{
  "assamese": {
    "n_samples": 500,
    "wer_raw": 39.24,
    "wer_norm": 39.01,
    "wer_numcanon": 39.01,
    "space_norm_wer": 34.63,
    "mer": 19.12,
    "cer_norm": 17.93,
    "empty_hypotheses": 1,
    "normalization_delta": {
      "raw_to_norm": 0.23,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 4.38,
      "norm_to_mer": 19.89
    }
  },
  "bengali": {
    "n_samples": 500,
    "wer_raw": 39.01,
    "wer_norm": 38.23,
    "wer_numcanon": 38.23,
    "space_norm_wer": 29.44,
    "mer": 11.66,
    "cer_norm": 11.52,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.78,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 8.79,
      "norm_to_mer": 26.57
    }
  },
  "english": {
    "n_samples": 500,
    "wer_raw": 28.35,
    "wer_norm": 14.85,
    "wer_numcanon": 14.85,
    "space_norm_wer": 10.25,
    "mer": 9.36,
    "cer_norm": 9.2,
    "empty_hypotheses": 18,
    "normalization_delta": {
      "raw_to_norm": 13.5,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 4.6,
      "norm_to_mer": 5.49
    }
  },
  "gujarati": {
    "n_samples": 500,
    "wer_raw": 31.06,
    "wer_norm": 30.76,
    "wer_numcanon": 30.76,
    "space_norm_wer": 23.83,
    "mer": 10.04,
    "cer_norm": 9.9,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.29,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 6.94,
      "norm_to_mer": 20.72
    }
  },
  "hindi": {
    "n_samples": 500,
    "wer_raw": 21.75,
    "wer_norm": 21.72,
    "wer_numcanon": 21.72,
    "space_norm_wer": 16.49,
    "mer": 8.43,
    "cer_norm": 7.95,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.03,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 5.23,
      "norm_to_mer": 13.29
    }
  },
  "kannada": {
    "n_samples": 500,
    "wer_raw": 48.1,
    "wer_norm": 47.74,
    "wer_numcanon": 47.74,
    "space_norm_wer": 37.12,
    "mer": 12.17,
    "cer_norm": 12.31,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.36,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.62,
      "norm_to_mer": 35.57
    }
  },
  "malayalam": {
    "n_samples": 500,
    "wer_raw": 53.79,
    "wer_norm": 52.3,
    "wer_numcanon": 52.3,
    "space_norm_wer": 40.9,
    "mer": 11.58,
    "cer_norm": 11.54,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 1.49,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 11.4,
      "norm_to_mer": 40.72
    }
  },
  "marathi": {
    "n_samples": 500,
    "wer_raw": 45.22,
    "wer_norm": 45.15,
    "wer_numcanon": 45.15,
    "space_norm_wer": 35.23,
    "mer": 14.25,
    "cer_norm": 14.23,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.07,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 9.92,
      "norm_to_mer": 30.89
    }
  },
  "odia": {
    "n_samples": 500,
    "wer_raw": 40.86,
    "wer_norm": 40.39,
    "wer_numcanon": 40.39,
    "space_norm_wer": 30.02,
    "mer": 12.69,
    "cer_norm": 12.7,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.46,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.37,
      "norm_to_mer": 27.71
    }
  },
  "punjabi": {
    "n_samples": 500,
    "wer_raw": 29.03,
    "wer_norm": 28.93,
    "wer_numcanon": 28.93,
    "space_norm_wer": 22.43,
    "mer": 9.72,
    "cer_norm": 9.42,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.1,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 6.5,
      "norm_to_mer": 19.21
    }
  },
  "tamil": {
    "n_samples": 500,
    "wer_raw": 50.16,
    "wer_norm": 49.03,
    "wer_numcanon": 49.03,
    "space_norm_wer": 39.38,
    "mer": 12.31,
    "cer_norm": 12.2,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 1.13,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 9.65,
      "norm_to_mer": 36.72
    }
  },
  "telugu": {
    "n_samples": 500,
    "wer_raw": 39.06,
    "wer_norm": 37.35,
    "wer_numcanon": 37.35,
    "space_norm_wer": 26.86,
    "mer": 8.14,
    "cer_norm": 8.38,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 1.7,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.49,
      "norm_to_mer": 29.21
    }
  },
  "__overall__": {
    "n_samples": 6000,
    "wer_raw": 38.67,
    "wer_norm": 37.07,
    "wer_numcanon": 37.07,
    "space_norm_wer": 28.96,
    "mer": 11.82,
    "cer_norm": 11.66
  },
  "__macro_avg__": {
    "n_languages": 12,
    "wer_raw": 38.8,
    "wer_norm": 37.12,
    "wer_numcanon": 37.12,
    "space_norm_wer": 28.88,
    "mer": 11.62,
    "cer_norm": 11.44
  },
  "__meta__": {
    "checkpoint": "/home/ubuntu/training/checkpoints/maya-asr-tdt-1.1b-ckpt-60000/model.ckpt",
    "checkpoint_name": "ckpt-60000",
    "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": 349.33,
    "total_audio_sec": 40354.46,
    "rtf": 0.0087,
    "timestamp": "2026-03-29T06:47:31Z",
    "gpu": "NVIDIA A100-SXM4-80GB",
    "framework": "nemo",
    "normalization_version": "v1",
    "jiwer_version": "unknown"
  }
}