voice-stt-dsh
Run the following command in DeepSeek Harness:
dsh plugin install vTRKA/voice-stt-dsh
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install vTRKA/voice-stt-dsh in your terminal to install this plugin in DeepSeek Harness; the source is hosted at https://github.com/vTRKA/voice-stt-dsh .
About this plugin
In DeepSeek Harness, the composer has always been a text-only experience. voice-stt-dsh closes that gap with a local microphone button and a keyboard shortcut. Click the button or press Ctrl+E, speak for a few seconds, click again or press the shortcut once more, and the transcript appears in your draft. Sending is still a deliberate, separate action the plugin will never trigger on your behalf.
The entire recognition pipeline runs offline. A local companion process receives the captured WAV file through 127.0.0.1 and hands it to the Parakeet TDT model, which returns the final text. No audio ever touches a cloud speech API. Model downloads require the publisher SHA-256 checksum, and the installer verifies it before replacing any existing file. You can rebind the shortcut via right-click or the settings page, as long as the new combination includes Ctrl, Alt, Shift, or Command.
If you run DSH Desktop or the Web profile on Windows x64, value data privacy, or work in an environment where audio cannot leave the machine, this plugin is a lightweight voice entry point for everyday chat. It needs no network, no account, and the model file persists across uninstall and reinstall so you never download it twice.
Use Cases
- Dictate chat messages hands-free instead of typing
- Transcribe speech locally in restricted or air-gapped workspaces
- Replace cloud speech-to-text APIs with a fully offline pipeline
Best For
- Privacy-focused DSH users who want audio to stay on-device
- DSH Desktop or Web-profile users on Windows x64
- Teams in regulated or offline environments needing voice input
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.