Upload 2 files

config_dit_mel_seed.yml

@@ -0,0 +1,79 @@
+log_dir: "./runs/run_dit_mel_seed"
+save_freq: 1
+log_interval: 10
+save_interval: 1000
+device: "cuda"
+epochs: 1000 # number of epochs for first stage training (pre-training)
+batch_size: 4
+batch_length: 100 # maximum duration of audio in a batch (in seconds)
+max_len: 80 # maximum number of frames
+pretrained_model: ""
+pretrained_encoder: ""
+load_only_params: False # set to true if do not want to load epoch numbers and optimizer parameters
+
+F0_path: "modules/JDC/bst.t7"
+
+preprocess_params:
+  sr: 22050
+  spect_params:
+    n_fft: 1024
+    win_length: 1024
+    hop_length: 256
+    n_mels: 80
+
+model_params:
+  dit_type: "DiT" # uDiT or DiT
+  reg_loss_type: "l2" # l1 or l2
+
+  speech_tokenizer:
+    path: "speech_tokenizer_v1.onnx"
+
+  style_encoder:
+    dim: 192
+    campplus_path: "campplus_cn_common.bin"
+
+  DAC:
+    encoder_dim: 64
+    encoder_rates: [2, 5, 5, 6]
+    decoder_dim: 1536
+    decoder_rates: [ 6, 5, 5, 2 ]
+    sr: 24000
+
+  length_regulator:
+    channels: 768
+    is_discrete: true
+    content_codebook_size: 4096
+    in_frame_rate: 50
+    out_frame_rate: 80
+    sampling_ratios: [1, 1, 1, 1]
+
+  DiT:
+    hidden_dim: 768
+    num_heads: 12
+    depth: 12
+    class_dropout_prob: 0.1
+    block_size: 4096
+    in_channels: 80
+    style_condition: true
+    final_layer_type: 'wavenet'
+    target: 'mel' # mel or codec
+    content_dim: 768
+    content_codebook_size: 1024
+    content_type: 'discrete'
+    f0_condition: false
+    n_f0_bins: 512
+    content_codebooks: 1
+    is_causal: false
+    long_skip_connection: true
+    zero_prompt_speech_token: false # for prompt component, do not input corresponding speech token
+
+  wavenet:
+    hidden_dim: 768
+    num_layers: 8
+    kernel_size: 5
+    dilation_rate: 1
+    p_dropout: 0.2
+    style_condition: true
+
+loss_params:
+  base_lr: 0.0001

hifigan.yml

@@ -0,0 +1,25 @@
+hift:
+    in_channels: 80
+    base_channels: 512
+    nb_harmonics: 8
+    sampling_rate: 22050
+    nsf_alpha: 0.1
+    nsf_sigma: 0.003
+    nsf_voiced_threshold: 10
+    upsample_rates: [8, 8]
+    upsample_kernel_sizes: [16, 16]
+    istft_params:
+        n_fft: 16
+        hop_len: 4
+    resblock_kernel_sizes: [3, 7, 11]
+    resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5], [1, 3, 5]]
+    source_resblock_kernel_sizes: [7, 11]
+    source_resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5]]
+    lrelu_slope: 0.1
+    audio_limit: 0.99
+f0_predictor:
+    num_class: 1
+    in_channels: 80
+    cond_channels: 512
+
+pretrained_model_path: "hift.pt"