In listening tasks, if the audio file or transcription is missing, the AI must strictly state that execution is impossible without the source audio file and is forbidden from attempting to guess the audio content.