{
  "assamese": {
    "n_samples": 500,
    "wer_raw": 39.89,
    "wer_norm": 39.69,
    "wer_numcanon": 39.69,
    "space_norm_wer": 35.17,
    "mer": 19.45,
    "cer_norm": 18.26,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.2,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 4.51,
      "norm_to_mer": 20.24
    }
  },
  "bengali": {
    "n_samples": 500,
    "wer_raw": 38.14,
    "wer_norm": 37.44,
    "wer_numcanon": 37.44,
    "space_norm_wer": 28.46,
    "mer": 11.14,
    "cer_norm": 10.99,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.7,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 8.98,
      "norm_to_mer": 26.3
    }
  },
  "english": {
    "n_samples": 500,
    "wer_raw": 27.88,
    "wer_norm": 14.31,
    "wer_numcanon": 14.31,
    "space_norm_wer": 9.65,
    "mer": 9.03,
    "cer_norm": 8.87,
    "empty_hypotheses": 15,
    "normalization_delta": {
      "raw_to_norm": 13.57,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 4.66,
      "norm_to_mer": 5.28
    }
  },
  "gujarati": {
    "n_samples": 500,
    "wer_raw": 31.02,
    "wer_norm": 30.42,
    "wer_numcanon": 30.42,
    "space_norm_wer": 23.91,
    "mer": 10.34,
    "cer_norm": 10.14,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.6,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 6.51,
      "norm_to_mer": 20.08
    }
  },
  "hindi": {
    "n_samples": 500,
    "wer_raw": 22.0,
    "wer_norm": 21.91,
    "wer_numcanon": 21.91,
    "space_norm_wer": 16.65,
    "mer": 8.03,
    "cer_norm": 7.61,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.09,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 5.26,
      "norm_to_mer": 13.88
    }
  },
  "kannada": {
    "n_samples": 500,
    "wer_raw": 48.45,
    "wer_norm": 48.25,
    "wer_numcanon": 48.25,
    "space_norm_wer": 36.86,
    "mer": 11.59,
    "cer_norm": 11.85,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.2,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 11.39,
      "norm_to_mer": 36.65
    }
  },
  "malayalam": {
    "n_samples": 500,
    "wer_raw": 53.29,
    "wer_norm": 52.11,
    "wer_numcanon": 52.11,
    "space_norm_wer": 40.86,
    "mer": 11.27,
    "cer_norm": 11.21,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 1.18,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 11.25,
      "norm_to_mer": 40.85
    }
  },
  "marathi": {
    "n_samples": 500,
    "wer_raw": 44.62,
    "wer_norm": 44.44,
    "wer_numcanon": 44.44,
    "space_norm_wer": 34.58,
    "mer": 13.94,
    "cer_norm": 13.82,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.18,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 9.87,
      "norm_to_mer": 30.5
    }
  },
  "odia": {
    "n_samples": 500,
    "wer_raw": 40.23,
    "wer_norm": 39.66,
    "wer_numcanon": 39.66,
    "space_norm_wer": 30.78,
    "mer": 13.7,
    "cer_norm": 13.65,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.57,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 8.88,
      "norm_to_mer": 25.96
    }
  },
  "punjabi": {
    "n_samples": 500,
    "wer_raw": 28.13,
    "wer_norm": 27.99,
    "wer_numcanon": 27.99,
    "space_norm_wer": 21.79,
    "mer": 9.62,
    "cer_norm": 9.28,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.14,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 6.2,
      "norm_to_mer": 18.37
    }
  },
  "tamil": {
    "n_samples": 500,
    "wer_raw": 50.06,
    "wer_norm": 49.49,
    "wer_numcanon": 49.49,
    "space_norm_wer": 39.49,
    "mer": 12.51,
    "cer_norm": 12.44,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 0.58,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.0,
      "norm_to_mer": 36.98
    }
  },
  "telugu": {
    "n_samples": 500,
    "wer_raw": 39.52,
    "wer_norm": 37.78,
    "wer_numcanon": 37.78,
    "space_norm_wer": 27.07,
    "mer": 8.18,
    "cer_norm": 8.41,
    "empty_hypotheses": 0,
    "normalization_delta": {
      "raw_to_norm": 1.74,
      "norm_to_numcanon": 0.0,
      "norm_to_space_norm": 10.71,
      "norm_to_mer": 29.6
    }
  },
  "__overall__": {
    "n_samples": 6000,
    "wer_raw": 38.49,
    "wer_norm": 36.93,
    "wer_numcanon": 36.93,
    "space_norm_wer": 28.87,
    "mer": 11.76,
    "cer_norm": 11.6
  },
  "__macro_avg__": {
    "n_languages": 12,
    "wer_raw": 38.6,
    "wer_norm": 36.96,
    "wer_numcanon": 36.96,
    "space_norm_wer": 28.77,
    "mer": 11.57,
    "cer_norm": 11.38
  },
  "__meta__": {
    "checkpoint": "/home/ubuntu/training/checkpoints/maya-asr-tdt-1.1b-ckpt-80000/model.ckpt",
    "checkpoint_name": "ckpt-80000",
    "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": 346.44,
    "total_audio_sec": 40354.46,
    "rtf": 0.0086,
    "timestamp": "2026-03-29T06:49:49Z",
    "gpu": "NVIDIA A100-SXM4-80GB",
    "framework": "nemo",
    "normalization_version": "v1",
    "jiwer_version": "unknown"
  }
}