Ningyao Voice Launcher
Install the bundled launcher into a user-chosen folder, then configure .env and run it.
Workflow
- Copy
assets/voice-chat-localinto the target folder. - Install dependencies with
npm installin the copied folder. - Copy
.env.exampleto.envand fill inOPENAI_API_KEY. - Start with
start-voice-chat.cmdorstart-voice-chat-bg.cmd.
Bundled Files
assets/voice-chat-local: Browser voice chat app template.scripts/install-launcher.ps1: Windows installer/copy helper.
Windows Install
Run:
powershell -ExecutionPolicy Bypass -File scripts/install-launcher.ps1 -Destination "$env:USERPROFILE\Desktop\ningyao-voice-chat"
The installer copies the template, skips .env, and can optionally run npm install.
Configuration
Set these in .env:
OPENAI_API_KEY: requiredOPENAI_BASE_URL: optional API-compatible endpointOPENAI_MODEL: text + vision capable model if screen analysis is neededOPENAI_TIMEOUT_MS: request timeoutPORT: local portSYSTEM_PROMPT: persona prompt
Notes
- Prefer Chrome or Edge for browser speech recognition and speech synthesis.
- The terminal panel is intentionally restricted to a small whitelist.
- If screen analysis is enabled, choose a model that supports image input.