Maya 2 Native: Open-Weight Speech Model Supports 12 Languages with Real-Time Streaming

Maya Research has unveiled Maya 2 Native, an open-weight speech generation model that supports 11 Indian languages plus English. The model offers real-time audio streaming and cross-lingual voice cloning, enabling developers to build applications that can speak multiple languages fluently. This release is significant for the AI community because it provides open access to high-quality speech technology, which is often dominated by proprietary systems. By covering a wide range of Indian languages, Maya 2 Native addresses a critical gap in voice AI for one of the world's most linguistically diverse regions. The ability to clone voices across languages also opens up new possibilities for personalized assistants, audiobook narration, and accessibility tools. With real-time streaming, the model can be integrated into interactive applications like virtual agents and live translation services. As open-weight models continue to gain traction, Maya 2 Native could accelerate innovation in multilingual voice interfaces and democratize access to speech AI for developers worldwide.