The whisper transcript api is returning the input prompt as the transcribed text instead of actually transcribing the audio. This happens for certain audio not all audio.
