HERTZKIT / PRACTICAL GUIDE

Compare mixes without loudness bias

Compare versions with matched loudness, equivalent regions and a concrete goal. Learn what to listen for before choosing your mix.

On this page6 sections

By Marcelo Borasi · Updated October 29, 2026

A louder version may seem clearer or more powerful even when the useful change is small. Separate listening level from the change you want to evaluate.

Prepare comparable versions

Export before and after with the same start and end. Different leading silence will not be automatically aligned: the comparator switches tracks at the same playback time, not by matching transients. Align B manually in the comparator, or correct it in your editor.

Enable Choose excerpt in the comparator and drag Start and End on the waveforms. Loop excerpt repeats that selection. If the versions start differently, use Adjust B start, choose Earlier or Later and enter seconds. Fine-adjustment buttons refine the offset. Both recordings remain complete; hatched areas show where one has no audio. Loudness matching still uses the whole tracks.

In the A/B comparator, try the piano at two levels. It is the same performance: B is about 6 dB louder. Start playback and switch A/B with loudness matching on, then off. Keys 1 and 2 select A and B when you are not editing a field. On a phone, Go to comparison and Back to files connect the two panels.

Compare in HertzKit

  1. Open the A/B comparator and choose both authorised versions. You can also load two test files: the same signal at different levels.
  2. Loading measures each track’s integrated loudness. Once both are ready, choose A or B and press play. Both versions share the listening position.
  3. Keep Match integrated loudness (LUFS) enabled and use main volume for comfortable listening. Each version’s adjustment is shown in dB; it is not applied to your files.
  4. Choose A or B during playback to switch without moving to another moment. Tap the waveforms or use Play from to seek. The timeline includes both complete recordings after alignment. Their endings may differ; no audio is trimmed or stretched.
  5. Listen for ten or twenty seconds with one question: vocal consonants, bass definition, drum attack or chorus space. Stop, rest and repeat before deciding.

Matching is based on the whole track, not the listening position. The comparator attenuates the louder version and reduces both if peak headroom requires it. Disabling matching restores their original levels. Loop all repeats the complete timeline. Enable Choose excerpt to repeat a specific passage instead. For blind listening and a saved decision, open Delivery review and A/B comparison.

Matching is not mastering

Matched playback does not fix EQ, distortion or dynamics. Integrated loudness also does not mean every second has equal intensity. Compare equivalent regions and listen to local changes. Silence does not provide a valid loudness reference.

Exercise: export your original mix and a version with one small adjustment. Describe the result without saying “louder”. If no useful improvement is apparent, keep the original.

Check mono too

Use the mono sum in Mix review and listen for disappearing elements. Do not seek a perfect correlation number. A stereo effect may be intentional; judge how it works on intended playback systems.

Keep versions and an explained report. A repeatable decision is more useful than a sequence of untested changes.

References and scope

Technical references: ITU-R BS.1770 and EBU Loudness. Browser measurements are indicative.

← All guides

A comparison with decisions

The Delivery review and A/B comparison adds shared filters, mono and blind preference listening. The report keeps what you heard and what you will retain. It is not an ABX test or a statistical result.

Open Delivery review and A/B comparison →