I am accessing OpenAI and Whisper with n8n. The transcription of an audio WAV file is not working as expected. I am receiving text output that I did not speak. It seems to me that the text might be coming from other users. Could there be an issue here?
You should trim silence from the start and end of the file and any long silences within by detection, along with normalizing the levels, or you will get hallucination output on the background noise.