ],
436 "dtype": "float16",
437 "offset_bytes": 296588,
438 "nbytes": 40960,
439 "sha256": "bd29cc35a0625b057e7f2cdd781eb51c06a2fa53fbb23d692b6253b4d8fac741"
440 },
441 {
442 "name": "frame_blocks.2.scale",
443 "shape": [
444 1
445 ],
446 "dtype": "float16",
447 "offset_bytes": 337548,
448 "nbytes": 2,
449 "sha256": "903212996fa1b93e35a7ef3876f32f17068470b3b9de74445b5d62012fd66af0"
450 },
451 {
452 "name": "frame_blocks.3.net.0.bias",
453 "shape": [
454 64
455 ],
456 "dtype": "float16",
457 "offset_bytes": 337550,
458 "nbytes": 128,
459 "sha256": "7033477864960a9c9541570a950a07c8dd248c402b6b0b7b96fedc9e5570430e"
460 },
461 {
462 "name": "frame_blocks.3.net.0.weight",
463 "shape": [
464 64,
465 64,
466 5
467 ],
468 "dtype": "float16",
469 "offset_bytes": 337678,
470 "nbytes": 40960,
471 "sha256": "e287ec9669cc9f498315565b3403ddcfc66d2d74494a044fc588cb11e26f18ca"
472 },
473 {
474 "name": "frame_blocks.3.net.2.bias",
475 "shape": [
476 64
477 ],
478 "dtype": "float16",
479 "offset_bytes": 378638,
480 "nbytes": 128,
481 "sha256": "750e933f96e710ffc3139c64e9663983258e8928f1424e8e9a7bce7ed7ac3cd2"
482 },
483 {
484 "name": "frame_blocks.3.net.2.weight",
485 "shape": [
486 64,
487 64,
488 5
489 ],
490 "dtype": "float16",
491 "offset_bytes": 378766,
492 "nbytes": 40960,
493 "sha256": "0225a89a2162ad5009302704af8e0954b63b435ebf72a350093b8734a6447468"
494 },
495 {
496 "name": "frame_blocks.3.scale",
497 "shape": [
498 1
499 ],
500 "dtype": "float16",
501 "offset_bytes": 419726,
502 "nbytes": 2,
503 "sha256": "b26968d357f99068dd2d31bcf952bcc5d653e178510b4c527bd43d9747b530e1"
504 },
505 {
506 "name": "frame_input_proj.bias",
507 "shape": [
508 64
509 ],
510 "dtype": "float16",
511 "offset_bytes": 419728,
512 "nbytes": 128,
513 "sha256": "b512aaf1f6449094b33d52632bebd6ee5ad6d49d5b5294af22d1209356295438"
514 },
515 {
516 "name": "frame_input_proj.weight",
517 "shape": [
518 64,
519 67,
520 1
521 ],
522 "dtype": "float16",
523 "offset_bytes": 419856,
524 "nbytes": 8576,
525 "sha256": "bec26b8cee2483c328230701d49129acb1ade5e4a3b57ea12935cf8243a9b61d"
526 },
527 {
528 "name": "output.bias",
529 "shape": [
530 192
531 ],
532 "dtype": "float16",
533 "offset_bytes": 428432,
534 "nbytes": 384,
535 "sha256": "16b3058b3c9f7269dbacfb6e5c6a33c75c7ef262903c308a695cd3b7c182d745"
536 },
537 {
538 "name": "output.weight",
539 "shape": [
540 192,
541 64,
542 1
543 ],
544 "dtype": "float16",
545 "offset_bytes": 428816,
546 "nbytes": 24576,
547 "sha256": "2b5d4e8c9cae417359d12bf7e687dc0857a78cfde1822989d0515b7ae443dd6d"
548 },
549 {
550 "name": "token_blocks.0.net.0.bias",
551 "shape": [
552 64
553 ],
554 "dtype": "float16",
555 "offset_bytes": 453392,
556 "nbytes": 128,
557 "sha256": "1210ca84d7a3ed4b7043fe3300b475a3cbd88e46ce5ab34a06adf1075fc76e81"
