Pratham developed a real-time text-to-speech audio feature for the pipecat repository, integrating Sarvam TTS with WebSocket support to enable live audio generation and streaming. He updated the Sarvam TTS service to communicate over WebSockets and created an example script demonstrating interruptible speech synthesis, allowing for dynamic user interaction during live conversations. This work leveraged Python and WebSockets, focusing on API integration and full stack development to enhance the responsiveness and flexibility of the TTS pipeline. The implementation addressed the need for real-time, interruptible audio output, laying a technical foundation for future streaming-enabled speech synthesis features in pipecat.