thanks to sh-lee-prml ❤

generator_discriminator_v1.1/D_1890000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:227fde67840990fa3f4540b34ce592add20896b966612a3a3571d89418d5707e
+size 499174277

generator_discriminator_v1.1/G_1890000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6cdc3b97b511d287caa6492d0112821f5e82937db623e8bb512b8b8edb718924
+size 1166678943

hierspeechpp_eng_kor/config.json

@@ -0,0 +1,67 @@
+{
+    "train": {
+      "log_interval": 200,
+      "eval_interval": 10000,
+      "save_interval": 10000,
+      "seed": 1234,
+      "epochs": 20000,
+      "learning_rate": 1e-4,
+      "betas": [0.8, 0.99],
+      "eps": 1e-9,
+      "batch_size": 20,
+      "fp16_run": false,
+      "lr_decay": 0.999,
+      "segment_size": 61440,
+      "init_lr_ratio": 1,
+      "warmup_epochs": 0,
+      "c_mel": 45,
+      "c_kl": 1.0,
+      "c_bi_kl": 0.5,
+      "c_mixup": 1,
+      "c_pho": 45.0,
+      "c_f0": 45,
+      "aug": true
+    },
+    "data": {
+      "train_filelist_path": "filelists/train_eng_kor_wav.txt",
+      "test_filelist_path": "filelists/test_hier_16k_72_wav.txt",
+      "text_cleaners":["english_cleaners2"],
+      "max_wav_value": 32768.0,
+      "sampling_rate": 16000,
+      "filter_length": 1280,
+      "hop_length": 320,
+      "win_length": 1280,
+      "n_mel_channels": 80,
+      "mel_fmin": 0,
+      "mel_fmax": 8000,
+      "add_blank": true,
+      "n_speakers": 0,
+      "cleaned_text": true,
+      "aug_rate": 1.0,
+      "top_db": 20
+    },
+    "model": {
+      "inter_channels": 192,
+      "hidden_channels": 192,
+      "filter_channels": 768,
+      "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": [4,5,4,2,2],
+      "upsample_initial_channel": 512,
+      "upsample_kernel_sizes": [8,11,8,4,4],
+      "n_layers_q": 3,
+      "use_spectral_norm": false,
+      "wav2vec_feature_layer": 7,
+      "gin_channels": 256,
+      "cfg": true,
+      "uncond_ratio": 0.1,
+      "prosody_size": 20
+  
+    }
+  }
+  

hierspeechpp_eng_kor/hierspeechpp_v1.1_ckpt.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fd704f9a4d11b62195caaaeb76ce759b12ea69a1c36b2716af44bd60bde6a64
+size 388999163

hierspeechpp_eng_kor/hierspeechpp_v1_ckpt.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0229501cd273135bce9ee0be09ddef18e079cb8dbe61cda665fec73afbfad0d4
+size 388970075

hierspeechpp_libritts460/config.json

@@ -0,0 +1,67 @@
+{
+    "train": {
+      "log_interval": 200,
+      "eval_interval": 10000,
+      "save_interval": 10000,
+      "seed": 1234,
+      "epochs": 20000,
+      "learning_rate": 1e-4,
+      "betas": [0.8, 0.99],
+      "eps": 1e-9,
+      "batch_size": 16,
+      "fp16_run": false,
+      "lr_decay": 0.999875,
+      "segment_size": 61440,
+      "init_lr_ratio": 1,
+      "warmup_epochs": 0,
+      "c_mel": 45,
+      "c_kl": 1.0,
+      "c_bi_kl": 0.5,
+      "c_mixup": 1,
+      "c_pho": 45.0,
+      "c_f0": 45,
+      "aug": true
+    },
+    "data": {
+      "train_filelist_path": "filelists/train_libritts_460_wav.txt",
+      "test_filelist_path": "filelists/test_hier_16k_72_wav.txt",
+      "text_cleaners":["english_cleaners2"],
+      "max_wav_value": 32768.0,
+      "sampling_rate": 16000,
+      "filter_length": 1280,
+      "hop_length": 320,
+      "win_length": 1280,
+      "n_mel_channels": 80,
+      "mel_fmin": 0,
+      "mel_fmax": 8000,
+      "add_blank": true,
+      "n_speakers": 0,
+      "cleaned_text": true,
+      "aug_rate": 1.0,
+      "top_db": 20
+    },
+    "model": {
+      "inter_channels": 192,
+      "hidden_channels": 192,
+      "filter_channels": 768,
+      "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": [4,5,4,2,2],
+      "upsample_initial_channel": 512,
+      "upsample_kernel_sizes": [8,11,8,4,4],
+      "n_layers_q": 3,
+      "use_spectral_norm": false,
+      "wav2vec_feature_layer": 7,
+      "gin_channels": 256,
+      "cfg": true,
+      "uncond_ratio": 0.1,
+      "prosody_size": 20
+  
+    }
+  }
+  

