Lightning V3

About

Introducing Lightning V3, the most sophisticated text-to-speech model from Smallest AI. It features a latency of 100 milliseconds and achieves a WVMOS score of 3.89. Supporting English, Hindi, Spanish, Tamil, and more than fifteen additional languages, V3 was chosen by listeners over OpenAI's GPT-4o-mini-TTS in 76.2% of comparisons. It delivers audio at 44.1 kHz and enables voice assistants, IVR systems, content creation, and conversational AI with natural, human-like speech. The model offers instant voice cloning from only ten seconds of audio. It is real-time, expressive, and ready for enterprise use.