Introduction

DeepSeek Harness (DSH) adopts a plugin-based architecture, and the Web client extends functionality through plugins. When building physical voice interaction scenarios, it is necessary not only to communicate with the backend gateway but also to display microphone status in the UI layer and provide a control entry. This section introduces @xp1024/dsh-voice-gateway, a voice gateway client plugin for the DSH Web client that handles physical microphone status indication, voiceprint management, and automatic interaction with the voice gateway.

Core Features

This plugin provides the following core capabilities for the Web interface:

  • Six-state real-time status indicator: Provides real-time feedback for six states: idle, listening, thinking, generating, speaking, and disconnected.
  • Voiceprint filtering and multi-speaker management: Supports wizard-based voiceprint recording and registration, manages multiple speaker profiles, and filters unauthorized voice input.
  • Bidirectional voice interaction: Supports the Auto-speak toggle and formats speech recognition results as [<speaker>] <text> to be injected into the current session.
  • Dynamic gateway address configuration: Allows modifying the voice gateway address in the UI settings panel, with the default address being http://127.0.0.1:8765.
  • Voice control button mounting: The voice control button is mounted on the right side of the chat input bar, providing a convenient operation entry.

Installation and Activation

Install and activate this plugin in a configured DSH Profile (such as web):

dsh plugin --profile web add @xp1024/dsh-voice-gateway

Typical Usage

Before using it, make sure the local physical voice gateway daemon is running.

  1. Start the local voice gateway:
git clone https://github.com/zq0951/local-voice-gateway.git
cd local-voice-gateway
./start.sh

The gateway listens on http://127.0.0.1:8765 and ws://127.0.0.1:8765/v1/events by default.

  1. Install the plugin:
    Run the installation command above.

  2. Start the Web interface:

dsh web --no-open --port 3080

After startup, open DSH in your browser. A microphone control button will appear on the right side of the chat input bar.

Use Cases and Notes

  • Use cases: Developers who need to perform voice interactions through a physical microphone on the Web and require voiceprint recognition or status feedback.
  • Notes: This plugin is a third-party open-source plugin and is not an official DeepSeek product. Before use, ensure that Local Voice Gateway is running locally (default port 8765). The plugin runs with the permissions of the current DSH process. It is recommended to review the source code and license before installation.

Conclusion

This plugin provides complete voice interaction UI support for the DSH Web client. For more details and source code, refer to the GitHub repository.