Add DiffRhythm: TTS/DiffRhythm/MuQ-MuLan-large/config.json
Browse files
TTS/DiffRhythm/MuQ-MuLan-large/config.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"mulan": {
|
| 3 |
+
"sr": 24000,
|
| 4 |
+
"clip_secs": 10,
|
| 5 |
+
"dim_latent": 512,
|
| 6 |
+
"decoupled_contrastive_learning": true,
|
| 7 |
+
"hierarchical_contrastive_loss": false,
|
| 8 |
+
"hierarchical_contrastive_loss_layers": null,
|
| 9 |
+
"sigmoid_contrastive_loss": false,
|
| 10 |
+
"rank_contrast": true
|
| 11 |
+
},
|
| 12 |
+
"audio_model": {
|
| 13 |
+
"name": "OpenMuQ/MuQ-large-msd-iter",
|
| 14 |
+
"model_dim": 1024,
|
| 15 |
+
"use_layer_idx": -1
|
| 16 |
+
},
|
| 17 |
+
"text_model": {
|
| 18 |
+
"name": "xlm-roberta-base",
|
| 19 |
+
"model_dim": null,
|
| 20 |
+
"use_layer_idx": -1
|
| 21 |
+
},
|
| 22 |
+
"audio_transformer": {
|
| 23 |
+
"dim": 768,
|
| 24 |
+
"tf_depth": 0,
|
| 25 |
+
"heads": 8,
|
| 26 |
+
"dim_head": 64,
|
| 27 |
+
"attn_dropout": 0,
|
| 28 |
+
"ff_dropout": 0,
|
| 29 |
+
"ff_mult": 4
|
| 30 |
+
},
|
| 31 |
+
"text_transformer": {
|
| 32 |
+
"dim": 768,
|
| 33 |
+
"tf_depth": 8,
|
| 34 |
+
"max_seq_len": 1024,
|
| 35 |
+
"dim_head": 64,
|
| 36 |
+
"heads": 8,
|
| 37 |
+
"attn_dropout": 0,
|
| 38 |
+
"ff_dropout": 0,
|
| 39 |
+
"ff_mult": 4
|
| 40 |
+
}
|
| 41 |
+
}
|