README.md

---
title: Q Live — the Serverless Voice
emoji: 🎙️
colorFrom: indigo
colorTo: purple
sdk: static
app_file: index.html
pinned: true
short_description: Talk to Q — a voice AI running 100% in your browser
---

# Q Live — the Serverless Voice

Tap the orb and **talk to Q**. Every model — the brain (BitNet‑2B), the ear (Moonshine), the voice
(Kokoro) — streams by content address from [HOLOGRAMTECH](https://huggingface.co/HOLOGRAMTECH),
is verified per block, and runs **entirely in your browser** on WebGPU/WASM.

**Zero inference server. Your voice never leaves your device. Works offline after first load.**

This is the real‑time speech‑to‑speech experience of a datacenter voice pipeline — with no datacenter.
Cerebras throws a wafer at latency; Q throws your own GPU and its idle time (it starts answering on your
confident partial while you're still finishing). One self‑verifying κ‑link boots the whole runtime in any
cold browser.

Built with the Hologram stack. Best in a Chromium browser with WebGPU.