Worked on the pipecat-ai/pipecat repository to deliver a major audio processing enhancement focused on multi-subscriber call quality. Developed a per-subscriber audio callback mechanism in the Vonage client, introducing dedicated resamplers for each subscriber to isolate audio streams and prevent cross-stream interference. Updated the listener interface to support individualized callbacks, enabling more granular audio handling in sessions with multiple participants. The implementation relied on Python and leveraged asynchronous programming, callback handling, and unit testing to ensure robust functionality. This work established a foundation for future per-user audio features, addressing the need for scalable, high-quality audio processing in complex communication scenarios.