558 },
559 {
560 "name": "token_blocks.0.net.0.weight",
561 "shape": [
562 64,
563 64,
564 5
565 ],
566 "dtype": "float16",
567 "offset_bytes": 453520,
568 "nbytes": 40960,
569 "sha256": "5542f272daa36288b07dfeefd77792eeca361c04e5a5754b6a2f2ac1190b59d6"
570 },
571 {
572 "name": "token_blocks.0.net.2.bias",
573 "shape": [
574 64
575 ],
576 "dtype": "float16",
577 "offset_bytes": 494480,
578 "nbytes": 128,
579 "sha256": "1b6f8907647ac150f8fbc374028c20774844bd5663f34656b56114f0e410222d"
580 },
581 {
582 "name": "token_blocks.0.net.2.weight",
583 "shape": [
584 64,
585 64,
586 5
587 ],
588 "dtype": "float16",
589 "offset_bytes": 494608,
590 "nbytes": 40960,
591 "sha256": "37f26492ac8814aabb45ac5875434cd9eaf4298d7b0d26b723bd1668394c9cd4"
592 },
593 {
594 "name": "token_blocks.0.scale",
595 "shape": [
596 1
597 ],
598 "dtype": "float16",
599 "offset_bytes": 535568,
600 "nbytes": 2,
601 "sha256": "76a8c8c02bf83b7a32554e070f81dcef3cba91eff7b50773334b2297820293d0"
602 },
603 {
604 "name": "token_blocks.1.net.0.bias",
605 "shape": [
606 64
607 ],
608 "dtype": "float16",
609 "offset_bytes": 535570,
610 "nbytes": 128,
611 "sha256": "feac1ed80a895dc92009e2f27624ad16a6eba305e09356b77f369a78dfb00a27"
612 },
613 {
614 "name": "token_blocks.1.net.0.weight",
615 "shape": [
616 64,
617 64,
618 5
619 ],
620 "dtype": "float16",
621 "offset_bytes": 535698,
622 "nbytes": 40960,
623 "sha256": "046d07c0fcff2df91dde2c377d4edd7dcb73e5589bc86235e8819002bbc5d3fa"
624 },
625 {
626 "name": "token_blocks.1.net.2.bias",
627 "shape": [
628 64
629 ],
630 "dtype": "float16",
631 "offset_bytes": 576658,
632 "nbytes": 128,
633 "sha256": "ff40bcc3b49ca5031fbfe3119ae04058083371f5639147a249671769de80708b"
634 },
635 {
636 "name": "token_blocks.1.net.2.weight",
637 "shape": [
638 64,
639 64,
640 5
641 ],
642 "dtype": "float16",
643 "offset_bytes": 576786,
644 "nbytes": 40960,
645 "sha256": "10da2dc95c6c2de5a640db3e45d8cb1734a2a09865cc0fb5c42eb58adc767e39"
646 },
647 {
648 "name": "token_blocks.1.scale",
649 "shape": [
650 1
651 ],
652 "dtype": "float16",
653 "offset_bytes": 617746,
654 "nbytes": 2,
655 "sha256": "a80190875742dde022b60d2314d51bb17d5c9d010f207ada388541f1ee7b1979"
656 },
657 {
658 "name": "token_blocks.2.net.0.bias",
659 "shape": [
660 64
661 ],
662 "dtype": "float16",
663 "offset_bytes": 617748,
664 "nbytes": 128,
665 "sha256": "902e2e1b41e83ff4be9bcfd52c79602ecbe473c89a1e0e24080d30cd052d6e10"
666 },
667 {
668 "name": "token_blocks.2.net.0.weight",
669 "shape": [
670 64,
671 64,
672 5
673 ],
674 "dtype": "float16",
675 "offset_bytes": 617876,
676 "nbytes": 40960,
677 "sha256": "00e8fcd019e5fa62816138e59ee44c187d1fda4554c6afe47acdf42efd070eea"
678 },
679 {
680 "name": "token_blocks.2.net.2.bias",
681 "shape": [
682 64
683 ],
684 "dtype": "float16",
