New audio models in the API + tools for voice agents

It’s getting harder to keep up!

Love the design of the demo site btw :heart_eyes: