audio segmentration in transcription apis is not a background task. So for transcribing longer audios the nginx gets timed out.