init

.gitattributes

@@ -0,0 +1,4 @@
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+g_* filter=lfs diff=lfs merge=lfs -text

checkpoints/MVoice/multitask_ratio_g930_A100_CQ/checkpoint_last.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05c2bc0235a9c3681b63fa698980b93bb52015f95ccd543867729be7d6bbb3e7
+size 15077046419

checkpoints/MVoice/multitask_ratio_liblarge_0924/checkpoint_last.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bedaf22e1b72d1e7e115784c1dbc26311c438a29096d572bab25a82fb80099b3
+size 9169919163

checkpoints/tokens/acoustic/ckpt_01135000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52b341c9dc01b019d126d56858261812ae8c0f92a7497d6842057bbef4ff7697
+size 291809077

checkpoints/tokens/acoustic/config.yaml

@@ -0,0 +1,131 @@
+
+########### model config ###########
+generator:
+  name: SoundStream
+  config:
+    n_filters: 32
+    D: 256
+    #target_bandwidths: [6,] # [1, 1.5, 2, 4, 6] # [0.5, 1, 1.5, 2, 4, 6]
+    target_bandwidths: [0.5, 1, 1.5, 2, 4]
+    ratios: [8, 5, 4, 2] # downsampling by 320
+    sample_rate: 16000
+    bins: 1024
+
+# Discriminator list
+#d_list: ['mpd', 'msd', 'mfd']
+d_list: ['mfd']
+
+mfd:
+  name: MultiFrequencyDiscriminator
+  config:
+    hop_lengths: [32, 64, 128, 256, 512, 1024]
+    hidden_channels: [64, 128, 256, 512, 512, 512]
+    domain: double
+    mel_scale: true
+    sample_rate: 16000
+
+mpd:
+  name: MultiPeriodDiscriminator
+  config:
+    period_sizes: [2, 3, 5, 7, 11]
+    period_kernel_size: 5
+
+msd:
+  name: MultiScaleDiscriminator
+  config:
+    num_scales: 3
+    pool_kernel_size: 4 
+    pool_stride: 2
+
+########### optimizer config ###########
+optimizer:
+  g:
+    name: AdamW
+    config:
+      lr: 2e-4
+      betas: [0.8, 0.99]
+      eps: 1.0e-6
+
+  d:
+    name: AdamW
+    config:
+      lr: 2e-4
+      betas: [0.8, 0.99]
+      eps: 1.0e-6
+
+lr_scheduler:
+  g:
+    name: ExponentialLR
+    config:
+      gamma: 0.999
+  d:
+    name: ExponentialLR
+    config:
+      gamma: 0.999
+
+########### criterion config ###########
+criterion:
+  g_criterion:
+    name: losses.generator_loss.GeneratorSTFTLoss
+    config:
+      use_mel_loss: false
+      #adv_criterion: LeastDLoss
+      adv_criterion: MSEGLoss
+      mel_loss_weight: 45
+      use_feature_match: true
+      feat_match_loss_weight: 20
+      use_full_stft_loss: true # Magnitude 
+      use_sub_stft_loss: true  # PQMF loss
+      full_stft_loss_weight: 1
+      sub_stft_loss_weight: 1
+      mel_scale_loss:
+        sampling_rate: 16000
+        n_fft: 1024
+        num_mels: 80
+        hop_size: 160
+        win_size: 800
+        fmin: 0
+      full_multi_scale_stft_loss: # Full-band multi-scale STFT loss.
+        fft_sizes: [512, 1024, 2048]
+        win_sizes: [480, 960, 1200]
+        hop_sizes: [120, 240, 300]
+      sub_multi_scale_stft_loss: # Sub-band multi-scale STFT loss.
+        num_bands: 6
+        fft_sizes: [128, 256, 256]
+        win_sizes: [80, 120, 200]
+        hop_sizes: [20, 40, 50]
+  
+  d_criterion:
+    name: losses.discriminator_loss.MSEDiscriminatorLoss
+    config: null
+
+  commit_loss_weight: 1. #1000
+
+########### training and data config ###########
+
+#training_file: "/apdcephfs_cq2/share_1603164/data/nlphuang/huangrongjie2/data/covost_for_encodec/train.lst"
+#validation_file: "/apdcephfs_cq2/share_1603164/data/nlphuang/huangrongjie2/data/covost_for_encodec_test/valid.lst"
+training_file: "/apdcephfs_cq3/share_1603164/data/nlphuang/data/sing/Voicelm/covost_for_encodec/train.lst"
+validation_file: "/apdcephfs_cq3/share_1603164/data/nlphuang/data/sing/Voicelm/covost_for_encodec/valid.lst"
+
+seed: 2333
+cudnn_deterministic: false
+tensorboard: true # whether to use tensorboard
+#checkpoint_interval: 5
+#summary_interval: 10
+#validation_interval: 10
+
+checkpoint_interval: 5000
+summary_interval: 100
+validation_interval: 5000
+
+num_epoches: 5000
+print_freq: 10
+discriminator_iter_start: 0  # start step after which we update discriminators
+num_ckpt_keep: 10
+
+segment_size: 24000
+audio_norm_scale: 1.0
+batch_size: 6
+num_workers: 8
+num_plots: 8

