sovits tiny config

configs_template/config_tiny_template.json

@@ -0,0 +1,77 @@
+{
+  "train": {
+    "log_interval": 200,
+    "eval_interval": 800,
+    "seed": 1234,
+    "epochs": 10000,
+    "learning_rate": 0.0001,
+    "betas": [
+      0.8,
+      0.99
+    ],
+    "eps": 1e-09,
+    "batch_size": 6,
+    "fp16_run": false,
+    "half_type": "fp16",
+    "lr_decay": 0.999875,
+    "segment_size": 10240,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0,
+    "use_sr": true,
+    "max_speclen": 512,
+    "port": "8001",
+    "keep_ckpts": 3,
+    "all_in_mem": false,
+    "vol_aug":false
+  },
+  "data": {
+    "training_files": "filelists/train.txt",
+    "validation_files": "filelists/val.txt",
+    "max_wav_value": 32768.0,
+    "sampling_rate": 44100,
+    "filter_length": 2048,
+    "hop_length": 512,
+    "win_length": 2048,
+    "n_mel_channels": 80,
+    "mel_fmin": 0.0,
+    "mel_fmax": 22050,
+    "unit_interpolate_mode":"nearest"
+  },
+  "model": {
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 512,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.1,
+    "resblock": "1",
+    "resblock_kernel_sizes": [3,7,11],
+    "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
+    "upsample_rates": [ 8, 8, 2, 2, 2],
+    "upsample_initial_channel": 400,
+    "upsample_kernel_sizes": [16,16, 4, 4, 4],
+    "n_layers_q": 3,
+    "n_flow_layer": 4,
+    "use_spectral_norm": false,
+    "gin_channels": 768,
+    "ssl_dim": 768,
+    "n_speakers": 200,
+    "vocoder_name":"nsf-hifigan",
+    "speech_encoder":"vec768l12",
+    "speaker_embedding":false,
+    "vol_embedding":false,
+    "use_depthwise_conv":true,
+    "flow_share_parameter": true,
+    "use_automatic_f0_prediction": true
+  },
+  "spk": {
+    "nyaru": 0,
+    "huiyu": 1,
+    "nen": 2,
+    "paimon": 3,
+    "yunhao": 4
+  }
+}

preprocess_flist_config.py

@@ -30,6 +30,7 @@ if __name__ == "__main__":
     parser.add_argument("--source_dir", type=str, default="./dataset/44k", help="path to source dir")
     parser.add_argument("--speech_encoder", type=str, default="vec768l12", help="choice a speech encoder|'vec768l12','vec256l9','hubertsoft','whisper-ppg','cnhubertlarge','dphubert','whisper-ppg-large','wavlmbase+'")
     parser.add_argument("--vol_aug", action="store_true", help="Whether to use volume embedding and volume augmentation")
+    parser.add_argument("--tiny", action="store_true", help="Whether to train sovits tiny")
     args = parser.parse_args()
     
     train = []
@@ -72,7 +73,7 @@ if __name__ == "__main__":
             f.write(wavpath + "\n")
 
 
-    d_config_template = du.load_config("configs_template/diffusion_template.yaml")
+    d_config_template = du.load_config("configs_template/diffusion_tiny_template.yaml") if args.tiny else du.load_config("configs_template/diffusion_template.yaml")
     d_config_template["model"]["n_spk"] = spk_id
     d_config_template["data"]["encoder"] = args.speech_encoder
     d_config_template["spk"] = spk_dict
@@ -97,6 +98,9 @@ if __name__ == "__main__":
     if args.vol_aug:
         config_template["train"]["vol_aug"] = config_template["model"]["vol_embedding"] = True
 
+    if args.tiny:
+        config_template["model"]["filter_channels"] = 512
+
     print("Writing configs/config.json")
     with open("configs/config.json", "w") as f:
         json.dump(config_template, f, indent=2)