A readable audit of what normalization did to your audio. Upload the file from before the change and the file from after it — a master before and after loudness normalization, a podcast pass before and after leveling, or any two stages of a chain — and the card measures both with a BS.1770-4 compliant meter, then lays the numbers side by side: integrated loudness, oversampled true peak, sample peak, loudness range (LRA) and duration, with the applied gain computed from the loudness delta. The waveform pair is drawn from the decoded PCM on a shared amplitude scale, so a level change or a squashed crest is visible at a glance, and a third-octave spectral comparison (long-term average spectra of both sides) shows which frequency regions gained or lost energy — a tip-off that more than pure gain was applied. The card ends with a verdict: faithful-gain when the change is a clean level move (loudness delta matches the peak delta and the loudness range survives), dynamics-compressed when LRA shrank by 3 LU or more, clipping-risk when the result peaks above 0 dBTP or 0 dBFS, and level-lowered / level-raised for plain downward or upward moves. Great for documenting delivery versions, proving a normalize pass behaved, or sharing an A/B with a client.