Push model using huggingface_hub.
Browse files- README.md +12 -0
- config.json +153 -0
- model.safetensors +3 -0
README.md
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: voicecraft
|
3 |
+
tags:
|
4 |
+
- text-to-speech
|
5 |
+
- pytorch_model_hub_mixin
|
6 |
+
- model_hub_mixin
|
7 |
+
repo_url: https://github.com/jasonppy/VoiceCraft
|
8 |
+
---
|
9 |
+
|
10 |
+
This model has been pushed to the Hub using **voicecraft**:
|
11 |
+
- Repo: https://github.com/jasonppy/VoiceCraft
|
12 |
+
- Docs: [More Information Needed]
|
config.json
ADDED
@@ -0,0 +1,153 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"args": null,
|
3 |
+
"audio_embedding_dim": 1024,
|
4 |
+
"audio_embedding_dropout": 0,
|
5 |
+
"audio_max_length": 20.0,
|
6 |
+
"audio_min_length": 2.0,
|
7 |
+
"audio_pad_token": 1026,
|
8 |
+
"audio_positional_embedding_dropout": 0.1,
|
9 |
+
"audio_vocab_size": "1024",
|
10 |
+
"batch_size": 100,
|
11 |
+
"clipping_update_period": 1000,
|
12 |
+
"codebook_weight": "[5,2,1,0.5,0.2,0.1,0.1,0.1]",
|
13 |
+
"d_model": 1024,
|
14 |
+
"dataset": "gigaspeech",
|
15 |
+
"dataset_dir": "/data/scratch/pyp/datasets/gigaspeech_phn_enc_manifest/xl",
|
16 |
+
"drop_long": 1,
|
17 |
+
"dynamic_batching": 1,
|
18 |
+
"early_stop_step": 3200,
|
19 |
+
"early_stop_threshold": -1.0,
|
20 |
+
"empty_token": 1024,
|
21 |
+
"encodec_folder_name": "encodec_d3cd8c2f_8cb1024_noise_epoch94",
|
22 |
+
"encodec_sr": 50,
|
23 |
+
"eog": 1025,
|
24 |
+
"eos": 1027,
|
25 |
+
"exp_dir": "/data/scratch/pyp/exp_pyp/VoiceCraft/gigaspeech/615M_8cb1024_se",
|
26 |
+
"gradient_accumulation_steps": 24,
|
27 |
+
"gradient_clip_val": 1.0,
|
28 |
+
"load_model_from": null,
|
29 |
+
"lr": 0.05,
|
30 |
+
"manifest_name": "manifest_large16khz_lessambi",
|
31 |
+
"mask_len_max": 600,
|
32 |
+
"mask_len_min": 1,
|
33 |
+
"mask_sample_dist": "poisson1",
|
34 |
+
"max_mask_portion": 0.9,
|
35 |
+
"max_n_spans": 3,
|
36 |
+
"max_num_tokens": 100000,
|
37 |
+
"min_gap": 5,
|
38 |
+
"n_codebooks": 8,
|
39 |
+
"n_special": 4,
|
40 |
+
"nhead": 16,
|
41 |
+
"num_buckets": 6,
|
42 |
+
"num_decoder_layers": 48,
|
43 |
+
"num_epochs": 10,
|
44 |
+
"num_steps": 50000,
|
45 |
+
"num_workers": 8,
|
46 |
+
"optimizer_name": "ScaledAdam",
|
47 |
+
"pad_x": 0,
|
48 |
+
"phn2num": {
|
49 |
+
"!": 17,
|
50 |
+
",": 64,
|
51 |
+
".": 77,
|
52 |
+
"<MUSIC>": 39,
|
53 |
+
"<NOISE>": 52,
|
54 |
+
"<OTHER>": 60,
|
55 |
+
"<SIL>": 53,
|
56 |
+
"?": 78,
|
57 |
+
"_": 15,
|
58 |
+
"a\u026a": 48,
|
59 |
+
"a\u026a\u0259": 56,
|
60 |
+
"a\u026a\u025a": 2,
|
61 |
+
"a\u028a": 36,