685 "offset_bytes": 658836,
686 "nbytes": 128,
687 "sha256": "4a712ae2a4e9784f3a5740b50c481550a8373037e651714acc92eb9019f20eb3"
688 },
689 {
690 "name": "token_blocks.2.net.2.weight",
691 "shape": [
692 64,
693 64,
694 5
695 ],
696 "dtype": "float16",
697 "offset_bytes": 658964,
698 "nbytes": 40960,
699 "sha256": "9ef7f1c31f052686a1cc01229d45e35533b62d27e39ea2337e19fc6c6fdbe8f9"
700 },
701 {
702 "name": "token_blocks.2.scale",
703 "shape": [
704 1
705 ],
706 "dtype": "float16",
707 "offset_bytes": 699924,
708 "nbytes": 2,
709 "sha256": "cb84947587ece86ac7a1e2439da49679363df9a9a65318e3d6575c66c50529d5"
710 },
711 {
712 "name": "token_input_proj.bias",
713 "shape": [
714 64
715 ],
716 "dtype": "float16",
717 "offset_bytes": 699926,
718 "nbytes": 128,
719 "sha256": "8d9225187b387623ef2fb7139e0acd03a5d16ff7dd3af7f0fac4d4406b057c5c"
720 },
721 {
722 "name": "token_input_proj.weight",
723 "shape": [
724 64,
725 66,
726 1
727 ],
728 "dtype": "float16",
729 "offset_bytes": 700054,
730 "nbytes": 8448,
731 "sha256": "e533de19df63ff612473941beff7e7089b13174298539df6e0ff65b3de22cd8f"
732 }
733 ]
734 },
735 "decoder": {
736 "checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-amy-medium-retrain-20260712/e3-lite731-ac64-joint/decoder-student.pt",
737 "checkpoint_sha256": "0f4e293d39de9f8c89fa963ff072b8c5fac4633751eaa8660556ba50e9f62e10",
738 "config": {
739 "in_channels": 192,
740 "source_in_channels": 192,
741 "channels": [
742 160,
743 80,
744 40,
745 20
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-amy-medium-retrain-20260712/a91b-decoder-10k/decoder-student.pt",
786 "decoder_init": {
787 "checkpoint": "artifacts/sub10m-search/root-a-piper-vits/en_US-amy-medium-retrain-20260712/a91b-decoder-10k/decoder-student.pt",
788 "decoder_parameters": 730721,
789 "matched_config": {
790 "in_channels": 192,
791 "channels": [
792 160,
793 80,
794 40,
795 20
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-amy-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": 730721,
900 "tensors": [
901 {
902 "name": "post.bias",
903 "shape": [
904 1
905 ],
906 "dtype": "float16",
907 "offset_bytes": 708502,
908 "nbytes": 2,
909 "sha256": "23e33d4e759ac6e120abd79ad0aad87d9cfa83551a3a2f8055fa034f1d2b52c1"
910 },
911 {
912 "name": "post.weight",
913 "shape": [
914 1,
915 20,
916 7
917 ],
918 "dtype": "float16",
919 "offset_bytes": 708504,
920 "nbytes": 280,
921 "sha256": "527270e9d160651f441c9ed8a1527ce19c19e76758e70974aa19614b580a5526"
922 },
923 {
924 "name": "pre.bias",
925 "shape": [
926 160
927 ],
928 "dtype": "float16",
929 "offset_bytes": 708784,
930 "nbytes": 320,
931 "sha256": "950430dd4dc48e41f5e264f4975b2bdce8f565a10cf68ff287295c07fddb70ea"
932 },
933 {
934 "name": "pre.weight",
935 "shape": [
936 160,
937 192,
