{
  "model_name": "nemotron-speech-streaming-en-0.6b",
  "sample_rate": 16000,
  "n_mels": 128,
  "subsampling_factor": 8,
  "att_context_size": [
    70,
    6
  ],
  "left_context": 70,
  "right_context": 6,
  "chunk_size_output_frames": 7,
  "drop_extra_pre_encoded": 2,
  "num_encoder_layers": 24,
  "hidden_dim": 1024,
  "conv_context": 8,
  "vocab_size": 1024,
  "blank_id": 1024,
  "preprocessor": {
    "window_size": 0.025,
    "window_stride": 0.01,
    "n_fft": 512,
    "normalize": "per_feature",
    "preemph": 0.97
  },
  "cache_shapes": {
    "cache_last_channel": [
      24,
      1,
      70,
      1024
    ],
    "cache_last_time": [
      24,
      1,
      1024,
      8
    ],
    "cache_last_channel_len": [
      1
    ]
  },
  "test_input": {
    "mel_shape": [
      1,
      128,
      65
    ],
    "mel_length": 65
  },
  "test_output": {
    "encoded_shape": [
      1,
      1024,
      7
    ],
    "encoded_len": 7
  }
}