{
    "framework": "pytorch",
    "task": "acoustic-noise-suppression",
    "pipeline": {
       "type": "speech_zipenhancer_ans_multiloss_16k_base"
    },
    "model": {
      "type": "speech_zipenhancer_ans_multiloss_16k_base",
      "dense_channel": 64,
      "num_tsconformers": 4,
      "former_name": "Zipformer2DualPathEncoder",
      "former_conf": {
          "num_encoder_layers": [1, 1, 1, 1],
          "downsampling_factor": [1, 2, 2, 1],
          "f_downsampling_factor": [1, 2, 2, 1],
          "encoder_dim": [64, 64, 64, 64],
          "pos_dim": 24,
          "num_heads": 4,
          "query_head_dim": 12,
          "pos_head_dim": 4,
          "value_head_dim": 8,
          "feedforward_dim": [256, 256, 256, 256],
          "cnn_module_kernel": 15,
          "causal": false,
          "encoder_unmasked_dim": 64,
          "warmup_batches": 4000.0
      },
          "batch_first": true,
      "model_num_spks": 1
    },
    "preprocessor": {}
}