938 7
939 ],
940 "dtype": "float16",
941 "offset_bytes": 709104,
942 "nbytes": 430080,
943 "sha256": "180e9d4d4e2ef44c0ab012e4357260328c53cb94222f67fa4023c707707d950c"
944 },
945 {
946 "name": "res0.0.blocks.0.conv1.bias",
947 "shape": [
948 80
949 ],
950 "dtype": "float16",
951 "offset_bytes": 1139184,
952 "nbytes": 160,
953 "sha256": "6e86de47484f12de33bab08bdcef69be2907c8f9feeafe524b66c78c2b1b7aa8"
954 },
955 {
956 "name": "res0.0.blocks.0.conv1.weight",
957 "shape": [
958 80,
959 80,
960 3
961 ],
962 "dtype": "float16",
963 "offset_bytes": 1139344,
964 "nbytes": 38400,
965 "sha256": "3b481c63aa93f766d84a6481548d4c5a95c8233223a8e31665bf76fcb365ba22"
966 },
967 {
968 "name": "res0.0.blocks.0.conv2.bias",
969 "shape": [
970 80
971 ],
972 "dtype": "float16",
973 "offset_bytes": 1177744,
974 "nbytes": 160,
975 "sha256": "cc0fc46f15d39e9989602be46f434244b407152f1f527236064974af238cc1df"
976 },
977 {
978 "name": "res0.0.blocks.0.conv2.weight",
979 "shape": [
980 80,
981 80,
982 3
983 ],
984 "dtype": "float16",
985 "offset_bytes": 1177904,
986 "nbytes": 38400,
987 "sha256": "11da25e31180f2cae135a75630926912942b8cf4bf9ecf7f56e6e9bff5a59cf8"
988 },
989 {
990 "name": "res0.0.blocks.1.conv1.bias",
991 "shape": [
992 80
993 ],
994 "dtype": "float16",
995 "offset_bytes": 1216304,
996 "nbytes": 160,
997 "sha256": "26885fb06e084d16062712599d598925e784321671ba50b5097bc3d49a978ec9"
998 },
999 {
1000 "name": "res0.0.blocks.1.conv1.weight",
1001 "shape": [
1002 80,
1003 80,
1004 5
1005 ],
1006 "dtype": "float16",
1007 "offset_bytes": 1216464,
1008 "nbytes": 64000,
1009 "sha256": "6225e5cbfe25e1bcab81d26fb3691c123e45615e0d018cc356facafd4b7f76cb"
1010 },
1011 {
1012 "name": "res0.0.blocks.1.conv2.bias",
1013 "shape": [
1014 80
1015 ],
1016 "dtype": "float16",
1017 "offset_bytes": 1280464,
1018 "nbytes": 160,
1019 "sha256": "27fcb1b4591ebaf2795dc9267a2a59328f3d37a43c79bedbeae901b417f302a8"
1020 },
1021 {
1022 "name": "res0.0.blocks.1.conv2.weight",
1023 "shape": [
1024 80,
1025 80,
1026 5
1027 ],
1028 "dtype": "float16",
1029 "offset_bytes": 1280624,
1030 "nbytes": 64000,
1031 "sha256": "cd8956b051105bc548a82eb7d5b3ddcde0a24cf56e31ae34dfd064019429e2fe"
1032 },
1033 {
1034 "name": "res0.0.blocks.2.conv1.bias",
1035 "shape": [
1036 80
1037 ],
1038 "dtype": "float16",
1039 "offset_bytes": 1344624,
1040 "nbytes": 160,
1041 "sha256": "9fab6ca1c4b1d3be2fed79550ac680ba7a74f6e5dc0939950431cf6144419e5c"
1042 },
1043 {
1044 "name": "res0.0.blocks.2.conv1.weight",
1045 "shape": [
1046 80,
1047 80,
1048 7
1049 ],
1050 "dtype": "float16",
1051 "offset_bytes": 1344784,
1052 "nbytes": 89600,
1053 "sha256": "018ffb3041d85616216a620007f96924d59f009ac91c7957b4cd4db3b62ecd4b"
1054 },
1055 {
1056 "name": "res0.0.blocks.2.conv2.bias",
1057 "shape": [
