hopetonp/voicemesh-dsh-voice
Voice Conversation by Hopeton adds live voice to DeepSeek Harness: choose a realtime speech provider (GPT Live, OpenAI Realtime, Gemini Live, xAI Grok) or a pipeline that combines local or hosted speech recognition, a conversational model and speech synthesis, driven from a control in the chat composer.
Voice Conversation by Hopeton adds live voice conversations to DeepSeek Harness. You can use a realtime speech provider or a pipeline that combines speech recognition, a conversational model and speech synthesis: pipeline recognition offers local Vosk or Parakeet or hosted Deepgram/Grok/OpenAI/Gemini/Fish Audio, synthesis offers local Kokoro or the same hosted set, and the conversational model can be any model already configured in Harness or an OpenAI-compatible endpoint (with presets for DeepSeek, OpenAI, Grok, Gemini, Groq, Mistral and OpenRouter). The settings editor has General, Voice provider and Conversation areas, and an optional 'Plain-language summaries' preference (off by default) has coding results explained in everyday language for both pipeline and live voice. The chat composer shows a voice control for an empty draft and Send while text is present.
Install
dsh plugin --profile web add @voicemesh/dsh-voice@0.1.0-beta.3Copied verbatim from the README's install block for the Web profile; the README also documents installing the same beta from Plugins -> Add plugin on Desktop.
Compatibility
Requires a DeepSeek Harness profile; restart the app or Web profile after installing or updating, enable the plugin in Plugins and open its settings editor. Choose between realtime speech (GPT Live, OpenAI Realtime, Gemini Live, xAI Grok) or a pipeline, then select providers and models; provider API keys are stored by DeepSeek Harness and the settings editor reports only that a key is saved. VoiceMesh does not download or select local pipeline speech models automatically; a local speech engine does not make the conversational model local. The README advises installing only one voice integration that contributes a microphone control. English README.
Details
- Repo: hopetonp/voicemesh-dsh-voice
- Category: Vision & Multimodal
- Stars: 0
- Version: GitHub repo hopetonp/voicemesh-dsh-voice (0★, MIT); npm package @voicemesh/dsh-voice at 0.1.0-beta.3 (registry check 2026-10-13; the registry repository field resolves back to this repo). Appears as 'Voice Conversation by Hopeton' in Harness.
Alternatives
54xkeee/dsh-vision · 54xkeee/dsh-youreyes · Cheng-cheng9669/dsh-deepseek-vision