Moonshine Micro: Speech Recognition and TTS in under 500KB for Microcontrollers
Summary
Moonshine Micro provides an open-source AI toolkit for microcontrollers, enabling VAD, STT, and neural TTS on resource-constrained devices. It outlines a compact memory budget (~470 KB RAM, ~3.6 MB Flash) and an end-to-end example on RP2350, using TensorFlow Lite Micro for neural computations. Suitable for embedded edge AI enthusiasts and SMBs looking to deploy voice apps on small hardware.