A mix analyzer built by an audio engineer, not a marketing team.

Most "AI mastering" tools are black boxes that hand-tune a few heuristics and call it intelligence. Unveil is the opposite: every measurement maps to a published broadcast standard, every reference curve is built from real commercial masters, and the source list is public.

What We Stand For

Four principles that drive every product decision.

Standards Over Vibes

Loudness is ITU-R BS.1770-4. True peak is 4× oversampled per Annex 2. Loudness Range follows EBU R128 / Tech 3342 with both gates applied. Where we don't meet a standard, we say so on the methodology page.

Published Reference Corpus

Genre reference curves are derived from a hand-curated list of commercially mastered tracks — and that list is published. Most analyzers ship hand-tuned heuristics and call them "AI." We measure real music and show our sources.

Actionable, Not Academic

Every fix tells you the exact frequency, exact dB change, and the exact plugin steps in your DAW. Localized to the worst-offending window in your track, not the whole song.

Built by a Producer

I am one person who got tired of mixes that sounded right on my monitors and wrong everywhere else. Every feature came from a real frustration in my own studio. If it doesn't solve a real problem I have, it doesn't ship.

The Stack

No mystery boxes. The engineering choices behind every measurement.

Loudness Engine

pyloudnorm implementing ITU-R BS.1770-4 K-weighted gating. Integrated LUFS, short-term and momentary loudness.

True Peak

4× polyphase oversampling via scipy.signal.resample_poly per BS.1770-4 Annex 2 — captures inter-sample peaks lossless TP meters report.

Loudness Range

EBU R128 / Tech 3342 — 3 s short-term blocks, 1 s hop, two-stage gating (-70 LUFS absolute + -20 LU relative), 95th − 10th percentile.

Spectral Analysis

audioFlux BFT (Hann window, linear scale) — 5–10× faster than librosa STFT on CPU.

Stem Separation

Spotify Demucs (htdemucs) — used for Fix My Mix biomimetic mastering and the Midify MIDI transcription pipeline.

Vocal Masking

Custom Bark-scale psychoacoustic masking detector. Constructs an adaptive M/S mastering chain based on vocal vs instrument energy overlap.

Genre Reference Corpus

Hand-curated list of commercially mastered tracks per genre. Per-band statistics (mean + standard deviation) drive z-score scoring with sigmoidal smoothing.

Infrastructure

Python FastAPI on Google Cloud Run, Firebase Auth + Firestore + Storage, signed URLs for audio artifacts.

The Team

Unveil is a one-person company.

David Corl

Founder · CEO · Designer · Engineer

B.S. Audio Production from Full Sail University and an Avid Certified Pro Tools Specialist. I built Unveil because I kept asking commercial mix tools the same question — "why does my mix sound off?" — and getting back generic spectrum pictures and AI-generated word salad instead of measurable answers.

Every line of code, every reference curve, every fix instruction — written, measured, or curated by me. If a feature exists, it's because I needed it at 2 AM in my own studio. If it's missing, I haven't found the time yet, not because I outsourced the decision.

Standards-grade analysis, free to try with a quick sign-up.

About the founder

David Corl is an audio engineer and the founder of Unveil. He holds a B.S. in Audio Production from Full Sail University and is an Avid Certified Pro Tools Specialist. Unveil is a one-person company: every measurement, reference curve and fix instruction is written, measured or curated by him.

Analyze a mix · Plug-ins · FAQ