1058 80
1059 ],
1060 "dtype": "float16",
1061 "offset_bytes": 1434384,
1062 "nbytes": 160,
1063 "sha256": "ff6721e2d183e2961bad78222345e3170ceb1ce86710746241f9f94ac631e04c"
1064 },
1065 {
1066 "name": "res0.0.blocks.2.conv2.weight",
1067 "shape": [
1068 80,
1069 80,
1070 7
1071 ],
1072 "dtype": "float16",
1073 "offset_bytes": 1434544,
1074 "nbytes": 89600,
1075 "sha256": "c2323d17d37ee6550f35c7ad2c8f44072944cbd471637cdd0a2f9bf33b6929ea"
1076 },
1077 {
1078 "name": "res1.0.blocks.0.conv1.bias",
1079 "shape": [
1080 40
1081 ],
1082 "dtype": "float16",
1083 "offset_bytes": 1524144,
1084 "nbytes": 80,
1085 "sha256": "68f5fd1726118c0adbc59aeb1ad903c784d95d1b0171f385bbff5593a56dcec8"
1086 },
1087 {
1088 "name": "res1.0.blocks.0.conv1.weight",
1089 "shape": [
1090 40,
1091 40,
1092 3
1093 ],
1094 "dtype": "float16",
1095 "offset_bytes": 1524224,
1096 "nbytes": 9600,
1097 "sha256": "fe486b030fab9f79849c43b971cc8d35e8ef1f9dcf5b2b6c8c2519c582ae9b63"
1098 },
1099 {
1100 "name": "res1.0.blocks.0.conv2.bias",
1101 "shape": [
1102 40
1103 ],
1104 "dtype": "float16",
1105 "offset_bytes": 1533824,
1106 "nbytes": 80,
1107 "sha256": "d498a1375e0a2317b939dc0f98ede4b4c8aeec6854954b2cdb1463c116beb61e"
1108 },
1109 {
1110 "name": "res1.0.blocks.0.conv2.weight",
1111 "shape": [
1112 40,
1113 40,
1114 3
1115 ],
1116 "dtype": "float16",
1117 "offset_bytes": 1533904,
1118 "nbytes": 9600,
1119 "sha256": "00e819d692007d5943d2ec446014df319128ec33ee178a6de7424418b67ce127"
1120 },
1121 {
1122 "name": "res1.0.blocks.1.conv1.bias",
1123 "shape": [
1124 40
1125 ],
1126 "dtype": "float16",
1127 "offset_bytes": 1543504,
1128 "nbytes": 80,
1129 "sha256": "df350b2f6ba664eff88d92b427300d29c1063b8ed7aa6dbe696300bd82dfd88a"
1130 },
1131 {
1132 "name": "res1.0.blocks.1.conv1.weight",
1133 "shape": [
1134 40,
1135 40,
1136 5
1137 ],
1138 "dtype": "float16",
1139 "offset_bytes": 1543584,
1140 "nbytes": 16000,
1141 "sha256": "81190bd5c228a87094577f9650fe1a505179645be0b0dd5d8c9349f0b04fcf58"
1142 },
1143 {
1144 "name": "res1.0.blocks.1.conv2.bias",
1145 "shape": [
1146 40
1147 ],
1148 "dtype": "float16",
1149 "offset_bytes": 1559584,
1150 "nbytes": 80,
1151 "sha256": "dc9aeef5f5f532210218e76db546889a4f231f540cab5b8d42ecb3e76dee58ae"
1152 },
1153 {
1154 "name": "res1.0.blocks.1.conv2.weight",
1155 "shape": [
1156 40,
1157 40,
1158 5
1159 ],
1160 "dtype": "float16",
1161 "offset_bytes": 1559664,
1162 "nbytes": 16000,
1163 "sha256": "1b18bf5cd4fb29cf01c989c6f2103dfff9e318745ee436df133ea86de646d4fb"
1164 },
1165 {
1166 "name": "res1.0.blocks.2.conv1.bias",
1167 "shape": [
1168 40
1169 ],
1170 "dtype": "float16",
1171 "offset_bytes": 1575664,
1172 "nbytes": 80,
1173 "sha256": "ef993f068a0a8310c0da94730dc3faa1560d38feee179ab0a1b80dc6fbc2277a"
1174 },
1175 {
