A new text-to-speech (TTS) system called sanoTTS has been released, notable for its extremely small parameter size and efficient performance. The smallest version boasts 294,000 parameters and can run on a $3 microcontroller, while a larger 1.46 million parameter model outperforms significantly larger systems. sanoTTS offers multiple languages and voices, achieving a low Word Error Rate (WER) and fast real-time factor (RTF) on embedded hardware. AI
IMPACT Enables high-quality speech synthesis on extremely resource-constrained devices.
RANK_REASON Release of a new, highly efficient TTS model with detailed performance metrics. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →