Babytalk: Offline STT and TTS on a $20 ESP32, no cloud needed
tlack · hn · 2026-10-10
A Show HN project called Babytalk runs fully offline speech-to-text and text-to-speech on $20 ESP32S3/P4 boards with no keyboard or screen — no cloud service and no fixed command vocabulary.
Key points:
- Built with Claude: high-performance 4-bit int kernels for ESP32S3/P4, quantizing fairly large STT/TTS models so they run surprisingly well on tiny boards
- Includes a finetuned model that copes with accents, coughing, and mumbling
- Works with Micropython (RAM constrained) and AtomVM (better)
A practical open starting point for offline voice interfaces on cheap embedded hardware.
More from Embodied
- AI concept: Pokémon Snap on a foldable iPhone with dual-screen camera UI — yungcontent · 2026-10-23
- HumanoidPose recovers humanoid robots' full-body joint states from RGB video alone — cholz · 2026-10-23
- Peeling the iPhone Duo's nano-texture film reveals a surprisingly uneven screen underneath — film_girl · 2026-10-23
- ALPS: a single visual encoder unifying state, language, and pixels for world models — cihangxie · 2026-10-23
- XPeng IRON does voice-free handshakes, re-inferring motion every 0.2s via VLA 2.0 — CyberRobooo · 2026-10-23
- Viral humanoid robot cucumber-cutting video debunked: 2x background speed exaggerates the shaking — jonstephens85 · 2026-10-23