During November 2025, Duwen Han developed expressive speech capabilities for the coze-dev/coze-js repository, focusing on enhancing the API to support nuanced voice emotions in speech synthesis. By introducing emotion and emotion scale properties to the speech request interface and updating both voice and audio output interfaces, Duwen enabled developers to configure emotional tone in synthesized speech. This work was implemented using TypeScript and leveraged full stack development skills to ensure seamless integration across the API surface. The feature lays groundwork for future enhancements, reflecting a focused and technically sound approach to expanding the expressiveness and flexibility of the Coze JS platform.