1176 "name": "res1.0.blocks.2.conv1.weight",
1177 "shape": [
1178 40,
1179 40,
1180 7
1181 ],
1182 "dtype": "float16",
1183 "offset_bytes": 1575744,
1184 "nbytes": 22400,
1185 "sha256": "12e995436981a0685d4923136e7c0c58f1c0d65cb9d816b6683d3aec50c0f875"
1186 },
1187 {
1188 "name": "res1.0.blocks.2.conv2.bias",
1189 "shape": [
1190 40
1191 ],
1192 "dtype": "float16",
1193 "offset_bytes": 1598144,
1194 "nbytes": 80,
1195 "sha256": "9e0517aceea430a84256c50c9a3ec6f4786729a353c1eb7f2ae31f7aec4556da"
1196 },
1197 {
1198 "name": "res1.0.blocks.2.conv2.weight",
1199 "shape": [
1200 40,
1201 40,
1202 7
1203 ],
1204 "dtype": "float16",
1205 "offset_bytes": 1598224,
1206 "nbytes": 22400,
1207 "sha256": "1050dd4d1488a04002f3089994d0542ea146493cd9850f07c2e613da5229683a"
1208 },
1209 {
1210 "name": "res2.0.blocks.0.conv1.bias",
1211 "shape": [
1212 20
1213 ],
1214 "dtype": "float16",
1215 "offset_bytes": 1620624,
1216 "nbytes": 40,
1217 "sha256": "076b8f3dba7197f4973798d2854dd6e9cf6a51c56151374176e64e4b93f4a60e"
1218 },
1219 {
1220 "name": "res2.0.blocks.0.conv1.weight",
1221 "shape": [
1222 20,
1223 20,
1224 3
1225 ],
1226 "dtype": "float16",
1227 "offset_bytes": 1620664,
1228 "nbytes": 2400,
1229 "sha256": "87a7e7f57b66cefc6b80d559aaf64322bd4122f46ae8b30a8733ff4c8a403e53"
1230 },
1231 {
1232 "name": "res2.0.blocks.0.conv2.bias",
1233 "shape": [
1234 20
1235 ],
1236 "dtype": "float16",
1237 "offset_bytes": 1623064,
1238 "nbytes": 40,
1239 "sha256": "16d6672419329a2d1ff05eb2fe43bfc0e58ba95b2c6c621edc3b8dd8b7953497"
1240 },
1241 {
1242 "name": "res2.0.blocks.0.conv2.weight",
1243 "shape": [
1244 20,
1245 20,
1246 3
1247 ],
1248 "dtype": "float16",
1249 "offset_bytes": 1623104,
1250 "nbytes": 2400,
1251 "sha256": "3ef3add11c10f63ffff4a1f7cea6b4e77888c9171db301fbfeaf8e72290c2571"
1252 },
1253 {
1254 "name": "res2.0.blocks.1.conv1.bias",
1255 "shape": [
1256 20
1257 ],
1258 "dtype": "float16",
1259 "offset_bytes": 1625504,
1260 "nbytes": 40,
1261 "sha256": "f24840db324d91389479cc9d7b29351c89b6d2cd9eef1d6605bfa8a207a577b0"
1262 },
1263 {
1264 "name": "res2.0.blocks.1.conv1.weight",
1265 "shape": [
1266 20,
1267 20,
1268 5
1269 ],
1270 "dtype": "float16",
1271 "offset_bytes": 1625544,
1272 "nbytes": 4000,
1273 "sha256": "d8fd5b255460101b7a6e3e22658bbcc60463349f8382fb6016de4fd9e0636c86"
1274 },
1275 {
1276 "name": "res2.0.blocks.1.conv2.bias",
1277 "shape": [
1278 20
1279 ],
1280 "dtype": "float16",
1281 "offset_bytes": 1629544,
1282 "nbytes": 40,
1283 "sha256": "3f101e7032725eb5ed1978988a547b208c4c38a57e2891a2c3632c087cb38ac1"
1284 },
1285 {
1286 "name": "res2.0.blocks.1.conv2.weight",
1287 "shape": [
1288 20,
1289 20,
1290 5
1291 ],
1292 "dtype": "float16",
1293 "offset_bytes": 1629584,
1294 "nbytes": 4000,
