Master Analysis
Upload an audio file to analyze loudness, peak levels, dynamic range, and frequency distribution. All processing happens in your browser.
Glossary
What every number in this tool means — in plain language.
Most of these numbers are in decibels (dB) on a scale where 0 is the absolute maximum a digital file can hold. That means the useful values are negative: a number like −14 sits 14 dB below the ceiling. More negative = quieter, with more headroom to spare. Closer to 0 = louder, with less safety margin. So “lower” (more negative) usually means safer, not worse.
- Integrated Loudness (LUFS)target: −16 to −13 LUFS for most streaming
The average perceived loudness of the whole track, measured the way human hearing weights it (ITU-R BS.1770).
Streaming services normalize everyone to a common loudness, so this decides how loud your track ends up next to others — and how much they'll turn it up or down.
Measured in LUFS below the 0 ceiling, so it's negative. −14 LUFS is louder than −20 LUFS.
- True Peak (dBTP)target: −1 dBTP or lower
The highest level the signal actually reaches between samples, found by reconstructing the waveform (4× oversampled).
Lossy encoders (MP3/AAC) and conversions can push peaks higher than the raw samples show. Leaving headroom prevents clipping and distortion after your track is encoded.
0 dBTP is the ceiling. Staying at or below −1 dBTP is the safe target; above 0 means it's already clipping.
- Loudness Range (LRA)target: about 4 to 12 LU for most music
How much the loudness varies across the track, in loudness units (EBU Tech 3342).
It's a quick read on dynamics over time — whether the song breathes (quiet verses, big choruses) or stays at one level throughout.
- Dynamic Range (DR)target: about 8 to 14 DR for modern music
The distance between the loudest peaks and the average loud level, in dB (TT/PMF method).
It captures the 'loudness war' tradeoff: heavy limiting raises loudness but flattens punch. Higher DR generally means more impact and clarity.
- Crest Factortarget: 6 dB or more (higher = punchier)
The gap between the peak level and the average (RMS) level, in dB.
Another view of how 'squashed' the master is. A bigger gap means sharper transients survived the limiting.
- Sample Peak (dBFS)
The highest individual sample value, before reconstructing between-sample peaks.
It's the classic peak meter reading. True Peak is the safer number to trust for streaming, since it catches inter-sample overshoots this misses.
0 dBFS is full scale; negative means below it.
- Max Momentary (LUFS)
The loudest 400-millisecond moment in the track.
Shows the most intense instant — useful for spotting a chorus or drop that spikes well above the overall average.
- Max Short-Term (LUFS)
The loudest 3-second window in the track.
A slightly longer view than momentary — closer to how a sustained loud section actually feels.
- Stereo Widthtarget: about 30% to 70%
How much energy sits in the sides versus the center of the stereo image.
Width adds space, but too much can make the mix collapse or sound thin when played in mono (phone speakers, clubs).
- Phase Correlationtarget: +0.5 to +1, and it should rarely dip below 0
How well the left and right channels agree, from +1 (identical/mono) through 0 (very wide) to −1 (out of phase).
Negative values mean parts of the mix cancel out when summed to mono — which happens on lots of playback systems. We read it in ~100 ms windows across the whole track (not one average), so a brief out-of-phase section shows up even when the overall number looks fine.
- Clipped Samples
The count of samples that hit the absolute ceiling and got flattened.
Clipping is hard digital distortion. Even a few clipped samples can add audible harshness, especially after lossy encoding.
- DC Offset
A constant shift of the whole waveform away from zero.
DC offset wastes headroom and can cause clicks at edits. It's usually tiny; a noticeable amount points to a hardware or plugin issue.
- Frequency Balance
How the track's energy is spread across the frequency bands, from sub-bass to treble.
It's the tonal fingerprint of your master — a creative choice, not a right-or-wrong score. It's just useful to know, so you can hear how that balance carries across different speakers if you want to.
- Brightness (Spectral Centroid)
The 'center of mass' of the frequencies present — a single number for how dark or bright the track sounds.
A quick tonal fingerprint. Useful for comparing your master against references in the same genre.
- Streaming Targets
Two separate checks per platform: how your loudness compares to its normalization target, and whether your true peak clears its ceiling.
Platforms turn everyone toward a common loudness, so being off-target mostly changes how much they adjust you — it's rarely a defect. The true-peak ceiling is the real gate: clear it and you're safe; exceed it and the platform's lossy encode can clip.
Color shows status: green = on target or peak-safe; yellow = your loudness is off, so the platform simply plays it a little quieter or louder to match other tracks (automatic, no quality lost); red = your true peak is over that platform's ceiling and could distort after encoding. Lowering your limiter ceiling to the strictest platform's value keeps you safe on all of them at once.