mirror of
https://github.com/debpalash/VoiceStudio.git
synced 2026-10-02 01:26:35 +08:00
debpalashandClaude Opus 5.5
9e7eef25ce
* docs(install): name the install request body and the real success check Follow-up to #2288 review: POST /models/install needs the repo_id, and the status endpoint drops finished jobs, so success is confirmed through GET /models installed=true. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> * docs(install): poll until jobs is empty, not for an empty body Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2.5 KiB
2.5 KiB
Install with a coding agent
Instructions for a coding agent (Claude Code, Codex, Cursor, …) asked to install VoiceStudio. The short prompt in the README points here. Complete the setup, not just a plan.
Rules
- Install the Electron desktop app. Tauri is archived; never install or launch it.
- Reuse existing data, settings and models. Ask before downloading a model: state its size and license.
- Keep cloud services and analytics opt-in.
- Verify the actual compute device; do not assume GPU support.
- Name any permission or manual step you cannot perform instead of skipping it.
Steps
- Load context. Read the install guide for this OS (macOS ·
Windows · Linux; Docker is a headless backend, not the desktop app),
performance and
skills/voicestudio/SKILL.md. If your agent supports skills, install it withnpx skills add debpalash/VoiceStudio. - Inspect the device. OS, CPU architecture, GPU, RAM/VRAM, free disk, and any existing VoiceStudio install, backend or downloaded models.
- Install.
- Default: the latest stable release asset named
VoiceStudio-Electron-*for this OS and architecture, or the install script. - From source: follow
electron/README.md—bun install,bun run setup:api,bun run devfrom the repository root. Let Electron supervise the backend; do not start a second one. - Migrating from Tauri: back up first and follow the migration guide.
- Default: the latest stable release asset named
- Configure. Pick a supported voice-cloning engine and acceleration that fit this device. Keep working defaults and install required dependencies.
- Verify. Start the app and check
/healthat the backend address (defaulthttp://localhost:3900); discover the API through/openapi.json. If the chosen engine's model shows"installed": falseinGET /models, state its size and license and get consent, then sendPOST /models/installwith that entry's{"repo_id": "…"}. PollGET /models/install/status: stop on afailedstate, and treat an emptyjobsarray as finished only onceGET /modelsreports"installed": true. Then generate a short clip with a bundled or authorized voice and confirm the audio file plays. - Report. Installed version, engine, actual device, data location, audio output path, and how to reopen the app.