滴滴资深语音算法工程师(J250903029)

1、负责语音理解和语音生成算法在滴滴场景的落地使用； 2、跟进最新技术，结合业务场景，提升语音识别、音频事件检测、声纹识别、语音合成等算法效果； 3、探索语音大模型或多模态大模型在语音理解及语音生成场景的应用范式。

1.负责GVoice游戏语音服务中多语种语音识别模型的定制化设计、训练与迭代优化，精通Conformer、Paraformer等端到端语音识别模型，显著提升不同语种识别准确率； 2.主导语音合成相关技术和算法的研发工作，包括模型训练、优化及项目全流程交付，确保语音合成效果达到行业领先水平； 3.针对多轮对话、全时全双工等跨语言交互场景，设计并优化适配多语种特性的语音处理方案，显著提升跨语言交互体验； 4.深入优化多语种模型在设备端与云端的推理性能，包括模型压缩、量化、流式部署等技术，平衡多语种支持与响应速度、资源利用率； 5.参与多语种语音数据集的构建、扩充与质量优化，搭建多语种专属评测体系，实现从训练到验证再到部署的完整闭环； 6.持续跟踪多语种语音领域前沿技术（如低资源语种建模、跨语种迁移学习等），探索并推动其在实际业务场景中的应用落地。

The Role Voice integration engineers within the vehicle software organization work side by side with various cross-functional teams to lead the design and development of voice interaction systems. This role utilizes hardware, software, and system design fundamentals to support product design, voice interaction behaviors definition, prototype bring-up, and validation, with a software focus. Voice integration engineers are responsible for overall functionality and performance of voice interaction systems throughout product life cycle. This role will be responsible for voice recognition, natural language processing, audio processing pipeline, voice commands, and other voice-related features in the vehicle. Responsibilities Distill vehicle product requirements into actionable software requirements for voice services Lead software architecture design of voice interaction products and features Define system hardware and software interfaces for voice processing systems Bring-up and debug early software and hardware prototypes through characterizing the voice systems and ensuring systems perform to

1、负责 Soul App 移动端社交功能（如语音房、发布工具等）的规划与设计； 2、结合大模型（LLM）能力，探索 AI 在语音交互及内容生产工具中的落地，提升用户互动效率； 3、分析用户需求，绘制产品原型图，编写相关需求文档，并深度参与 UX/UI 交互逻辑的打磨； 4、收集并整理产品反馈与使用数据，通过 AI 技术或功能优化持续改善使用体验，提升用户留存； 5、协同算法、技术、设计、运营团队，共同推进 AI 创新项目从 0 到 1 的落地与迭代。