dsh-voice-ai-girlfriend
beiyege-01
语音 AI 女友(Voice AI girlfriend for DeepSeek Harness):Whisper 语音输入 + Qwen3-TTS 声音克隆 + 句子级流式朗读 + 数字人动画窗。插话/排队双模式,说话即打断。
SAOS-36244210Updated yesterday
陪看:和智能体一起看视频 —— 跟着真实播放位置,每段结束前把分镜画面 + 带秒台词交过去;悬浮窗里能出声,也听得见房间。A DSH plugin that watches a video together, in sync with the real playback: a storyboard image + subtitle lines per segment, spoken replies and live room audio.
Practical score
87 / 100
Setup
Ready to use
License
Apache-2.0
Language
JavaScript
Copy the command, or hand the whole task to your harness agent.
git clone https://github.com/SAOS-3624/dsh-watch-together.gitPlease install the DeepSeek Harness plugin "dsh-watch-together" for me. - GitHub repository: SAOS-3624/dsh-watch-together (https://github.com/SAOS-3624/dsh-watch-together) - Type: a Cordis plugin (installed into a DSH profile) - The plugin declares a DSH versi…
A cleaned excerpt from the project's own documentation.
dsh-watch-together 陪智能体 一起看视频 :跟着你的真实播放进度,在你快看完这一段时,把这一段的 画面分镜 和 带秒台词 一起递过去;它还能在悬浮窗里出声、也听得见房间里的声音 —— 它看到的是画面,不是一段转写文本。 名字全小写,带横杠 —— 和 dsh-video-storyboard 同一个风格。 English TL;DR — a DSH plugin that watches a video together with the agent, in sync with the real playback position a browser extension reports. Each segment arrives as one storyboard image plus its subtitle lines just before that segment ends; the floating window carries spoken replies (through whatever voice adapter you plug in) and live room audio. The agent gets picture, text and sound — not a transcript. Code, identifiers and the tool schema are English; the docs below are Chinese. - 它是什么 :一个 DSH 插件,只注册一个工具 watch together - 它不是什么 :不是全片概览,不是播放器,也不带 TTS 引擎(声音自己接,见 §接声音) - 依赖 :…
Open on GitHubSimilar tools from the same corner of the ecosystem.
beiyege-01
语音 AI 女友(Voice AI girlfriend for DeepSeek Harness):Whisper 语音输入 + Qwen3-TTS 声音克隆 + 句子级流式朗读 + 数字人动画窗。插话/排队双模式,说话即打断。
zzy6-a
DSH Computer Use:让 Agent 看见并操作 Windows 桌面(Windows 原生 / WSL 自动识别);视觉通道 + 鼠标键盘 + Codex 风格蓝色覆盖层,Esc 随时中止。
SAOS-3624
Scene-aware storyboard for DeepSeek Harness: 9-16 numbered cells with timestamps + transcript + an audio envelope, packed into ONE image (384 tokens). Two tools: video_storyboard and video_zoom.