BokeBox

English

简体中文

English

简体中文

Theme

FAQ ​

Product ​

How is this different from “TTS reading an article”? ​

Content is rewritten into spoken structure (open, highlights, close) under your persona, not raw read-aloud.

Where is my data? ​

Default single-user private deploy — jobs, progress, media under local storage/.
Your configured AI provider receives API traffic for ASR / chat / TTS as needed.

License? ​

LGPL-3.0-only — https://github.com/vastsa/BokeBox

Install & deploy ​

Minimum env? ​

Node >= 22.5, pnpm 9.x — or Docker.

Can’t pull the image? ​

  • Network to ghcr.io
  • ./start.sh docker.local
  • In CN: ./start.sh docker.cn

Change port? ​

.env → PORT (default 8787). Configuration.

MCP URL wrong behind proxy? ​

Set PUBLIC_BASE_URL=https://your.domain.

Models ​

Must I use OpenAI official? ​

No. Any OpenAI-compatible Chat / ASR / TTS via OPENAI_BASE_URL.

Demo mode? ​

When critical models are missing, degraded/demo paths may run — check get_system_health and UI hints.

Jobs ​

Stuck on a stage? ​

  1. Job detail error text
  2. Matching model (ASR / Chat / TTS)
  3. Retry / re-run from that step
  4. Server logs

Only regenerate audio? ​

Re-run from TTS / synthesize, reusing the script if still present.

URL fetch fails? ​

Login walls / anti-bot → local upload or a custom Source plugin.

Plugins ​

Where do they live? ​

text
storage/plugins/{source|asr|tts|schedule}/<dir>/

Rescan in Settings or POST /api/...-plugins/rescan.

High-risk plugins off by default? ​

Yes when riskLevel: high.

MCP ​

Token leaked? ​

Rotate/reset under Settings → MCP; scrub client configs from public repos.

Docs site ​

bash
pnpm docs:dev

Language switcher: 简体中文 / English in the nav.

Settings map: Settings.

Still stuck? ​

Issues: https://github.com/vastsa/BokeBox/issues
Include version, run mode (Docker/pnpm), redacted logs.