# dsh-voice-plugin [English](README.md) | [中文](README.zh.md) > 如果你是 agent,请先读 AGENTS.md。 给 DeepSeek Harness 输入框加一个麦克风:录音、确认,文字就进入草稿。 ## 安装 ```sh dsh plugin --profile web add dsh-voice-plugin # 或从 GitHub 直装: dsh plugin --profile web add github:yafangwang9/dsh-voice-plugin ``` 然后重启 `dsh web`。 ## 配置 打开 **设置 → Plugins → voice-plugin**,选一个模型: - 中文 → **豆包语音** - 英文或其他语言 → **OpenAI** 只填对应模型的 API Key。还没配置时,点麦克风只会提示去设置。 ## 许可 MIT