Objective

• Translate Alice's voice for Bob to hear in Bob's language. Translate Bob's voice for Alice to hear in Alice's language.
• Neither person should hear translation of their own voice.
• Alice and Bob could be in the same room physically or in different rooms.
• Neither person should hear noise due to closed loop between a mic and speaker.

Zero-code solution

1. Open: Realtime API in openAI playground in macOS Safari. Input: macOS mic
2. Open: Zoom. Input: Loopback Audio. Output: macOS speaker
3. Open: Rogue Amoeba Loopback.app. Create new device. Safari 1&2 -> Channels 1&2

Do this on only one device for translation one way. Do this on both devices for translation both ways.

Once you have this setup working, you can also connect headphones for better noise cancellation if both people are in the same room. Only change required is Zoom Output: Headphones.

Prepend each prompt with "translate to French/Chinese/etc" either by speaking these 3 words aloud, or by writing an app that can do it automatically. (I can host this if there's demand.)