Show HN: Real-time AI (audio/video in, voice out) on an M3 Pro with Gemma E2B

Build a modular framework that allows developers to swap local inference backends for real-time multimodal interaction on edge hardware. There is a lack of clean, unified abstraction layers for low-latency audio-to-audio flows outside of closed-source ecosystem alternatives.

Suggested repo: parlor-flow

"Low-latency multimodal AI pipeline that runs locally on your Apple Silicon."

Estimated effort: 40h