A second opinion on your mix — before it leaves the studio.
The kind an experienced engineer gives on first listen — turned into a score, a list of specific problems, and the fixes for the ones we can fix for you.
Files used to compute your score, then deleted on schedule. Analysis rows stay — your audio doesn't.
ITU-R BS.1770-5 loudness, 4× true-peak, published streaming targets. Every number ties to a public standard.
Same input, same output, every time. The chain is biquad + FIR EQ, glue compression, true-peak limiting — not a model.
Your mix sounds great in the studio. Will it survive the real world?
A mix is release-ready when it still sounds right outside the room you made it in. On a car stereo. On a pair of earbuds. On a friend’s phone speaker. In a club.
Most mixes aren’t there on the first pass. The kick disappears in the car. The vocal turns harsh on phones. The bass gets weird on Bluetooth speakers. MixLens listens for the things that usually cause that — and tells you exactly what they are.
Five things we check on every mix.
The same things an experienced engineer hears in the first ten seconds — translated into plain English.
How loud the mix is overall, and whether the loudest moments will distort once the file gets compressed for streaming. Streaming services adjust every track to the same target — go too far over and they turn you down, ungracefully.
Whether the low / mid / high frequencies are in proportion. Too much bass makes a mix muddy. Too little air makes it sound dull. Hard to hear on the speakers you mixed on — your room lies to you.
Whether the bass and sub are working together or fighting. A common failure: the kick and the bass occupy the same frequency, and the whole low end turns into mud — especially on small speakers.
Whether the stereo image is healthy — and whether the bass is mono. When low-end ends up out of phase between left and right, it disappears the moment something plays in mono: phones, club PAs, Bluetooth speakers.
Whether kicks and snares still punch, or whether you’ve squashed everything flat with a limiter. A flat mix never feels alive — and you can’t add the punch back later.
Diagnostics first. Final polish second.
Final-polish tools aim at a louder, polished bounce. MixLens aims at release readiness: measurable reasons a mix fails in mono, on earbuds, after codecs, and in the car — with named metrics behind every warning. Many people use loudness tools for the final file and MixLens to debug what still sounds wrong upstream.
If your mix already translates, you do not need another pass. If it does not, this is the quicker way to find the real problem before you hand it to a plugin or an engineer.
- →You see what is off (phase, tonal balance, true peak, dynamics) — not just a new file to A/B.
- →You keep mix decisions; we suggest fixes and only render when you opt into corrective steps.
- →Built for the “why does this fall apart in the car?” moment — before you ship the file.
We tell you what’s wrong — and let you fix it with one click.
Once we’ve measured your mix, we run it through a list of well-known mix problems. Each one is a specific pattern — “the bass is out of phase below 150 Hz”, “the mix is louder than streaming targets”, “the kick has nowhere to breathe through the limiter.”
When a pattern fires, you get a card explaining what it is, why it matters, and the exact parametric settings — bell freq, comp threshold, limiter ceiling — you can dial into any DAW. Want MixLens to do it for you? Toggle Auto-fix and hit Preview to audition the change on a 10-second slice in the browser, then Render when you’re happy.
No AI guessing. Every score traces back to numbers you can see in the Diagnostics tab.
Below 150 Hz, your left and right channels are pulling against each other. On anything that plays in mono — phone speakers, club PAs, most Bluetooth speakers — the bass collapses or disappears.
Lock everything below 120 Hz to mono on the mix bus, before any saturation or limiting. Stereo image up top stays intact.
A calibrated band — not a number to chase.
A raw 0–100 invites people to chase 100 like a probability. It isn’t one. We surface a band instead: Ready, Almost, or Needs work — calibrated to the platform you’re shipping to.
Multiple issues. Don’t ship yet.
One or two specific fixes — listed in the report.
Holds up against your target platform’s spec.
Same mix, scored against every platform’s spec.
A mix at −10 LUFS / −0.5 dBTP is great for club, marginal for streaming, and broken for broadcast. One score can’t say all three — a matrix can. MixLens runs the same rules engine against each target so you see exactly where the master is ready and where it needs more work.
Loudness + true-peak rules anchor to whichever target you’ve set on the project; tonal balance, low-end, stereo, and dynamics stay platform-agnostic because they’re mix-quality signals, not delivery specs. Every threshold has a citation in the methodology page.
What the numbers mean
Every metric you'll see in MixLens ties back to a published standard. Quick primers below — each links to a deeper post or the full methodology page.
