⬢github C · 17 ★ +10 since we first saw it · pushed 9 h ago · NOASSERTION
tlack/babytalk
Optimized ESP32-S3/P4 fully offline speech to text and text to speech system for Micropython and AtomVM, featuring a finetuned STT model for noisy environments, plus a data generation/field capture/model finetuning framework. Includes high performance 4/8bit int kernels for Xtensa/RiscV PIE vector units
BabyTalk is a library that runs fully offline speech-to-text and text-to-speech on ESP32-S3/P4 microcontrollers, transcribing arbitrary English sentences without a command list or cloud. It works from C, MicroPython, or Erlang/Elixir via AtomVM, uses quantized 4/8-bit models that run from flash within tiny RAM budgets, and includes tooling for data capture and model fine-tuning.
Why now: It was just featured on Hacker News as a Show HN post, drawing attention for achieving general-purpose on-chip speech recognition on a microcontroller without cloud services.
Who it is for: Embedded and IoT developers building private, offline voice interfaces on ESP32 hardware, such as keyboardless LoRa radios or voice-controlled devices.
asratomvmedge-aielixirerlangesp-idfesp32esp32-p4esp32-s3int4
Stars over our 24 snapshots: 7 to 17, since 5 h ago.
Where people talked about it
API: https://socialmediatrends-api.osmike.com/v1/repos/tlack/babytalk