Developed a text preprocessing module for the vllm-project/vllm-omni repository, focusing on enhancing the TTS demo by improving input quality and reliability. The module was engineered in Python and leverages regex-based sanitization to remove unwanted characters, format numbers and currencies, and enforce proper punctuation. This approach ensures cleaner, more consistent input for text-to-speech conversion, reducing input-related errors and supporting a smoother user experience. Integration with Gradio facilitated seamless demo interactions, while strong Git hygiene practices ensured traceable, well-documented contributions. The work established a robust foundation for future TTS enhancements, emphasizing input normalization and maintainable, testable code through unit testing.