{
  "assamese": {
    "n_samples": 500,
    "wer_raw": 36.17,
    "wer_norm": 35.85,
    "wer_numcanon": 35.85,
    "space_norm_wer": 31.38,
    "mer": 17.42,
    "cer_norm": 16.37,
    "empty_hypotheses": 2,
    "normalization_delta": {
      "raw_to_norm": 0.31,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 4.47,
      "norm_to_mer": 18.43
    }
  },
  "bengali": {
    "n_samples": 500,
    "wer_raw": 34.98,
    "wer_norm": 34.25,
    "wer_numcanon": 34.25,
    "space_norm_wer": 26.0,
    "mer": 10.14,
    "cer_norm": 10.09,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.73,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 8.26,
      "norm_to_mer": 24.11
    }
  },
  "english": {
    "n_samples": 500,
    "wer_raw": 28.9,
    "wer_norm": 15.43,
    "wer_numcanon": 15.43,
    "space_norm_wer": 10.4,
    "mer": 9.56,
    "cer_norm": 9.5,
    "empty_hypotheses": 16,
    "normalization_delta": {
      "raw_to_norm": 13.48,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 5.03,
      "norm_to_mer": 5.87
    }
  },
  "gujarati": {
    "n_samples": 500,
    "wer_raw": 27.36,
    "wer_norm": 27.3,
    "wer_numcanon": 27.3,
    "space_norm_wer": 21.17,
    "mer": 8.58,
    "cer_norm": 8.45,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.05,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 6.13,
      "norm_to_mer": 18.72
    }
  },
  "hindi": {
    "n_samples": 500,
    "wer_raw": 19.66,
    "wer_norm": 19.63,
    "wer_numcanon": 19.63,
    "space_norm_wer": 14.73,
    "mer": 7.09,
    "cer_norm": 6.71,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.03,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 4.9,
      "norm_to_mer": 12.54
    }
  },
  "kannada": {
    "n_samples": 500,
    "wer_raw": 42.75,
    "wer_norm": 42.62,
    "wer_numcanon": 42.62,
    "space_norm_wer": 32.23,
    "mer": 9.81,
    "cer_norm": 10.06,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.13,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.39,
      "norm_to_mer": 32.81
    }
  },
  "malayalam": {
    "n_samples": 500,
    "wer_raw": 48.95,
    "wer_norm": 48.36,
    "wer_numcanon": 48.36,
    "space_norm_wer": 37.64,
    "mer": 10.02,
    "cer_norm": 10.03,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.59,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.72,
      "norm_to_mer": 38.35
    }
  },
  "marathi": {
    "n_samples": 500,
    "wer_raw": 40.75,
    "wer_norm": 40.71,
    "wer_numcanon": 40.71,
    "space_norm_wer": 31.11,
    "mer": 12.03,
    "cer_norm": 12.06,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.04,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 9.6,
      "norm_to_mer": 28.68
    }
  },
  "odia": {
    "n_samples": 500,
    "wer_raw": 36.76,
    "wer_norm": 36.5,
    "wer_numcanon": 36.5,
    "space_norm_wer": 25.95,
    "mer": 10.51,
    "cer_norm": 10.64,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.26,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.56,
      "norm_to_mer": 25.99
    }
  },
  "punjabi": {
    "n_samples": 500,
    "wer_raw": 25.62,
    "wer_norm": 25.57,
    "wer_numcanon": 25.57,
    "space_norm_wer": 19.5,
    "mer": 8.13,
    "cer_norm": 7.83,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.05,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 6.06,
      "norm_to_mer": 17.44
    }
  },
  "tamil": {
    "n_samples": 500,
    "wer_raw": 45.41,
    "wer_norm": 44.92,
    "wer_numcanon": 44.92,
    "space_norm_wer": 35.48,
    "mer": 10.84,
    "cer_norm": 10.84,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.48,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 9.45,
      "norm_to_mer": 34.08
    }
  },
  "telugu": {
    "n_samples": 500,
    "wer_raw": 34.49,
    "wer_norm": 33.56,
    "wer_numcanon": 33.56,
    "space_norm_wer": 23.9,
    "mer": 7.06,
    "cer_norm": 7.28,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.93,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 9.66,
      "norm_to_mer": 26.5
    }
  },
  "__overall__": {
    "n_samples": 6000,
    "wer_raw": 35.03,
    "wer_norm": 33.68,
    "wer_numcanon": 33.68,
    "space_norm_wer": 25.87,
    "mer": 10.26,
    "cer_norm": 10.17
  },
  "__macro_avg__": {
    "n_languages": 12,
    "wer_raw": 35.15,
    "wer_norm": 33.73,
    "wer_numcanon": 33.73,
    "space_norm_wer": 25.79,
    "mer": 10.1,
    "cer_norm": 9.99
  },
  "__meta__": {
    "checkpoint": "/home/ubuntu/training/checkpoints/maya-asr-tdt-1.1b-ckpt-120000/model.ckpt",
    "checkpoint_name": "ckpt-120000",
    "model_id": "parakeet-tdt-1.1b-lang",
    "model_type": "maya-asr-hybrid-tdt-1.1b",
    "dataset": "BayAreaBoys/indic-asr-benchmark-6k",
    "batch_size": 32,
    "inference_time_sec": 350.69,
    "total_audio_sec": 40354.46,
    "rtf": 0.0087,
    "timestamp": "2026-03-29T06:52:04Z",
    "gpu": "NVIDIA A100-SXM4-80GB",
    "framework": "nemo",
    "normalization_version": "v1",
    "jiwer_version": "unknown"
  }
}