init

music_tokenizer/config.json

@@ -1,7 +1,7 @@
 {
     "resblock": "1",
     "num_gpus": 8,
-    "batch_size": 128,
+    "batch_size": 140,
     "learning_rate": 0.00002,
     "adam_b1": 0.5,
     "adam_b2": 0.9,
@@ -11,17 +11,17 @@
     "upsample_rates": [8,5,4,2],
     "upsample_kernel_sizes": [16,11,8,4],
     "upsample_initial_channel": 512,
-    "resblock_kernel_sizes": [3,7,11],
-    "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
+    "resblock_kernel_sizes": [3,5,7,9,11,13],
+    "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5], [1,3,5], [1,3,5], [1,3,5]],
 
-    "segment_size": 24000,
-    "num_mels": 128,
+    "segment_size": 48000,
+    "num_mels": 80,
     "num_freq": 1024,
     "n_fft": 1024,
     "hop_size": 240,
     "win_size": 1024,
 
-    "sampling_rate": 24000,
+    "sampling_rate": 48000,
 
     "n_code_groups": 2,
     "n_codes": 1024,
@@ -29,10 +29,10 @@
     "commitment_loss_lambda": 0.25,
 
     "fmin": 0,
-    "fmax": 12000,
+    "fmax": 48000,
     "fmax_for_loss": null,
 
-    "num_workers": 32,
+    "num_workers": 24,
 
     "dist_config": {
         "dist_backend": "nccl",
@@ -40,4 +40,3 @@
         "world_size": 1
     }
 }
-

music_tokenizer/model.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:08d26564f725b5b4a38d0c9ea3d8d5e7df42173f15eeb4453930c2bfb93ef783
-size 254691363
+oid sha256:ba71efdc50378baf9776d607eb11566907c3810e6f221c316719c02591135626
+size 537087507