AI Briefing
KO

Supertonic 3 Launches as Ultra-Lightweight On-Device TTS, Supporting 31 Languages and Emotion Tags

·2026.05.15 10:05

Key point

Supertonic 3 has been released as an on-device TTS supporting 31 languages and emotion tags.

Details

Supertonic 3 has been released as an on-device TTS supporting 31 languages.

  • Emotional expression can be controlled by inserting 10 types of emotion tags such as laugh, breath, and scream into text.
  • It improves pronunciation accuracy, reduces word repetition/omission failures, and also improves voice cloning performance.
  • The model size is 99M parameters.
  • It emphasizes complete privacy protection and no network latency, and can run on browsers, PC, mobile, and even Raspberry Pi.
  • It provides 10 preset voices, and is available for commercial use under the OpenRAIL-M license.
speech-synthesison-device-aittsopen-modelsvoice-cloningmultilingual-ai
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

Google AI BlogGoogle Releases Gemini 3.8 Flash TTS: Voice Customization via Natural Language PromptsGeekNewsOpen-source 'VoiceStudio' released with support for local voice cloning, dubbing, and transcriptionRedditLoudKit: Local TTS with Voice Cloning, 10-Language Support, and SDKs for Python, Swift, Go, Rust, and TypeScriptRedditTontaube Releases Open TTS ModelHuggingFace BlogNVIDIA Releases Magpie TTSElevenLabsThe Key to Natural AI Voice: Defining Prosody and Its Application in TTS