01STEP 1 OF 4
Decode
Drag in a WAV or MP3 — decoded to raw audio in the browser. The file never leaves your device.
Upload a bounce, choose references, and Attune measures the differences across punch, movement, tone, clarity, space, and mastering. This page shows what happens at each step.
Runs in your browser · your audio never leaves your device · no install
Drag in a WAV or MP3 — decoded to raw audio in the browser. The file never leaves your device.
A WebAssembly engine extracts 50+ acoustic metrics, separates 4 stems, and detects tempo and sections.
Each metric is placed on the statistical distribution of your chosen collection of pro references.
Gaps become a six-axis score and plain-language findings — each with a production move to try.
Lows controlled, or building up?
Nothing here is simplified for the page — these are the real measurements Attune extracts from every bounce, grouped by what they listen for.
A quiet intro and a long outro drag whole-track averages toward mush. So you choose the region: analyze the full mix, or drag a selection around any moment — the drop, a breakdown — and Attune scores only that. Crucially, your selection is compared against the matching section of every reference, not their whole tracks.
Every reference is profiled end-to-end. Compare your whole bounce against whole pro tracks.
Every reference also carries a drop profile — so your drop is judged against their drops, like-for-like.
Breakdown, chorus and intro profiles are being added to the reference corpus as it grows.
Attune never decides what "good" sounds like. It measures your value, then shows where it sits in the spread of real pro releases in the collection you picked.
Your value, in physical units. Low-mid energy and how dynamically it moves — measured on your drop, not a whole-track average.
Dropped onto the distribution. Where does it fall among the 142 references in this collection? The dashed line is their median.
Distance becomes the axis score. +1.4σ from the median → a “watch” finding. The wider the gap, the louder the finding.
Different isn't wrong. The score tells you where you sit versus the genre — it never tells you to conform. A big gap is a flag to listen, and decide whether it's a choice or a problem.
Beneath the six axes, each section is a 30-dimensional acoustic fingerprint — spectral bands, per-band stereo width, groove, dynamics, tempo. Attune projects the whole library into a 2-D map; your mix drops in as a single point. The nearest dots are your nearest references — that's how the matches are found.
Live in the reference library today — pan, zoom, recolour by genre, BPM or any single dimension, and pull up "more like this" from any point.
Open a finished analysis on a sample track — no upload, no signup. Walk the whole flow in your browser.
The score is the start. Two surfaces turn a flagged gap into something you can actually fix.
Each axis unfolds into the real measurement behind it — your trace over the reference's — plus the specific production move: sidechain EQ at 200–400 Hz, ease the bus comp, narrow the air band. Not "add clarity."
Stream a reference next to your mix with loudness auto-matched and any stem or frequency band soloed — all on-device. So what you hear is the mix, not the master.
So a comparison is honest, here's exactly what Attune aligns before you A/B — and what it deliberately leaves alone.
New collections, features, and mix-down tips — about once a month. No spam.
The DSP pipeline runs locally in WebAssembly — your audio file stays in your browser. The reference library lives server-side as pre-computed metric distributions; only the analysis numbers travel up — to find your collection, and to keep those distributions current.
The score names the question. Your ears learn the answer.