It's included in the audio stream sent to the transcriber and can reduce transcription accuracy, so it’s best to disable it for tasks requiring precise input like CSAT scores..