FFmpeg for audio analysis. Rather than relying on a third-party API, the platform runs parallel FFmpeg processes server-side -loudness metering (EBU R128), 4-band spectrum analysis, stereo correlation via aphasemeter, and master detection heuristics. This keeps per-analysis costs near zero and gives full control over the metrics pipeline.
OpenAI for personality, not measurement. The AI layer interprets the raw metrics and generates the "roast" -translating "-8.2 LUFS with 4.1 LU range" into language producers actually understand. The technical analysis is deterministic; only the presentation is generative.
Next.js 14 on a single Hetzner VPS. No Vercel, no serverless. PostgreSQL, Redis, PM2, and Nginx on one box. This keeps infrastructure costs under €10/month while handling the current traffic comfortably.
Programmatic SEO at scale. 210+ pages across 11 content categories (genres, mixing problems, instruments, DAW guides, comparisons) -all generated from structured TypeScript data files with shared components. This drives organic discovery without manual content creation for each page.
Stripe Checkout with webhook-driven fulfillment. Payment confirmation triggers report generation and email delivery automatically -no polling, no manual intervention.