{
  "dataset": {
    "bpe_model": "multilingual_zh_ja_yue_char_del.tiktoken",
    "mel": {
      "hop_length": 256,
      "mel_fmin": 0,
      "n_fft": 1024,
      "n_mels": 100,
      "normalize": false,
      "sample_rate": 24000,
      "win_length": 1024
    },
    "sample_rate": 24000,
    "squeeze": false,
    "tokenizer_type": "tiktoken"
  },
  "emo_matrix": "feat2.pt",
  "emo_num": [
    3,
    17,
    2,
    8,
    4,
    5,
    10,
    24
  ],
  "gpt": {
    "condition_module": {
      "attention_heads": 8,
      "input_layer": "conv2d2",
      "linear_units": 2048,
      "num_blocks": 6,
      "output_size": 512,
      "perceiver_mult": 2
    },
    "condition_type": "conformer_perceiver",
    "emo_condition_module": {
      "attention_heads": 4,
      "input_layer": "conv2d2",
      "linear_units": 1024,
      "num_blocks": 4,
      "output_size": 512,
      "perceiver_mult": 2
    },
    "heads": 20,
    "layers": 24,
    "max_mel_tokens": 1815,
    "max_text_tokens": 600,
    "mel_length_compression": 1024,
    "model_dim": 1280,
    "number_mel_codes": 8194,
    "number_text_tokens": 60509,
    "start_mel_token": 8192,
    "start_text_token": 0,
    "stop_mel_token": 8193,
    "stop_text_token": 1,
    "train_solo_embeddings": false,
    "use_mel_codes_as_input": true
  },
  "gpt_checkpoint": "checkpoint_5000.pth",
  "qwen_emo_path": "qwen0.6bemo4-merge/",
  "s2mel": {
    "DiT": {
      "add_resblock_in_transformer": false,
      "block_size": 8192,
      "class_dropout_prob": 0.1,
      "content_codebook_size": 1024,
      "content_codebooks": 1,
      "content_dim": 512,
      "content_type": "discrete",
      "depth": 13,
      "f0_condition": false,
      "final_layer_type": "wavenet",
      "hidden_dim": 512,
      "in_channels": 80,
      "is_causal": false,
      "long_skip_connection": true,
      "n_f0_bins": 512,
      "num_heads": 8,
      "style_as_token": false,
      "style_condition": true,
      "target": "mel",
      "time_as_token": false,
      "uvit_skip_connection": true,
      "zero_prompt_speech_token": false
    },
    "dit_type": "DiT",
    "length_regulator": {
      "channels": 512,
      "content_codebook_size": 2048,
      "f0_condition": false,
      "in_channels": 1024,
      "is_discrete": false,
      "n_codebooks": 1,
      "n_f0_bins": 512,
      "quantizer_dropout": 0.0,
      "sampling_ratios": [
        1,
        1,
        1,
        1
      ],
      "vector_quantize": false
    },
    "preprocess_params": {
      "spect_params": {
        "fmax": "None",
        "fmin": 0,
        "hop_length": 256,
        "n_fft": 1024,
        "n_mels": 80,
        "win_length": 1024
      },
      "sr": 22050
    },
    "reg_loss_type": "l1",
    "style_encoder": {
      "dim": 192
    },
    "wavenet": {
      "dilation_rate": 1,
      "hidden_dim": 512,
      "kernel_size": 5,
      "num_layers": 8,
      "p_dropout": 0.2,
      "style_condition": true
    }
  },
  "s2mel_checkpoint": "DiT_gptlatent_10000.pth",
  "semantic_codec": {
    "codebook_dim": 8,
    "codebook_size": 8192,
    "frame_rate": 25,
    "hidden_size": 1024,
    "vocos_dim": 384,
    "vocos_intermediate_dim": 2048,
    "vocos_num_layers": 12
  },
  "spk_matrix": "feat1.pt",
  "supported_languages": [
    "zh",
    "en",
    "ja",
    "es",
    "ar"
  ],
  "version": 2.5,
  "vocoder": {
    "name": "bigvgan_generator.pt",
    "type": "bigvgan"
  },
  "w2v_stat": "wav2vec2bert_stats.pt"
}