swanada / config.json
gnumanth's picture
Release Swanada v1.0.0: 16 kHz Multilingual Indic Conformer-CTC INT8 Models
c9dc2ac verified
Raw History Blame Contribute Delete
2.67 kB
{
"name": "swanada",
"version": "1.0.0",
"pipeline_tag": "automatic-speech-recognition",
"sample_rate": 16000,
"frontend": {
"type": "log_mel_filterbank",
"features": 80,
"fft_size": 512,
"hop_length": 160,
"win_length": 400,
"window_fn": "hann",
"f_min": 0.0,
"f_max": 8000.0,
"preprocessor_model": "preprocessor.onnx"
},
"acoustic_model": {
"architecture": "Conformer-CTC",
"d_model": 512,
"n_layers": 17,
"n_heads": 8,
"conv_kernel_size": 31,
"subsampling_factor": 4,
"quantization": "INT8_DYNAMIC_QUINT8",
"time_stride_ms": 40.0
},
"languages": {
"sa": { "name": "Sanskrit", "script": "Devanagari", "model": "models/onnx/swanada_sa_ctc_int8.onnx", "vocab": "models/vocabularies/sa_vocab.json", "vocab_size": 257 },
"en": { "name": "Indian English", "script": "Latin", "model": "models/onnx/swanada_en_ctc_int8.onnx", "vocab": "models/vocabularies/en_vocab.json", "vocab_size": 257 },
"hi": { "name": "Hindi", "script": "Devanagari", "model": "models/onnx/swanada_hi_ctc_int8.onnx", "vocab": "models/vocabularies/hi_vocab.json", "vocab_size": 257 },
"kn": { "name": "Kannada", "script": "Kannada", "model": "models/onnx/swanada_kn_ctc_int8.onnx", "vocab": "models/vocabularies/kn_vocab.json", "vocab_size": 257 },
"mr": { "name": "Marathi", "script": "Devanagari", "model": "models/onnx/swanada_mr_ctc_int8.onnx", "vocab": "models/vocabularies/mr_vocab.json", "vocab_size": 257 },
"bn": { "name": "Bengali", "script": "Bengali", "model": "models/onnx/swanada_bn_ctc_int8.onnx", "vocab": "models/vocabularies/bn_vocab.json", "vocab_size": 257 },
"ta": { "name": "Tamil", "script": "Tamil", "model": "models/onnx/swanada_ta_ctc_int8.onnx", "vocab": "models/vocabularies/ta_vocab.json", "vocab_size": 257 },
"te": { "name": "Telugu", "script": "Telugu", "model": "models/onnx/swanada_te_ctc_int8.onnx", "vocab": "models/vocabularies/te_vocab.json", "vocab_size": 257 },
"ml": { "name": "Malayalam", "script": "Malayalam", "model": "models/onnx/swanada_ml_ctc_int8.onnx", "vocab": "models/vocabularies/ml_vocab.json", "vocab_size": 257 },
"gu": { "name": "Gujarati", "script": "Gujarati", "model": "models/onnx/swanada_gu_ctc_int8.onnx", "vocab": "models/vocabularies/gu_vocab.json", "vocab_size": 257 },
"pa": { "name": "Punjabi", "script": "Gurmukhi", "model": "models/onnx/swanada_pa_ctc_int8.onnx", "vocab": "models/vocabularies/pa_vocab.json", "vocab_size": 257 },
"or": { "name": "Odia", "script": "Odia", "model": "models/onnx/swanada_or_ctc_int8.onnx", "vocab": "models/vocabularies/or_vocab.json", "vocab_size": 257 }
}
}