{
"id": "audio-feature-extraction",
"name": "Audio Feature Extraction",
"version": "1.0.0",
"description": "Extract audio features (spectrograms, MFCCs) for speech recognition and audio classification",
"steps": [
{
"id": "extract_features",
"step_type": "audio_features",
"config": {
"feature_type": "mel_spectrogram",
"sample_rate": 16000,
"n_fft": 2048,
"hop_length": 512,
"n_mels": 128,
"f_min": 0.0,
"f_max": 8000.0,
"normalize": true
},
"cache_results": true,
"description": "Extract Mel spectrogram features from audio"
}
],
"metadata": {
"created_at": "2025-12-31T00:00:00Z",
"author": "audio-ml-team",
"target_model": "Wav2Vec2/Whisper/Audio Classifier",
"dataset": "LibriSpeech/AudioSet",
"use_case": "Speech recognition and audio classification",
"recommended_batch_size": 16,
"notes": "Optimized for speech and audio understanding tasks"
}
}