Perceptual loudness measurement (ITU-R BS.1770-5). Every streaming platform normalizes to it.
Inter-sample peaks that lossy codecs can clip on. Why every spec asks for ≤ −1.0 dBTP.
Whether your mix still sounds right outside your studio — phones, cars, earbuds, PAs.
How to actually use a reference track (it's not a copy target — it's a measurement tool).
What the −14 LUFS / −1 dBTP spec actually means for your master and how to hit it.
The published thresholds MixLens checks against — per platform, per genre, with citations.
Answers to the questions producers actually ask
What is release readiness?⌄
A mix is release-ready when it still sounds right outside the room you made it in — on a phone speaker, a car stereo, earbuds, a Bluetooth speaker, a club PA. MixLens measures the most common reasons mixes fail to hold up (mono compat below 150 Hz, low-mid buildup, true-peak overshoot, LUFS too loud or quiet, dynamic-range crush) and surfaces a calibrated band — Ready, Almost, or Needs work — for each platform you might ship to (Spotify, Apple, YouTube, club). The loudness and true-peak rules anchor to your chosen target so a club master at −9 LUFS isn’t penalized for not being −14.
What is LUFS, and why does every streaming platform use −14?⌄
LUFS (Loudness Units relative to Full Scale) is a perceptual loudness measurement defined by ITU-R BS.1770-5. It uses K-weighting (a curve that mimics the ear's mid-frequency sensitivity) and a gating algorithm that excludes silence. Spotify, Apple Music, YouTube, Tidal, and SoundCloud all normalize streams to roughly −14 LUFS so different tracks play at similar perceived volume. Master quieter than −14 and the platform turns you up; master louder and it turns you down, but you've already lost transients to compression.
What is true peak (dBTP) and why does it matter for streaming?⌄
True peak measures the peaks that occur between sample points when the signal is reconstructed by a DAC or a lossy codec (Spotify Ogg, Apple AAC). A sample-domain peak of 0 dBFS can have inter-sample peaks well above 0 dBTP. Every major platform asks for a −1.0 dBTP ceiling so its codec doesn't clip your master.
Why does my mix sound different on phone speakers?⌄
Almost always one of three things: (1) the low end is out-of-phase below 150 Hz and disappears on mono playback, (2) there's too much energy in the 200–500 Hz range, which sounds boomy on small speakers, or (3) the vocal isn't forward enough in the mix. MixLens checks all three and tells you which apply to your mix.
Does MixLens use AI to master my track?⌄
No. Every score traces back to a named metric (LUFS, true peak, band PSD share, stereo correlation, etc.) and every recommendation traces back to a named rule firing on that metric. The corrective rendering is deterministic DSP — mono-below crossover, low-mid shelf cut, biquad / FIR matching EQ, bus glue compression, loudness normalize, true-peak limiter. Same input, same output, every time. No training on your audio.
Is my audio kept private?⌄
Yes. We don't train on uploaded audio, we don't fingerprint it, and we don't resell it. Default retention is 24 hours on the free plan, 30 days on Pro. After that the audio file is deleted from storage; the analysis row stays so your score history is intact.
How is MixLens different from a plugin?⌄
Plugins react to whatever you're playing through them. MixLens analyzes the entire bounce against published streaming standards (Spotify, Apple, YouTube, EBU R128) and tells you specifically why a track is or isn’t release-ready. It also renders a corrected version deterministically — the same sort of chain an experienced engineer would apply, automated.
For the person who records — and isn’t a mix engineer.
You know your way around a DAW. You can hear when something’s off, but can’t always say exactly what. You want a sanity check before you upload to Spotify, send to a label, or play it for a friend who’ll be too polite to tell you it sounds muddy.
- Solo producers and recordists
- Bedroom & home-studio mixers
- Artists self-releasing on streaming
- Anyone without a trusted engineer on speed dial
- Working release engineers
- Pro mix engineers with full studios
- Replacing trained ears entirely
If you’re a working engineer, your ears already do this. But if you’re still building those reflexes, MixLens points at the same things — with the receipts.
Methodology & library versions →
Exactly which DSP runs, what specs we follow (ITU-R BS.1770-5, RBJ biquads, 4× true peak), which libraries (pyloudnorm 0.1.1, scipy 1.14.1), and where every target number comes from. Open the receipts.
Drop a bounce. Get the score, the diagnostics, and the fix — before you ship it.
Free forever for analysis. Three corrective renders a month included. The file you upload is used only to compute your score and your fixes, then deleted on schedule — your audio is yours.