checkpoints/tokens/semantic/xlsr2_300m.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93d6cbc5543d3724a261dcf9d8ec162da5b52a840324233c0eaae6b1ff147a0f
+size 3808857131

checkpoints/tokens/semantic/xlsr300m_4_0_layer12_km_300h.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e8d6ba94b2b222ca121ade6fdd12d1cde057a6f71ea0bc1d6b4be0677b49d639
+size 4100733

checkpoints/vocoder/sing/config.json

@@ -0,0 +1,49 @@
+{
+    "resblock": "1",
+    "num_gpus": 0,
+    "batch_size": 12,
+    "learning_rate": 0.0001,
+    "adam_b1": 0.8,
+    "adam_b2": 0.99,
+    "lr_decay": 0.999,
+    "seed": 1234,
+    "upsample_rates": [5,4,2,2,2,2],
+    "upsample_kernel_sizes": [9,8,4,4,4,4],
+    "upsample_initial_channel": 1536,
+    "resblock_kernel_sizes": [3,7,11],
+    "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
+
+    "code_num": 1024,
+    "codebook_num": 3,
+    "code_emb_dim": 256,
+    "unit_hop_size": 320,
+
+    "activation": "snakebeta",
+    "snake_logscale": true,
+
+    "resolutions": [[1024, 120, 600], [2048, 240, 1200], [512, 50, 240]],
+    "mpd_reshapes": [2, 3, 5, 7, 11],
+    "use_spectral_norm": false,
+    "discriminator_channel_mult": 1,
+
+    "segment_size": 9600,
+    "num_mels": 80,
+    "num_freq": 1025,
+    "n_fft": 1024,
+    "hop_size": 256,
+    "win_size": 1024,
+
+    "sampling_rate": 16000,
+
+    "fmin": 0,
+    "fmax": 8000,
+    "fmax_for_loss": null,
+
+    "num_workers": 8,
+
+    "dist_config": {
+        "dist_backend": "nccl",
+        "dist_url": "tcp://localhost:54321",
+        "world_size": 1
+    }
+}

checkpoints/vocoder/sing/g_01550000

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:47d9aec16158c8f20e1998806f79e7edd61582c6fc86fe35230f462b8b08aa8f
+size 486674328

checkpoints/vocoder/speech/config.json

@@ -0,0 +1,49 @@
+{
+    "resblock": "1",
+    "num_gpus": 0,
+    "batch_size": 32,
+    "learning_rate": 0.0001,
+    "adam_b1": 0.8,
+    "adam_b2": 0.99,
+    "lr_decay": 0.999,
+    "seed": 1234,
+    "upsample_rates": [5,4,2,2,2,2],
+    "upsample_kernel_sizes": [9,8,4,4,4,4],
+    "upsample_initial_channel": 1536,
+    "resblock_kernel_sizes": [3,7,11],
+    "resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
+
+    "code_num": 1024,
+    "codebook_num": 3,
+    "code_emb_dim": 256,
+    "unit_hop_size": 320,
+
+    "activation": "snakebeta",
+    "snake_logscale": true,
+
+    "resolutions": [[1024, 120, 600], [2048, 240, 1200], [512, 50, 240]],
+    "mpd_reshapes": [2, 3, 5, 7, 11],
+    "use_spectral_norm": false,
+    "discriminator_channel_mult": 1,
+
+    "segment_size": 9600,
+    "num_mels": 80,
+    "num_freq": 1025,
+    "n_fft": 1024,
+    "hop_size": 256,
+    "win_size": 1024,
+
+    "sampling_rate": 16000,
+
+    "fmin": 0,
+    "fmax": 8000,
+    "fmax_for_loss": null,
+
+    "num_workers": 0,
+
+    "dist_config": {
+        "dist_backend": "nccl",
+        "dist_url": "tcp://localhost:54321",
+        "world_size": 1
+    }
+}

checkpoints/vocoder/speech/g_00490000

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19f442fb73276d6c2cd7b24a648d566ac762c3e221470f4c160cb612baec142a
+size 486674328