Upload folder using huggingface_hub
Browse files- added_tokens.json +3 -0
- config.json +82 -0
- openvino_model.bin +3 -0
- openvino_model.xml +0 -0
- special_tokens_map.json +4 -0
- tokenizer_config.json +30 -0
- vocab.json +40 -0
added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<unk>": 38
|
| 3 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "facebook/mms-tts-eng",
|
| 3 |
+
"activation_dropout": 0.1,
|
| 4 |
+
"architectures": [
|
| 5 |
+
"VitsModel"
|
| 6 |
+
],
|
| 7 |
+
"attention_dropout": 0.1,
|
| 8 |
+
"depth_separable_channels": 2,
|
| 9 |
+
"depth_separable_num_layers": 3,
|
| 10 |
+
"duration_predictor_dropout": 0.5,
|
| 11 |
+
"duration_predictor_filter_channels": 256,
|
| 12 |
+
"duration_predictor_flow_bins": 10,
|
| 13 |
+
"duration_predictor_kernel_size": 3,
|
| 14 |
+
"duration_predictor_num_flows": 4,
|
| 15 |
+
"duration_predictor_tail_bound": 5.0,
|
| 16 |
+
"ffn_dim": 768,
|
| 17 |
+
"ffn_kernel_size": 3,
|
| 18 |
+
"flow_size": 192,
|
| 19 |
+
"hidden_act": "relu",
|
| 20 |
+
"hidden_dropout": 0.1,
|
| 21 |
+
"hidden_size": 192,
|
| 22 |
+
"initializer_range": 0.02,
|
| 23 |
+
"layer_norm_eps": 1e-05,
|
| 24 |
+
"layerdrop": 0.1,
|
| 25 |
+
"leaky_relu_slope": 0.1,
|
| 26 |
+
"model_type": "vits",
|
| 27 |
+
"noise_scale": 0.667,
|
| 28 |
+
"noise_scale_duration": 0.8,
|
| 29 |
+
"num_attention_heads": 2,
|
| 30 |
+
"num_hidden_layers": 6,
|
| 31 |
+
"num_speakers": 1,
|
| 32 |
+
"posterior_encoder_num_wavenet_layers": 16,
|
| 33 |
+
"prior_encoder_num_flows": 4,
|
| 34 |
+
"prior_encoder_num_wavenet_layers": 4,
|
| 35 |
+
"resblock_dilation_sizes": [
|
| 36 |
+
[
|
| 37 |
+
1,
|
| 38 |
+
3,
|
| 39 |
+
5
|
| 40 |
+
],
|
| 41 |
+
[
|
| 42 |
+
1,
|
| 43 |
+
3,
|
| 44 |
+
5
|
| 45 |
+
],
|
| 46 |
+
[
|
| 47 |
+
1,
|
| 48 |
+
3,
|
| 49 |
+
5
|
| 50 |
+
]
|
| 51 |
+
],
|
| 52 |
+
"resblock_kernel_sizes": [
|
| 53 |
+
3,
|
| 54 |
+
7,
|
| 55 |
+
11
|
| 56 |
+
],
|
| 57 |
+
"sampling_rate": 16000,
|
| 58 |
+
"speaker_embedding_size": 0,
|
| 59 |
+
"speaking_rate": 1.0,
|
| 60 |
+
"spectrogram_bins": 513,
|
| 61 |
+
"transformers_version": "4.41.2",
|
| 62 |
+
"upsample_initial_channel": 512,
|
| 63 |
+
"upsample_kernel_sizes": [
|
| 64 |
+
16,
|
| 65 |
+
16,
|
| 66 |
+
4,
|
| 67 |
+
4
|
| 68 |
+
],
|
| 69 |
+
"upsample_rates": [
|
| 70 |
+
8,
|
| 71 |
+
8,
|
| 72 |
+
2,
|
| 73 |
+
2
|
| 74 |
+
],
|
| 75 |
+
"use_bias": true,
|
| 76 |
+
"use_stochastic_duration_prediction": true,
|
| 77 |
+
"vocab_size": 38,
|
| 78 |
+
"wavenet_dilation_rate": 1,
|
| 79 |
+
"wavenet_dropout": 0.0,
|
| 80 |
+
"wavenet_kernel_size": 5,
|
| 81 |
+
"window_size": 4
|
| 82 |
+
}
|
openvino_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cca040c8475a0e941cbf890dfcd33cb7856fc2561047073be11a42d8f461ab93
|
| 3 |
+
size 113093124
|
openvino_model.xml
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"pad_token": "k",
|
| 3 |
+
"unk_token": "<unk>"
|
| 4 |
+
}
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_blank": true,
|
| 3 |
+
"added_tokens_decoder": {
|
| 4 |
+
"0": {
|
| 5 |
+
"content": "k",
|
| 6 |
+
"lstrip": false,
|
| 7 |
+
"normalized": false,
|
| 8 |
+
"rstrip": false,
|
| 9 |
+
"single_word": false,
|
| 10 |
+
"special": true
|
| 11 |
+
},
|
| 12 |
+
"38": {
|
| 13 |
+
"content": "<unk>",
|
| 14 |
+
"lstrip": false,
|
| 15 |
+
"normalized": false,
|
| 16 |
+
"rstrip": false,
|
| 17 |
+
"single_word": false,
|
| 18 |
+
"special": true
|
| 19 |
+
}
|
| 20 |
+
},
|
| 21 |
+
"clean_up_tokenization_spaces": true,
|
| 22 |
+
"is_uroman": false,
|
| 23 |
+
"language": "eng",
|
| 24 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 25 |
+
"normalize": true,
|
| 26 |
+
"pad_token": "k",
|
| 27 |
+
"phonemize": false,
|
| 28 |
+
"tokenizer_class": "VitsTokenizer",
|
| 29 |
+
"unk_token": "<unk>"
|
| 30 |
+
}
|
vocab.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
" ": 19,
|
| 3 |
+
"'": 1,
|
| 4 |
+
"-": 14,
|
| 5 |
+
"0": 23,
|
| 6 |
+
"1": 15,
|
| 7 |
+
"2": 28,
|
| 8 |
+
"3": 11,
|
| 9 |
+
"4": 27,
|
| 10 |
+
"5": 35,
|
| 11 |
+
"6": 36,
|
| 12 |
+
"_": 30,
|
| 13 |
+
"a": 26,
|
| 14 |
+
"b": 24,
|
| 15 |
+
"c": 12,
|
| 16 |
+
"d": 5,
|
| 17 |
+
"e": 7,
|
| 18 |
+
"f": 20,
|
| 19 |
+
"g": 37,
|
| 20 |
+
"h": 6,
|
| 21 |
+
"i": 18,
|
| 22 |
+
"j": 16,
|
| 23 |
+
"k": 0,
|
| 24 |
+
"l": 21,
|
| 25 |
+
"m": 17,
|
| 26 |
+
"n": 29,
|
| 27 |
+
"o": 22,
|
| 28 |
+
"p": 13,
|
| 29 |
+
"q": 34,
|
| 30 |
+
"r": 25,
|
| 31 |
+
"s": 8,
|
| 32 |
+
"t": 33,
|
| 33 |
+
"u": 4,
|
| 34 |
+
"v": 32,
|
| 35 |
+
"w": 9,
|
| 36 |
+
"x": 31,
|
| 37 |
+
"y": 3,
|
| 38 |
+
"z": 2,
|
| 39 |
+
"–": 10
|
| 40 |
+
}
|