Waveform
Master DNA
File DNA
Technical observations
These are descriptions of what was measured. Master DNA does not grade, score or judge a master.
How loud is this master over time? Momentary, short-term and gated integrated loudness measured to ITU-R BS.1770-4, plus the loudness range and where the quiet and loud material sits.
Loudness over time
Loudness distribution
Head / tail & noise floor
Loudness normalization
Gain is the difference between the reference loudness and this master's integrated loudness. The projected true peak assumes that gain is applied with no limiting.
How much does the energy move? Crest factor, RMS energy and peak behaviour, including the 4× oversampled inter-sample peak estimate and full-scale run detection.
Dynamic DNA
Crest factor = block sample-peak (dBFS) − block RMS (dBFS). LRA follows EBU Tech 3342 gating on 3 s short-term loudness.
RMS energy
Peak & true peak
Clipping indicators
A run of consecutive samples at or above −0.02 dBFS is counted as a potential clipping event. Consecutive identical high-level samples are counted as flat-topped regions. These are indicators, not a diagnosis of audible distortion.
How is the stereo image behaving? Correlation, mid/side balance, width and what happens when the master is actually summed to mono.
Stereo field
Width index = side RMS / (mid RMS + side RMS), as a percentage: 0% is a mono signal, 50% is equal mid and side energy. The vector scope plots real decimated L/R sample pairs from this file.
Mono compatibility
Mono loss = stereo block level − mono-sum block level. A cancellation region loses more than 3 dB. Mono preview plays this master actually summed to mono.
Correlation / phase
Where is the energy in the spectrum? Average spectrum, band distribution, spectral character over time and a time × frequency map of this master.
Frequency spectrum
Spectral peaks
| Frequency | Level | Prominence |
|---|
Tonal balance
| Band | Range | Average | Peak | Share |
|---|
Band levels are measured from the averaged power spectrum of the mid signal. Share is the band's proportion of total spectral energy between 20 Hz and 20 kHz. These are measurements, not targets.
Tonal balance curve
Spectral character
Flatness is the ratio of the geometric to the arithmetic mean of the power spectrum, measured per frame and averaged. Read it comparatively: lower means more tonal and peaky, higher means more noise-like. The scale is not calibrated so noise reads 1 — an unsmoothed periodogram of white noise settles near 0.56. Tilt is a least-squares fit of level against log frequency from 100 Hz to 10 kHz.
Spectral history
Low-end analysis
| Range | Average | Peak | Share of 20–300 Hz | Note |
|---|
High-end analysis
| Range | Average | Peak | Share of 5–20 kHz |
|---|
Where are the unusual technical events? Every entry below is a measured region of this file. Select one to move the playhead to it. None of these are faults by definition — they are places worth listening to.
Measured events
Event summary
Thresholds are fixed and stated in each section: true peak above −1 dBTP, correlation below 0, mono loss above 3 dB, sample runs at or above −0.02 dBFS, short-term loudness steps above 6 LU.
How does this master compare with documented references? There is no single streaming standard. Each service applies its own playback normalization, and listeners can often switch it off. These are reference figures, not mastering targets.
Streaming check
Broadcast reference
| Parameter | EBU R 128 | This master | Difference |
|---|
EBU R 128 is a broadcast delivery specification. It is shown as a reference for context, not as a target for music distribution, and a music master measuring away from it is not wrong.
All reference information
| Platform | Normalization behaviour | Reference loudness | True-peak guidance | This master | Source / status |
|---|
This table describes platform normalization behaviour, which is not the same thing as a recommended mastering practice. Playback loudness is also a listener-controlled setting on several services. Where no fixed figure is published by the platform, the row says so. Verify against each platform's own documentation before making delivery decisions.
Difference from reference
Positive means this master is louder than the reference and will be turned down by that service; negative means it is quieter and may be turned up where the service does that. No pass or fail is implied.
Where exactly does everything happen? One shared time axis. Click anywhere to move the playhead; toggle layers to compare only what you need.
Your audio is analyzed locally in your browser.
No audio is uploaded or sent to a server.