Skip to content
Features

What MixAudit measures,
and how.

Built around a single premise: a producer mid-session doesn't need a score. They need to know what's wrong, why it matters, and what to do next.

34Measurements
30+Genre nodes
11Congestion bands
5Streaming targets
3–6Observations per analysis
01

Diagnostics

What the engine detects: spectral character, dynamic behaviour, spatial properties, and clarity. Observations are synthesised from these measurements.

01

Structured Observations

Every analysis produces 3–6 ranked observations, not a score. Each observation names the problem, explains why it matters to the listener, and describes a concrete fix. Critical issues surface first; informational notes last.

CRITICALWARNINGINFO
02

Genre-aware Thresholds

Loudness, dynamics, and spectral targets are calibrated per genre against reference datasets. A hip-hop mix and a classical recording are measured against different baselines. 30+ genre nodes inherit through a DAG so subgenres get accurate parent thresholds automatically.

30+ genresDAG inheritanceCalibrated against corpus
03

Spectral Tilt Analysis

Fits a spectral tilt model across 32 bands from 80 Hz to 12 kHz. Identifies energy holes and excess peaks within that range. Avoids false positives from extrapolation artifacts outside the calibrated window.

32 bands80 Hz – 12 kHzHole detection
04

Congestion Analysis

Detects masking and frequency congestion across 11 critical bands. When multiple elements compete in the same band, the observation names the likely culprits — kick vs. bass, vocals vs. guitars, synths vs. presence.

11 bandsMasking detection
05

Stereo Field

Measures mid/side energy balance, stereo correlation, and effective width across the full frequency range. Flags mono compatibility issues and excess width that will cause translation problems on earbuds and mono playback.

M/S balanceCorrelationMono compatibility
06

Spectral Trend

Tracks energy evolution across 4 frequency bands over time — not just a static snapshot. Detects brightness fade in the high end and bass swell in the low end as the mix progresses.

Time-varyingBrightness fadeBass swell
07

Reverb Wash & RT60

Estimates reverb tail density and decay time in three bands: low, mid, and high. Flags builds where excess low-end reverb accumulates into mud, or high-end wash smears transient definition.

RT60 by bandReverb wash score
08

Transient Density

Quantifies attack density, transient sharpness, and sustain shape. Distinguishes a flat over-limited signal from one with preserved punch. Reports impact score and onset rate relative to genre reference.

Onset ratePunch scoreAttack shape
09

Harmonic Distortion

Estimates total harmonic distortion by band across the low-mid and midrange. Separates analogue-style saturation from digital clipping.

THD by bandEven-order ratioClipping fraction
02

Loudness & Delivery

Loudness metering, dynamic range quantification, and delivery compliance. Every metric your master needs before it hits a platform.

10

Dynamic Range

Reports integrated LUFS, short-term LUFS, LRA, and RMS. Detects compression artifacts and loss of macro-dynamics that survive loudness normalisation.

Integrated LUFSLRARMS
11

Loudness History

Short-term LUFS plotted in 3-second windows with a 1-second hop across the full track. Shows where the mix loses energy or peaks unexpectedly.

3 s windows1 s hopFull track
12

Crest Factor

Peak-to-average ratio in dB. Separates heavily limited mixes from those with preserved headroom and punch.

Peak / RMSGenre-relative
13

EBU R128 Compliance

Checks the mix against EBU R128 integrated loudness, true peak, and LRA targets. Reports compliance status and exact delta to the target window.

-23 LUFS targetTP -1 dBTPLRA <= 20 LU
14

True Peak Detection

Inter-sample peak measurement per channel to ITU-R BS.1770-4. Reports maximum true peak and margin to 0, -1, and -2 dBTP.

ITU-R BS.1770-4Inter-samplePer-channel
15

Section Loudness

Measures LUFS, RMS, peak, and crest factor per time section. Flags sections that deviate significantly from the track median.

