Speech Recognition and TTS in less than 500kb

	Speech Recognition and TTS in less than 500kb(github.com)
	545 points by petewarden 5 days ago \| 79 comments
	tl;dr: Moonshine Micro is an open-source (MIT-licensed) voice AI toolkit for microcontrollers, bundling voice activity detection, speech-to-text, and neural text-to-speech into ~470 KB of RAM and ~3.6 MB of flash. It targets the 80-cent Raspberry Pi RP2350 as its reference platform, uses TensorFlow Lite Micro for inference, and ships with an end-to-end example for voice-driven Wi-Fi setup on an MCU.
	HN Discussion: ↑Enthusiasm about the toolkit's capabilities and ease of use for hobby/practical projects ~Skepticism about real-world accuracy despite the small footprint •Curiosity about how it compares to existing low-memory TTS solutions like flite or nanotts •Questioning the utility of voice interfaces in general based on personal experience ↑Sharing related work or extensions built on top of the project