Upload 2 files

config.json

@@ -1,9 +1,9 @@
 {
   "train": {
-    "log_interval": 500,
-    "eval_interval": 500,
+    "log_interval": 1000,
+    "eval_interval": 1000,
     "seed": 42,
-    "epochs": 1000,
+    "epochs": 10000,
     "learning_rate": 0.0001,
     "betas": [
       0.8,
@@ -11,14 +11,19 @@
     ],
     "eps": 1e-09,
     "batch_size": 8,
-    "fp16_run": false,
+    "bf16_run": false,
     "lr_decay": 0.99995,
     "segment_size": 16384,
     "init_lr_ratio": 1,
     "warmup_epochs": 0,
     "c_mel": 45,
     "c_kl": 1.0,
-    "skip_optimizer": true
+    "c_commit": 100,
+    "skip_optimizer": true,
+    "freeze_ZH_bert": false,
+    "freeze_JP_bert": false,
+    "freeze_EN_bert": false,
+    "freeze_emo": false
   },
   "data": {
     "training_files": "Data/glut/filelists/train.list",
@@ -32,14 +37,13 @@
     "mel_fmin": 0.0,
     "mel_fmax": null,
     "add_blank": true,
-    "n_speakers": 896,
+    "n_speakers": 4,
     "cleaned_text": true,
     "spk2id": {
       "16": 0,
       "G": 1,
       "M": 2,
-      "S": 3,
-      "Y": 4
+      "Y": 3
     }
   },
   "model": {
@@ -94,7 +98,14 @@
     ],
     "n_layers_q": 3,
     "use_spectral_norm": false,
-    "gin_channels": 256
+    "gin_channels": 512,
+    "slm": {
+      "model": "./slm/wavlm-base-plus",
+      "sr": 16000,
+      "hidden": 768,
+      "nlayers": 13,
+      "initial_channel": 64
+    }
   },
-  "version": "2.1"
+  "version": "2.3"
 }

config.yml

@@ -3,11 +3,12 @@ bert_gen:
   device: cuda
   num_processes: 2
   use_multi_device: false
-dataset_path: Data/glut
+dataset_path: Data\glut
 emo_gen:
   config_path: config.json
   device: cuda
   num_processes: 2
+  use_multi_device: false
 mirror: ''
 openi_token: ''
 preprocess_text:
@@ -18,23 +19,43 @@ preprocess_text:
   train_path: filelists/train.list
   transcription_path: filelists/short_character_anno.list
   val_path: filelists/val.list
-  val_per_spk: 4
+  val_per_lang: 4
 resample:
-  in_dir: custom_character_voice
-  out_dir: custom_character_voice
+  in_dir: raw
+  out_dir: wavs
   sampling_rate: 44100
 server:
   device: cuda
   models:
-  - config: Data/glut/config.json
+  - config: ''
     device: cuda
     language: ZH
-    model: Data/glut/models/G_1000.pth
+    model: ''
+    speakers:
+    - length_scale: 1
+      noise_scale: 0.6
+      noise_scale_w: 0.8
+      sdp_ratio: 0.2
+      speaker: "\u79D1\u6BD4"
+    - length_scale: 0.5
+      noise_scale: 0.7
+      noise_scale_w: 0.8
+      sdp_ratio: 0.3
+      speaker: "\u4E94\u6761\u609F"
+    - length_scale: 1.2
+      noise_scale: 0.6
+      noise_scale_w: 0.8
+      sdp_ratio: 0.2
+      speaker: "\u5B89\u500D\u664B\u4E09"
+  - config: ''
+    device: cpu
+    language: JP
+    model: ''
     speakers: []
   port: 5000
 train_ms:
   base:
-    model_image: "Bert-VITS2_2.1-Emo\u5E95\u6A21"
+    model_image: "Bert-VITS2_2.3\u5E95\u6A21"
     repo_id: Stardust_minus/Bert-VITS2
     use_base_model: false
   config_path: config.json
@@ -46,16 +67,16 @@ train_ms:
     WORLD_SIZE: 1
   keep_ckpts: 10
   model: models
-  num_workers: 4
+  num_workers: 2
   spec_cache: true
 translate:
   app_key: ''
   secret_key: ''
 webui:
-  config_path: configs/config.json
+  config_path: config.json
   debug: false
   device: cuda
   language_identification_library: langid
-  model: genshin/models/G_8000.pth
+  model: models/G_8000.pth
   port: 7860
   share: false