I'm using AWS transcribe. I would love to eventually train my own model specifically for each podcast. For this specific instance, since Joe is moving the Spotify soon, I'm not sure the investment is worth it.