Audio has to be sent to do the recognition. The only question is whether the company then turns around and saves it and sends it for further analysis. You can't

Audio has to be sent to do the recognition. The only question is whether the company then turns around and saves it and sends it for further analysis. You can't track that step.