Haru 4B: open-weight Apache 2.0 full-duplex speech model that listens and talks simultaneously
thisguyknowsai · x · 2026-11-17
Higurashi released Haru 4B, an end-to-end full-duplex speech-to-speech model with Apache 2.0 weights that runs on 12GB of GPU memory. Unlike cascaded ASR-LLM-TTS systems, it processes audio directly, emitting one frame every 133ms so listening and speaking never block each other, enabling backchanneling, interruptions and natural turn-taking. Character is set via a text card and voice via a reference clip; it supports 9 languages. The model card cautions it's a research base model, not a reliable factual assistant, and can clone voices—only use voices you have rights to.
More from Models
- Hermes App Screenshot Exposes Meta's Unreleased muse-wonder-1.0 Model with 1M Context — Teknium · 2026-11-17
- Agent token usage surges but price war leaves model vendors' revenue flat — FinanceYF5 · 2026-11-17
- Claude secretly banned 'blockchain' and every word needed to explain the project — RileyRalmuto · 2026-11-17
- tiny-voca specs: 4.5M params, 21MB ONNX, Apache 2.0 licensed — thisdudelikesAI · 2026-11-17
- 18MB browser-run TTS model tiny-voca makes paid English TTS look silly — thisdudelikesAI · 2026-11-17
- After MazeBench gets benchmaxxed, developers turn to secretive AI benchmarks — amplifiedamp · 2026-11-17