1295 "sha256": "2621dcf02182e19297c7b3397bdcde85635c26d479d0060d93c0d405cc4fbe7f"
1296 },
1297 {
1298 "name": "res2.0.blocks.2.conv1.bias",
1299 "shape": [
1300 20
1301 ],
1302 "dtype": "float16",
1303 "offset_bytes": 1633584,
1304 "nbytes": 40,
1305 "sha256": "645399f7b965444e80e0940936f7760c977cb9a5e64285353c2b49671d638f99"
1306 },
1307 {
1308 "name": "res2.0.blocks.2.conv1.weight",
1309 "shape": [
1310 20,
1311 20,
1312 7
1313 ],
1314 "dtype": "float16",
1315 "offset_bytes": 1633624,
1316 "nbytes": 5600,
1317 "sha256": "eea5719c23fb5052eba402d14c42a7803570ac5b3fbbb8e3856b296067dec862"
1318 },
1319 {
1320 "name": "res2.0.blocks.2.conv2.bias",
1321 "shape": [
1322 20
1323 ],
1324 "dtype": "float16",
1325 "offset_bytes": 1639224,
1326 "nbytes": 40,
1327 "sha256": "57e4076d88f650d100b2ae749a37ee2e46941a7b8f37ad157513089be595dc6a"
1328 },
1329 {
1330 "name": "res2.0.blocks.2.conv2.weight",
1331 "shape": [
1332 20,
1333 20,
1334 7
1335 ],
1336 "dtype": "float16",
1337 "offset_bytes": 1639264,
1338 "nbytes": 5600,
1339 "sha256": "5dcda2a32aa5275cbad69231095a94361fef82259b489f0bdb3d036a3f9ed449"
1340 },
1341 {
1342 "name": "up0.bias",
1343 "shape": [
1344 80
1345 ],
1346 "dtype": "float16",
1347 "offset_bytes": 1644864,
1348 "nbytes": 160,
1349 "sha256": "78d204195b03fa0c36a5e65cd12e0572d8e1be01ee59f431258c02ff6e2ed1a7"
1350 },
1351 {
1352 "name": "up0.weight",
1353 "shape": [
1354 160,
1355 80,
1356 16
1357 ],
1358 "dtype": "float16",
1359 "offset_bytes": 1645024,
1360 "nbytes": 409600,
1361 "sha256": "79e06da0175af3661c0181bbe7434269117c43f3740e866e0380e32adc6f1b6a"
1362 },
1363 {
1364 "name": "up1.bias",
1365 "shape": [
1366 40
1367 ],
1368 "dtype": "float16",
1369 "offset_bytes": 2054624,
1370 "nbytes": 80,
1371 "sha256": "c16a2a0bbe30a108ab7110f3fa68c7f990ba292a42be6e4b323c695f04e844f6"
1372 },
1373 {
1374 "name": "up1.weight",
1375 "shape": [
1376 80,
1377 40,
1378 16
1379 ],
1380 "dtype": "float16",
1381 "offset_bytes": 2054704,
1382 "nbytes": 102400,
1383 "sha256": "d2c62ae24c8344de62abfd2d07c4b439f87f8e703fcff053e2fca1126ab4b5b3"
1384 },
1385 {
1386 "name": "up2.bias",
1387 "shape": [
1388 20
1389 ],
1390 "dtype": "float16",
1391 "offset_bytes": 2157104,
1392 "nbytes": 40,
1393 "sha256": "31e404c496558d4ec29af3e8aa9cb201391666d3b86aff68121ac9c7f546a5a5"
1394 },
1395 {
1396 "name": "up2.weight",
1397 "shape": [
1398 40,
1399 20,
1400 8
1401 ],
1402 "dtype": "float16",
1403 "offset_bytes": 2157144,
1404 "nbytes": 12800,
1405 "sha256": "93cd0f24dcf9b37b31bc44f632a835f4431ce638dbf68f497b7edb740b105a68"
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": 154,
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": 2174826,
1443 "target_max_parameters": 1500000,
1444 "under_parameter_budget": true,
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