hierspeechpp_libritts460/hierspeechpp_lt460_ckpt.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f6928ad968454061b30fbeb024a4bbd9bfcadfd5e2660c213b95ab83fa3f171
+size 389002475

hierspeechpp_libritts960/config.json

@@ -0,0 +1,67 @@
+{
+    "train": {
+      "log_interval": 200,
+      "eval_interval": 10000,
+      "save_interval": 10000,
+      "seed": 1234,
+      "epochs": 20000,
+      "learning_rate": 1e-4,
+      "betas": [0.8, 0.99],
+      "eps": 1e-9,
+      "batch_size": 16,
+      "fp16_run": false,
+      "lr_decay": 0.999875,
+      "segment_size": 61440,
+      "init_lr_ratio": 1,
+      "warmup_epochs": 0,
+      "c_mel": 45,
+      "c_kl": 1.0,
+      "c_bi_kl": 0.5,
+      "c_mixup": 1,
+      "c_pho": 45.0,
+      "c_f0": 45,
+      "aug": true
+    },
+    "data": {
+      "train_filelist_path": "filelists/train_libritts_all_wav.txt",
+      "test_filelist_path": "filelists/test_libritts_all_wav.txt",
+      "text_cleaners":["english_cleaners2"],
+      "max_wav_value": 32768.0,
+      "sampling_rate": 16000,
+      "filter_length": 1280,
+      "hop_length": 320,
+      "win_length": 1280,
+      "n_mel_channels": 80,
+      "mel_fmin": 0,
+      "mel_fmax": 8000,
+      "add_blank": true,
+      "n_speakers": 0,
+      "cleaned_text": true,
+      "aug_rate": 1.0,
+      "top_db": 20
+    },
+    "model": {
+      "inter_channels": 192,
+      "hidden_channels": 192,
+      "filter_channels": 768,
+      "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": [4,5,4,2,2],
+      "upsample_initial_channel": 512,
+      "upsample_kernel_sizes": [8,11,8,4,4],
+      "n_layers_q": 3,
+      "use_spectral_norm": false,
+      "wav2vec_feature_layer": 7,
+      "gin_channels": 256,
+      "cfg": true,
+      "uncond_ratio": 0.1,
+      "prosody_size": 20
+  
+    }
+  }
+  

hierspeechpp_libritts960/hierspeechpp_lt960_ckpt.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:609465feba7cee4a1dd8e25e456459cb5ca0f4cb00d97fb1f7007e422d1a6fd6
+size 389002475

ttv_libritts_v1/config.json

@@ -0,0 +1,55 @@
+{
+    "train": {
+      "log_interval": 200,
+      "eval_interval": 10000,
+      "save_interval": 10000,
+      "seed": 1234,
+      "epochs": 20000,
+      "learning_rate": 2e-4,
+      "betas": [0.8, 0.99],
+      "eps": 1e-9,
+      "batch_size": 32,
+      "fp16_run": false,
+      "lr_decay": 0.999,
+      "segment_size": 192000,
+      "init_lr_ratio": 1,
+      "warmup_epochs": 0,
+      "c_mel": 10,
+      "c_kl": 0.1,
+      "c_pho": 45.0,
+      "c_f0": 1,
+      "aug": true
+    },
+    "data": {
+      "train_filelist_path": "filelists/train_wav.txt",
+      "test_filelist_path": "filelists/test_wav.txt",
+      "text_cleaners":["english_cleaners2"],
+      "max_wav_value": 32768.0,
+      "sampling_rate": 16000,
+      "filter_length": 1280,
+      "hop_length": 320,
+      "win_length": 1280,
+      "n_mel_channels": 80,
+      "mel_fmin": 0,
+      "mel_fmax": 8000,
+      "add_blank": true,
+      "n_speakers": 0,
+      "cleaned_text": true,
+      "aug_rate": 1.0,
+      "top_db": 20
+    },
+    "model": {
+        "inter_channels": 256,
+        "hidden_channels": 256,
+        "filter_channels": 1024,
+        "n_heads": 4,
+        "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]],
+        "use_spectral_norm": false
+      }
+  }
+  

ttv_libritts_v1/ttv_lt960_ckpt.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e408e2c6a72551da0ca616f43e0f989449ff5e0d0ca6229b27dea3dc36aa6c1
+size 434482782