He he. Thanks for the suggestion. Just wanted to show that an embedding model can transcribe. Maybe we should call it emscribe :D