{
  "variant": "nar",
  "topology": "nar",
  "sample_audio": "sample_audio.wav",
  "sample_audio_sha256": "6ee3e432f4ce88415747f0549628bfef0df742365813d38b18bf28067a40cd51",
  "sample_audio_bytes": 539884,
  "sample_rate": 16000,
  "duration_s": 16.87,
  "audio_frontend": {
    "n_fft": 512,
    "hop_length": 160,
    "win_length": 400,
    "n_mels": 80,
    "frame_stack": 2,
    "log_compression": "log10 with -8 dB clamp",
    "preemphasis": "none"
  },
  "input_features_shape": [
    1,
    843,
    160
  ],
  "input_features_dtype": "float32",
  "audio_embeds_shape": [
    1,
    171,
    2048
  ],
  "audio_embeds_dtype": "float32",
  "extra_files": [
    "expected_attention_mask.npy"
  ],
  "tolerance_suggestions": {
    "input_features_vs_processor": {
      "rtol": 0.001,
      "atol": 0.0001
    },
    "audio_embeds_vs_fp32_encoder": {
      "rtol": 0.0001,
      "atol": 1e-05
    }
  },
  "fixture_files": [
    "sample_audio.wav",
    "expected_input_features.npy",
    "expected_attention_mask.npy",
    "expected_audio_embeds.npy"
  ]
}