This looks really cool! Can't wait to try it when I'm listening to podcasts on my walk; are you using a paid API for the transcript or are you using an open source solution yourselves?