{
  "modelId": "assamese-indicconformer-ctc-onnx",
  "modelName": "Assamese Offline Voice Model",
  "modelVersion": "1.0.0",
  "bucketName": "keyboard-voice-models",
  "destinationDirectory": "assamese/",
  "downloadDomain": "https://voice.camorabug.com",
  "files": [
    {
      "fileName": "model_ctc.onnx",
      "objectKey": "assamese/model_ctc.onnx",
      "fileSizeBytes": 493060285,
      "sha256": "7e088fe668fbd6d5ab9fb9bc688f5894fdadb7aa53d6ce1e4c9c9dfe3a33bf58",
      "contentType": "application/octet-stream",
      "downloadUrl": "https://voice.camorabug.com/assamese/model_ctc.onnx",
      "originSourceUrl": "https://angsh.in/rodali_voice_onnx_data/download_onnx.php"
    },
    {
      "fileName": "assamese_ctc_vocab.json",
      "objectKey": "assamese/assamese_ctc_vocab.json",
      "fileSizeBytes": 3302,
      "sha256": "6cf4ba9cf509b90672827f44bf17f5753d28ff932542cd7572bff32b4f80075a",
      "contentType": "application/json",
      "downloadUrl": "https://voice.camorabug.com/assamese/assamese_ctc_vocab.json"
    }
  ],
  "dependencies": {
    "runtime": "com.microsoft.onnxruntime:onnxruntime-android:1.20.0",
    "onnxIrVersion": 8,
    "opsetVersion": 16,
    "requiredFiles": [
      "model_ctc.onnx",
      "assamese_ctc_vocab.json"
    ]
  },
  "inputTensorSpecifications": [
    {
      "name": "audio_signal",
      "dataType": "FLOAT32",
      "shape": [
        "batch",
        80,
        "time"
      ],
      "description": "80-band log-mel filterbank features normalized per feature"
    },
    {
      "name": "length",
      "dataType": "INT64",
      "shape": [
        "batch"
      ],
      "description": "Number of time frames T in audio_signal"
    }
  ],
  "outputTensorSpecifications": [
    {
      "name": "logprobs",
      "dataType": "FLOAT32",
      "shape": [
        "batch",
        "time_subsampled",
        5633
      ],
      "description": "Frame-level log-probabilities across 5632 multilingual tokens + blank token at index 5632"
    }
  ],
  "audioSampleRateHz": 16000,
  "preprocessingRequirements": {
    "audioEncoding": "PCM_16BIT",
    "channels": 1,
    "sampleRateHz": 16000,
    "preemphasis": 0.97,
    "nFft": 512,
    "winLength": 400,
    "hopLength": 160,
    "window": "hann_symmetric",
    "nMels": 80,
    "fMinHz": 0.0,
    "fMaxHz": 8000.0,
    "melScale": "slaney",
    "magPower": 2.0,
    "logZeroGuard": 5.960464477539063e-08,
    "normalization": "per_feature"
  },
  "decoderInformation": {
    "type": "CTC_GREEDY",
    "assameseTokenIdRange": [
      0,
      255
    ],
    "unkTokenId": 0,
    "blankTokenId": 5632,
    "wordBoundaryPrefix": "▁",
    "collapseRepeatedTokens": true
  }
}