我们给 DeepSeek Harness 加上了语音对话能力(实时全双工 · v1.10.0 起实验性接入) #263
x-lixu
started this conversation in
Show Your Plugins!
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
我们给 DeepSeek Harness 加上了语音对话能力
大家好,我们是 QwenAudio/qwen-audio-agent 的维护者(一个全双工实时语音的社区开源项目,2100+ stars)。今天发布的 v1.10.0,给 DeepSeek Harness 带来了语音对话能力——装上之后,你的 dsh 就有了"耳朵和嘴":用语音驱动 Harness 干活,它也会把结果说给你听。
dsh 获得的语音能力
接入方式
通过
dshCLI 与dsh-acp-demoACP 适配器(原生 ACP 集成),桌面悬浮球/CLI 中一键安装这两个组件,配置 DeepSeek API Key 后即可用语音驱动 Harness。试用(需 Node 20+ 与 DeepSeek API Key):
桌面端选择 DeepSeek 后台(标注"experimental"),或 CLI 中
qwenaudio install deepseek一键安装后以qwenaudio --backend deepseek启动。🖥️ 桌面端:悬浮球形态,零命令行
不想碰终端的话,桌面端更省事(macOS / Windows / Linux,支持自动更新,Releases 页直接下载安装包):
欢迎体验并反馈 Harness 侧的兼容性问题:https://github.com/QwenAudio/qwen-audio-agent
English summary: We added voice conversation to DeepSeek Harness. Qwen Audio Agent v1.10.0 plugs into
dshvia thedsh-acp-demoACP adapter (one-click install): full-duplex voice with barge-in, spoken permission confirmations, automatic task result announcements, and local wake word. A desktop orb app (macOS/Windows/Linux) offers a zero-command-line experience: graphical backend setup, live task progress cards, and custom skins. Repo: https://github.com/QwenAudio/qwen-audio-agentAll reactions