|
62 |
+
"b": 20,
|
63 |
+
"d": 72,
|
64 |
+
"d\u0292": 57,
|
65 |
+
"e\u026a": 6,
|
66 |
+
"f": 69,
|
67 |
+
"h": 14,
|
68 |
+
"i": 27,
|
69 |
+
"i\u0259": 42,
|
70 |
+
"i\u02d0": 68,
|
71 |
+
"i\u02d0\u02d0": 51,
|
72 |
+
"j": 67,
|
73 |
+
"k": 41,
|
74 |
+
"l": 63,
|
75 |
+
"m": 9,
|
76 |
+
"n": 23,
|
77 |
+
"n\u02b2": 8,
|
78 |
+
"o\u028a": 25,
|
79 |
+
"o\u02d0": 74,
|
80 |
+
"o\u02d0\u0279": 40,
|
81 |
+
"p": 34,
|
82 |
+
"r": 79,
|
83 |
+
"s": 66,
|
84 |
+
"t": 73,
|
85 |
+
"t\u0283": 75,
|
86 |
+
"u": 1,
|
87 |
+
"u\u02d0": 47,
|
88 |
+
"v": 31,
|
89 |
+
"w": 19,
|
90 |
+
"x": 4,
|
91 |
+
"z": 22,
|
92 |
+
"\u00e6": 32,
|
93 |
+
"\u00e6\u00e6": 50,
|
94 |
+
"\u00e7": 10,
|
95 |
+
"\u00f0": 7,
|
96 |
+
"\u014b": 58,
|
97 |
+
"\u0250": 70,
|
98 |
+
"\u0250\u0250": 71,
|
99 |
+
"\u0251": 61,
|
100 |
+
"\u0251\u02d0": 0,
|
101 |
+
"\u0251\u02d0\u0279": 44,
|
102 |
+
"\u0254": 3,
|
103 |
+
"\u0254\u026a": 13,
|
104 |
+
"\u0254\u02d0": 29,
|
105 |
+
"\u0254\u02d0\u0279": 33,
|
106 |
+
"\u0259": 54,
|
107 |
+
"\u0259l": 16,
|
108 |
+
"\u025a": 35,
|
109 |
+
"\u025b": 18,
|
110 |
+
"\u025b\u0279": 11,
|
111 |
+
"\u025c\u02d0": 21,
|
112 |
+
"\u0261": 49,
|
113 |
+
"\u0261\u02b2": 37,
|
114 |
+
"\u026a": 65,
|
115 |
+
"\u026a\u0279": 76,
|
116 |
+
"\u026c": 46,
|
117 |
+
"\u0279": 5,
|
118 |
+
"\u027e": 24,
|
119 |
+
"\u0283": 26,
|
120 |
+
"\u028a": 43,
|
121 |
+
"\u028a\u0279": 28,
|
122 |
+
"\u028c": 38,
|
123 |
+
"\u0292": 55,
|
124 |
+
"\u0294": 59,
|
125 |
+
"\u0303": 45,
|
126 |
+
"\u0329": 12,
|
127 |
+
"\u03b8": 30,
|
128 |
+
"\u1d7b": 62
|
129 |
+
},
|
130 |
+
"phn_folder_name": "phonemes",
|
131 |
+
"precision": "float16",
|
132 |
+
"print_every_n_steps": 400,
|
133 |
+
"pseudo_epoch_size": 3000,
|
134 |
+
"reduce_lr_start_epoch": 4,
|
135 |
+
"reduce_lr_start_step": 3000,
|
136 |
+
"reduced_eog": 1,
|
137 |
+
"resume": false,
|
138 |
+
"seed": 1,
|
139 |
+
"shuffle_mask_embedding": 0,
|
140 |
+
"special_first": 0,
|
141 |
+
"tb_write_every_n_steps": 100,
|
142 |
+
"text_embedding_dropout": 0.1,
|
143 |
+
"text_max_length": 400,
|
144 |
+
"text_min_length": 10.0,
|
145 |
+
"text_pad_token": 100,
|
146 |
+
"text_positional_embedding_dropout": 0.1,
|
147 |
+
"text_vocab_size": 100,
|
148 |
+
"trm_dropout": 0.1,
|
149 |
+
"val_every_n_steps": 1600,
|
150 |
+
"val_max_num_tokens": 6000,
|
151 |
+
"warmup_fraction": 0.01,
|
152 |
+
"weight_decay": 0.01
|
153 |
+
}
|
model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d4730a85b239b5150bca01fc2e93aa64835583dfc23a1f94ff4a6b4cd8681635
|
3 |
+
size 2486330704
|