audio_preprocessor_config.json

{
  "audio_end_token": "<|audio_end|>",
  "audio_start_token": "<|audio_start|>",
  "audio_token": "<|AUDIO_PAD|>",
  "auto_map": {
    "AutoFeatureExtractor": "audio_processing_hyperclovax_seed.HyperCLOVAXSeedAudioProcessor"
  },
  "chunk_length": 30,
  "chunk_unit": 80,
  "discrete_audio_end_token": "<|discrete_audio_end|>",
  "discrete_audio_start_token": "<|discrete_audio_start|>",
  "discrete_audio_token": "<|DISCRETE_AUDIO_PAD|>",
  "dither": 0.0,
  "feature_extractor_type": "HyperCLOVAXSeedAudioProcessor",
  "feature_size": 128,
  "hop_length": 160,
  "min_chunk_size": 1600,
  "n_fft": 400,
  "n_samples": 480000,
  "nb_max_frames": 3000,
  "padding_side": "right",
  "padding_value": 0.0,
  "pool_kernel_size": 5,
  "pool_stride": 5,
  "return_attention_mask": true,
  "sampling_rate": 16000,
  "use_discrete_token": false,
  "video_audio_pool_size": 25
}