Per-section LUFSFlagged sections
16

Streaming Normalization

Reports integrated LUFS, true peak, and LRA against Spotify, Apple Music, YouTube, Tidal, and SoundCloud targets. Shows exactly how each platform will adjust your level.

SpotifyApple MusicYouTubeTidalSoundCloud
03

Stems

Analysis that requires individual stem files loaded in the macOS desktop app. The engine examines relationships between elements, not just the printed sum.

Stem analysis runs locally in the macOS desktop app. Detection uses signal-processing heuristics — spectral discontinuity and phase irregularity scoring — not trained machine learning models.

17

Stems Support

Desktop

Drop individual stem files — kick, bass, vocals, full mix — and the engine analyses masking, phase relationships, and element balance directly between them. Stem-confirmed observations carry higher confidence than stereo-only detections.

18

Masking & Congestion

Desktop

Identifies which stem pairs create masking at which Bark bands, scores severity per pair, and counts how many stems crowd a given frequency region.

Bark bandsPair-level scoring
19

Kick / Bass Relationship

Desktop

Analyses sub-overlap between kick and bass stems, centroid frequency of each, frequency separation, zero-lag cross-correlation, and minimum lag in milliseconds. Detects sidechain ducking.

Sub overlapCentroid separationDucking detection
20

Phase Alignment

Desktop

Cross-correlates every stem pair and reports lag in samples and milliseconds with a confidence score per pair. Catches time alignment issues between close-miked and room mics.

Lag in samples / msPer-pair confidence
21

Stem Balance

Desktop

Reports LUFS, crest factor, sample peak, and true peak per stem. Shows level offsets relative to the mix render.

Per-stem LUFSCrest factorTrue peak
22

Stem Distortion

Desktop

Measures THD percent and clipping fraction per stem independently. Identifies which stem is the worst distortion offender.

Per-stem THDClipping fraction
23

Stem Transients

Desktop

Estimates BPM from stem onset patterns, measures onset density per bar per stem, and reports per-stem attack time in milliseconds.

BPM estimateOnset densityAttack ms
25

Dynamic EQ Recommendations

Desktop

Generates per-stem dynamic EQ suggestions with specific frequency, Q factor, gain in dB, attack and release times, and threshold.

Freq + Q + gainAttack / releasePer-stem
04

Workspace

The interface layer: DAW-style navigation, visualisation tools, and workflow features built for iterative work across multiple revisions.

26

DAW-style 3-panel Workspace

Observations panel on the left, primary analysis view in the centre, detail inspector on the right. Hover an observation to highlight the corresponding region in the spectrum or waveform. Built for engineers who read dashboards, not consumers swiping through scores.

27

Waveform Navigation

Click anywhere on the waveform timeline to jump to that position in the audio player. Observations are mapped to time ranges so you can immediately audition the flagged section.

28

Spectrogram Heatmap

Time-frequency energy map rendered as a 128x120 STFT heatmap. Mid and side spectra rendered separately so stereo imbalances are spatially obvious.

STFTMid / Side128 x 120
29

Loudness Timeline

Short-term LUFS visualised as a continuous line over the waveform view. Hover a region on the timeline to see the exact LUFS reading at that point.

30

Revision Comparison

Load two versions of the same mix side by side. The comparison view shows deltas in loudness, LRA, spectral balance, and which observations changed between revisions.

31

Reference Workflow

Load a reference track alongside your mix. The inspector shows spectral, dynamic, and loudness gaps relative to the reference rather than absolute targets.

32

Analysis History

Every analysis is stored locally with its full observation set and metrics. Browse past sessions, re-open any previous result without re-analysing.

33

Shareable Reports

Generate a read-only share link for any analysis. The recipient sees the full observation set, spectrogram, and metrics in the browser — no account required.

34

Offline Analysis

Desktop

The macOS desktop app runs the full analysis engine locally. No files leave your machine. No internet connection required after installation.

We use privacy-safe analytics to understand usage and improve product quality. You can change this later in Privacy Settings.