in api/run_eval.py [0:0]
def transcribe_with_retry( model_name: str, audio_file_path: Optional[str], sample: dict, max_retries=10, use_url=False,