During March 2025, contributed to the ggml-org/llama.cpp repository by developing a new feature that introduced speaker file support to the Text-to-Speech (TTS) system. This enhancement enabled customizable audio generation tailored to individual speaker profiles, expanding the personalization capabilities of the TTS pipeline. The implementation was achieved through a focused and low-risk change set, ensuring stability and maintainability within the existing codebase. Leveraging expertise in C++, audio processing, and text-to-speech technologies, the work aligned with the project’s roadmap for user personalization and established a foundation for future speaker-based enhancements without introducing regressions or instability.