For large transcriptions, it would be more efficient if we could specify to receive the json output without the list of words. At large scale this will save us meaningful bandwidth + each request should execute faster.