in src/huggingface_inference_toolkit/serialization/audio_utils.py [0:0]
def deserialize(body): return {"inputs": bytes(body)}