Introduction

The core design philosophy of DeepSeek Harness (DSH) is “everything is a plugin.” When building agents or workflows involving multimedia content, calling third-party AI services often requires dealing with cumbersome tasks such as API key management and maintaining account login status.

The plugin yu-wenchao/dsh-tool-lipsync is designed to solve this pain point. It registers 5 tools for DSH, allowing users to call a free open API directly in the conversation flow to generate lip-sync talking videos. The plugin requires no account registration and no API key configuration, lowering the barrier to AI video generation.

Core Features

The plugin provides the following tools and capabilities:

  • Voice management: Lists preset voices, supports filtering by language or keyword, and covers 500+ languages, including Chinese, English, and Japanese.
  • Asset management: Lists sample face, pet, and cartoon assets.
  • Task submission: Submits a lip-sync video generation task; supports up to 20 seconds of video.
  • Status monitoring: Polls generation status, with a default interval of 5-6 seconds.
  • Final video download: Downloads the generated MP4 video to the local workspace.

Installation and Enablement

In the DSH Web profile environment, run the following command to install it:

dsh plugin --profile web add github:yu-wenchao/dsh-tool-lipsync

After installation, restart DSH to activate the plugin.

Typical Usage

In a DSH conversation, you can directly describe your requirements. The plugin will automatically call the above 5 tools to work together.

  1. Requirement description: Specify the voice, assets, and content.
  2. Automatic execution: The system automatically lists voices, lists assets, submits the task, polls progress, and downloads the final video.

Example:

Please use a Japanese female voice with the sample face asset to generate a 15-second talking video with the text “こんにちは、世界”.

If you want to use a custom face image, you can specify an image URL:

Use this image as the face https://example.com/face.jpg with an English male voice saying “This is a test”.

Notes and Ecosystem

  • Version requirement: The DSH version must be >= 0.1.0-rc.6.
  • Free tier limitations: The free tier output is a low-quality preview; the high-quality, watermark-free original video requires login or payment.
  • Service stability: This plugin depends on third-party free services. Their free quotas and API endpoints may be adjusted at any time; generation speed is usually between 1 and 3 minutes, depending on server load.
  • Copyright and compliance: Please pay attention to copyright issues related to assets and portraits.
  • Ecosystem positioning: This plugin is maintained by community developers. The plugin directory has no affiliation with DeepSeek’s official team.

Conclusion

By encapsulating underlying API calls, this plugin provides DSH users with a low-barrier entry point for video generation. If you need to develop automated workflows involving lip-sync videos, see its GitHub repository: https://github.com/yu-wenchao/dsh-tool-lipsync.