| { | |
| "vocoder": { | |
| "resblock_dilation_sizes": [ | |
| [ | |
| 1, | |
| 3, | |
| 5 | |
| ], | |
| [ | |
| 1, | |
| 3, | |
| 5 | |
| ], | |
| [ | |
| 1, | |
| 3, | |
| 5 | |
| ] | |
| ], | |
| "stereo": true, | |
| "upsample_rates": [ | |
| 6, | |
| 5, | |
| 2, | |
| 2, | |
| 2 | |
| ], | |
| "resblock_kernel_sizes": [ | |
| 3, | |
| 7, | |
| 11 | |
| ], | |
| "upsample_kernel_sizes": [ | |
| 16, | |
| 15, | |
| 8, | |
| 4, | |
| 4 | |
| ], | |
| "resblock": "1", | |
| "upsample_initial_channel": 1024 | |
| }, | |
| "_class_name": "LTX2Vocoder" | |
| } | |