babytalk

ESP32 上离线跑转写和合成，只用 84.5KB 内存 语音模型留在 flash 里按需读，不整块拷进 RAM，转写和合成轮流用同一块内存。MicroPython、C、AtomVM 都能调。 tlack/babytalk t.co

Home voice devices should handle basic commands when the internet drops. BabyTalk runs English speech recognition and synthesis offline on ESP32-S3/P4. Its reported noise tests are a reminder to test beside a running fan, not just at a quiet desk. tlack/babytalk

Show HN: Babytalk: Offline speech to text and text to speech on ESP32 Discussion

Show HN: Babytalk: Offline speech to text and text to speech on ESP32 comments · posted on 2026.10.09 at 17:33:56 (c=1, p=4)

Show HN: Babytalk: Offline speech to text and text to speech on ESP32 I have been playing with a lot of $20 devices with no keyboards or screens. I wanted to use speech to interact with them, but without it having to run through a cloud service or use a fixed command vocabulary. I worked with Claude to implement high performance 4bit int kernels for the ESP32S3 and P4 and quantize some larger STT and TTS models. Surprisingly, it's quite usable on these tiny boards. Also includes a finetuned model that can better cope with my diesel drones, hacking coughs, and mumbling. Works with Micropython (RAM scarcity issues) and AtomVM (better).