MasteringPro — AI Audio Mastering

Browser-based mastering console by Hardban Records Lab. All DSP (EQ, multiband compression, limiting, mid/side, saturation, metering, offline render) runs client-side via the Web Audio API. A small backend service adds the AI features: chain generation, QA validation, reference matching, and the conversational Copilot.

┌──────────────────────────┐       HTTPS POST /api/*        ┌───────────────────────────┐
│  Frontend (Vercel)       │  ──────────────────────────▶   │  API (VPS, Docker)        │
│  Vite + React + shadcn    │                                │  Node / Hono proxy        │
│  Web Audio DSP engine     │  ◀──────────────────────────   │  → OpenRouter (AI models) │
└──────────────────────────┘        JSON config              └───────────────────────────┘
   masteringpro.hardbanrecordslab.online    api.masteringpro.hardbanrecordslab.online

Layer	Tech	Hosting
Frontend	Vite, React 18, TypeScript, shadcn/ui	Vercel (static SPA)
Backend	Node 20, Hono	VPS (Docker; host nginx terminates TLS)
AI	OpenRouter (OpenAI-compatible)	free models first, cheap paid fallback

On the HBRL VPS, host nginx + certbot already handle TLS, so deploy the API with docker-compose.vps.yml (API container only, published on 127.0.0.1:8787). docker-compose.yml + Caddyfile are for a standalone host. There is no database — session history and presets live in the browser (localStorage).

Repository layout

/                          Frontend (Vite app)
├─ public/                 Brand assets: logo.png, favicons, PWA icons, manifest
├─ src/lib/aiApi.ts        Client for the backend API
├─ vercel.json             SPA rewrites + asset caching
├─ server/                 Backend service
│  ├─ src/                 Hono app: /api/ai-mastering, /api/ai-copilot
│  ├─ Dockerfile
│  └─ .env.example         Backend config (AI key, model chain, rate limits, CORS)
├─ docker-compose.vps.yml  API container only — for the HBRL VPS (nginx does TLS)
├─ docker-compose.yml      Standalone-host stack: api + Caddy (auto HTTPS)
├─ Caddyfile
├─ .env.example            Compose config for the Caddy variant (API_DOMAIN, ACME_EMAIL)
└─ DEPLOYMENT.md           Step-by-step deploy guide

Local development

Two processes. Backend first:

cd server
cp .env.example .env          # set AI_API_KEY (get one at https://openrouter.ai/keys)
npm install
npm run dev                    # http://localhost:8787

Frontend:

npm install
npm run dev                    # http://localhost:8080  → auto-targets localhost:8787

No frontend .env is needed locally. Set VITE_API_URL only for deployed builds.

Deploy

See DEPLOYMENT.md for the complete guide. On the HBRL VPS the authoritative runbook is HBRL-VPS/VPS-ACTION-PLAN.md → "Wdrożenie MasterPRO". Summary:

1. VPS (API) — point api.masteringpro.hardbanrecordslab.online at the server, fill server/.env (or Infisical), then docker compose -f docker-compose.vps.yml up -d --build. Add an nginx vhost + certbot for TLS.
2. Vercel (frontend) — import the repo, set VITE_API_URL=https://api.masteringpro.hardbanrecordslab.online, deploy. Add the final frontend URL to ALLOWED_ORIGINS in server/.env and re-run docker compose -f docker-compose.vps.yml up -d.

Scripts

Command	What it does
npm run dev	Frontend dev server (Vite)
npm run build	Production build → dist/
npm test	Vitest unit tests
npm run lint	ESLint
cd server && npm run dev	Backend with --watch