Minimax's fastest AI speech model — ultra-low latency voice generation with refined prosody and multilingual support for demanding real-time voice interfaces and API integrations