{
  "assamese": {
    "n_samples": 500,
    "wer_raw": 35.0,
    "wer_norm": 34.66,
    "wer_numcanon": 34.66,
    "space_norm_wer": 30.09,
    "mer": 16.28,
    "cer_norm": 15.23,
    "empty_hypotheses": 1,
    "normalization_delta": {
      "raw_to_norm": 0.34,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 4.57,
      "norm_to_mer": 18.38
    }
  },
  "bengali": {
    "n_samples": 500,
    "wer_raw": 34.54,
    "wer_norm": 33.81,
    "wer_numcanon": 33.81,
    "space_norm_wer": 25.68,
    "mer": 9.86,
    "cer_norm": 9.81,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.73,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 8.13,
      "norm_to_mer": 23.95
    }
  },
  "english": {
    "n_samples": 500,
    "wer_raw": 28.27,
    "wer_norm": 15.19,
    "wer_numcanon": 15.19,
    "space_norm_wer": 10.49,
    "mer": 9.62,
    "cer_norm": 9.44,
    "empty_hypotheses": 22,
    "normalization_delta": {
      "raw_to_norm": 13.09,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 4.7,
      "norm_to_mer": 5.57
    }
  },
  "gujarati": {
    "n_samples": 500,
    "wer_raw": 26.5,
    "wer_norm": 26.41,
    "wer_numcanon": 26.41,
    "space_norm_wer": 20.06,
    "mer": 7.69,
    "cer_norm": 7.62,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.09,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 6.35,
      "norm_to_mer": 18.72
    }
  },
  "hindi": {
    "n_samples": 500,
    "wer_raw": 19.82,
    "wer_norm": 19.78,
    "wer_numcanon": 19.78,
    "space_norm_wer": 14.54,
    "mer": 6.88,
    "cer_norm": 6.61,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.03,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 5.24,
      "norm_to_mer": 12.91
    }
  },
  "kannada": {
    "n_samples": 500,
    "wer_raw": 43.62,
    "wer_norm": 43.44,
    "wer_numcanon": 43.44,
    "space_norm_wer": 32.58,
    "mer": 10.02,
    "cer_norm": 10.29,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.18,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.86,
      "norm_to_mer": 33.42
    }
  },
  "malayalam": {
    "n_samples": 500,
    "wer_raw": 48.93,
    "wer_norm": 48.18,
    "wer_numcanon": 48.18,
    "space_norm_wer": 37.42,
    "mer": 9.84,
    "cer_norm": 9.87,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.75,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.76,
      "norm_to_mer": 38.33
    }
  },
  "marathi": {
    "n_samples": 500,
    "wer_raw": 40.77,
    "wer_norm": 40.71,
    "wer_numcanon": 40.71,
    "space_norm_wer": 31.04,
    "mer": 12.08,
    "cer_norm": 12.18,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.05,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 9.67,
      "norm_to_mer": 28.63
    }
  },
  "odia": {
    "n_samples": 500,
    "wer_raw": 36.22,
    "wer_norm": 35.84,
    "wer_numcanon": 35.84,
    "space_norm_wer": 25.52,
    "mer": 9.74,
    "cer_norm": 9.83,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.39,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.32,
      "norm_to_mer": 26.1
    }
  },
  "punjabi": {
    "n_samples": 500,
    "wer_raw": 25.48,
    "wer_norm": 25.46,
    "wer_numcanon": 25.46,
    "space_norm_wer": 19.09,
    "mer": 7.77,
    "cer_norm": 7.53,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.02,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 6.37,
      "norm_to_mer": 17.69
    }
  },
  "tamil": {
    "n_samples": 500,
    "wer_raw": 45.63,
    "wer_norm": 45.3,
    "wer_numcanon": 45.3,
    "space_norm_wer": 35.26,
    "mer": 10.56,
    "cer_norm": 10.61,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.33,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.04,
      "norm_to_mer": 34.74
    }
  },
  "telugu": {
    "n_samples": 500,
    "wer_raw": 34.89,
    "wer_norm": 33.97,
    "wer_numcanon": 33.97,
    "space_norm_wer": 24.16,
    "mer": 7.05,
    "cer_norm": 7.31,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.93,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 9.81,
      "norm_to_mer": 26.92
    }
  },
  "__overall__": {
    "n_samples": 6000,
    "wer_raw": 34.83,
    "wer_norm": 33.49,
    "wer_numcanon": 33.49,
    "space_norm_wer": 25.54,
    "mer": 9.95,
    "cer_norm": 9.88
  },
  "__macro_avg__": {
    "n_languages": 12,
    "wer_raw": 34.97,
    "wer_norm": 33.56,
    "wer_numcanon": 33.56,
    "space_norm_wer": 25.49,
    "mer": 9.78,
    "cer_norm": 9.69
  },
  "__meta__": {
    "checkpoint": "/home/ubuntu/training/checkpoints/maya-asr-tdt-1.1b-ckpt-200000/model.ckpt",
    "checkpoint_name": "ckpt-200000",
    "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": 343.3,
    "total_audio_sec": 40354.46,
    "rtf": 0.0085,
    "timestamp": "2026-03-29T07:00:17Z",
    "gpu": "NVIDIA A100-SXM4-80GB",
    "framework": "nemo",
    "normalization_version": "v1",
    "jiwer_version": "unknown"
  }
}