bf16/speaker_encoder/preprocessor_config.json

{
  "feature_extractor_type": "EcapaTdnnFeatureExtractor",
  "auto_map": {
    "AutoFeatureExtractor": "feature_extraction_ecapa_tdnn.EcapaTdnnFeatureExtractor"
  },
  "sample_rate": 24000,
  "n_fft": 1024,
  "hop_length": 256,
  "n_mels": 128,
  "fmin": 0,
  "fmax": 12000
}