added refinegan pretrain download, re-enabled selection on training UI, adjusted training UI defaults

Author: AznamirWoW

rvc/lib/tools/prerequisites_download.py

@@ -16,7 +16,16 @@
             "f0G40k.pth",
             "f0G48k.pth",
         ],
-    )
+    ),
+]
+pretraineds_refinegan_list = [
+    (
+        "refinegan/",
+        [
+            "f0D32k.pth",
+            "f0G32k.pth",
+        ],
+    ),
 ]
 models_list = [("predictors/", ["rmvpe.pt", "fcpe.pt"])]
 embedders_list = [("embedders/contentvec/", ["pytorch_model.bin", "config.json"])]
@@ -26,6 +35,7 @@
 
 folder_mapping_list = {
     "pretrained_v2/": "rvc/models/pretraineds/hifi-gan/",
+    "refinegan/": "rvc/models/pretraineds/refinegan/",
     "embedders/contentvec/": "rvc/models/embedders/contentvec/",
     "predictors/": "rvc/models/predictors/",
     "formant/": "rvc/models/formant/",
@@ -118,6 +128,7 @@ def calculate_total_size(
     if exe and os.name == "nt":
         total_size += get_file_size_if_missing(executables_list)
     total_size += get_file_size_if_missing(pretraineds_hifigan)
+    total_size += get_file_size_if_missing(pretraineds_refinegan_list)
     return total_size
 
 
@@ -149,5 +160,6 @@ def prequisites_download_pipeline(
                     print("No executables needed")
             if pretraineds_hifigan:
                 download_mapping_files(pretraineds_hifigan_list, global_bar)
+                download_mapping_files(pretraineds_refinegan_list, global_bar)
     else:
         pass

rvc/train/train.py

@@ -70,6 +70,10 @@
 bf16_adamw = False
 disc_version = "v2"
 
+if vocoder == "RefineGAN":
+    disc_version = "v3"
+    multiscale_mel_loss = True
+
 current_dir = os.getcwd()
 
 try:

tabs/train/train.py

@@ -328,7 +328,7 @@ def train_tab():
                     choices=["RVC", "Applio"],
                     value="RVC",
                     interactive=True,
-                    visible=False,  # to be visible once pretraineds are ready
+                    visible=False,
                 )
             with gr.Column():
                 sampling_rate = gr.Radio(
@@ -343,10 +343,10 @@ def train_tab():
                     info=i18n(
                         "Choose the vocoder for audio synthesis:\n- **HiFi-GAN**: Default option, compatible with all clients.\n- **MRF HiFi-GAN**: Higher fidelity, Applio-only.\n- **RefineGAN**: Superior audio quality, Applio-only."
                     ),
-                    choices=["HiFi-GAN", "MRF HiFi-GAN", "RefineGAN"],
+                    choices=["HiFi-GAN", "RefineGAN"], #"MRF HiFi-GAN", ],
                     value="HiFi-GAN",
-                    interactive=False,
-                    visible=False,  # to be visible once pretraineds are ready
+                    interactive=True,
+                    visible=True,
                 )
         with gr.Accordion(
             i18n("Advanced Settings"),
@@ -451,7 +451,7 @@ def train_tab():
                     info=i18n(
                         "It's recommended to deactivate this option if your dataset has already been processed."
                     ),
-                    value=True,
+                    value=False,
                     interactive=True,
                     visible=True,
                 )
@@ -462,7 +462,7 @@ def train_tab():
                         "Audio normalization: Select 'none' if the files are already normalized, 'pre' to normalize the entire input file at once, or 'post' to normalize each slice individually."
                     ),
                     choices=["none", "pre", "post"],
-                    value="none",
+                    value="post",
                     interactive=True,
                     visible=True,
                 )