Introduction

DeepSeek Harness (DSH) adopts an “everything is a plugin” architecture, allowing developers to flexibly extend functionality as needed. In local model inference or agent development scenarios, voice input is an important means of improving interaction efficiency. This article introduces a plugin focused on speech-to-text.

Plugin Purpose

dsh-dictation is a DSH plugin maintained by user WSL043. It provides local multilingual recognition capabilities and integrates with the Codex Desktop dictation feature to write recognition results into an editable draft.

Core Features

  • Local multilingual recognition
  • Codex Desktop dictation

Installation and Enablement

The official docs do not provide explicit installation commands. Because retrieving the README failed, specific installation steps cannot be confirmed. It is recommended to visit the GitHub repository directly to review the source code or documentation. The plugin runs with the permissions of the current DSH process; its source code and license should be checked before installation.

Typical Usage

Specific usage commands and operational workflows cannot currently be confirmed. The feature description does not include reproducible examples.

Applicable Scenarios and Notes

  • Suitable for developers who need local voice input.
  • Privilege Note: The plugin runs with the permissions of the current dsh process. Be sure to check the source code and license before installation to ensure compliance with security and regulatory requirements.

Conclusion

This plugin provides local multilingual recognition and dictation capabilities. For more details, visit the plugin directory page or the GitHub repository.