Developed and delivered a new text-to-speech feature for the vllm-omni repository, introducing Qwen3 TTS voice options and configurations. The work centered on creating a model recipe in Python that enables configurable voice cloning and text-to-speech capabilities, expanding the product’s accessibility and potential use cases. The implementation involved API development and integration of advanced TTS technologies, with careful adherence to contribution standards and collaborative workflows, as reflected in multi-signer commits. No major bugs were reported or fixed during this period. The result broadened the repository’s support for voice-enabled applications and enhanced flexibility for diverse user experiences.