1
DeepSeek 开口说话了：灰度测试 AI 语音对话，支持四种音色
IT之家 9 月 12 日消息，据IT之家小伙伴反馈， DeepSeek 灰度测试语音对话 ，测试到的用户在 App 右上角会有一个小喇叭按钮。 此外，DeepSeek 的设置页面会出现一个“ 朗读音色 ”选项，用户可以选择四种音色： 贝壳：百变活泼 白浪：明朗坚定 海星：俏皮甜美 暗潮：低沉浑厚
2
DuplexChat: Constructing Speaker-Separated Full-Duplex Dialogue Speech at Scale for Spoken Dialogue Language Modeling
推荐阅读这篇论文：DuplexChat构建大规模说话人分离的全双工对话语音数据，为口语对话语言建模提供有力支撑，方法新颖实用。
arXiv:2607.04941v1 Announce Type: new Abstract: Full-duplex spoken dialogue models are trained on conversational speech in which each speaker is repre…
3
Show HN: Japanese Language AI Tutor in 3D classroom
3D教室里与AI教师实时语音互动，按你的水平纠正日语发音，沉浸式练口语新体验。
Article URL: https://unihongo.com Comments URL: https://news.ycombinator.com/item?id=48669612 Points: 3 # Comments: 0
4
Joint Speech and Text Training for LLM-Based End-to-End Spoken Dialogue State Tracking
端到端口语对话状态追踪新突破：联合语音与文本训练，显著提升LLM在语音场景下的跟踪准确率。
arXiv:2511.22503v2 Announce Type: replace Abstract: End-to-end spoken dialogue state tracking (DST) is made difficult by the tandem of having to handl…
5
Proactive for Uncertainty: Cause-Aware Error Diagnosis and Interactive Clarification for Spoken Dialogue Systems
因果感知的错误诊断与交互澄清方法，让语音对话系统主动应对不确定性，提升鲁棒性。
arXiv:2605.25404v1 Announce Type: new Abstract: Cascaded Automatic Speech Recognition -- Large Language Model (ASR-LLM) pipelines remain popular for i…
6
KAME: Tandem Architecture for Enhancing Knowledge in Real-Time Speech-to-Speech Conversational AI
提出KAME双架构协同，实时语音对话AI兼顾流畅交互与深度知识，突破现有模型语义瓶颈
arXiv:2510.02327v2 Announce Type: replace-cross Abstract: Real-time speech-to-speech (S2S) models excel at generating natural, low-latency conversatio…
7
谷歌升级 Workspace：语音对话协作 AI 检索邮件、起草文档、整理笔记
语音对话协作，AI智能检索邮件、起草文档、整理笔记，让办公效率倍增
IT之家 5 月 20 日消息，在今日的 2026 谷歌 I/O 开发者大会上，谷歌 CEO 桑达尔 · 皮查伊（Sundar Pichai）宣布升级 Google Workspace， 把 Gmail、Docs、Keep 等工具变成可语音对话的协作助手。 IT之家注：新功能包括可用自然语音检索邮件…