GPT-Live-1 provides a new capability within the OpenAI API focused on voice interactions. The system is designed to facilitate natural, full-duplex conversations, meaning it supports both listening and responding in real-time. The API now includes stronger instruction following, improving the model's ability to understand and execute user commands within a voice context. Custom voices are supported, allowing for tailored audio experiences. Telephony support is also integrated, opening possibilities for voice-based applications within telephony systems.
Source: https://openai.com/index/introducing-gpt-live-1-in-the-api
