The feature extractor is designed to extract features from raw audio data, and convert them into tensors.