I released sanoTTS: smallest complete TTS stack in 294k params (337 KB) that runs on $3 microcontroller and a 1.46m one that beats models 3x and 10x it's size

Explain Like I'm Normal

A developer released sanoTTS, a neural text-to-speech stack small enough to run on a $3 microcontroller with only 512kb of RAM. The 294k parameter model is roughly 1000x smaller than Kokoro while maintaining high voice quality and low word error rates. This massive efficiency gain allows for high-quality synthetic speech on low-power hardware and web browsers without a backend.