For a new project, cleaning before the final master is a practical starting workflow: you can finish the sound after processing. It is not a proven way to obtain a lower AI score. If you already have a master, compare alternatives from preserved source files and check the exact export you intend to keep.

Choose the order around the work you still need to do

If the mix still needs tonal or level decisions, keep those decisions available after Cleaner. If the master is already approved, applying Cleaner afterward creates a new version that needs another listening review. Do not attach the old master’s approval or detector report to the new file.

This guide concerns sequence. For the separate question of whether mastering changes detection cues, read mastering and AI fingerprints. Suno Remaster is also a separate workflow; it should not be silently substituted for a mastering pass in a comparison.

Workflow A: clean, then finish the master

  1. Archive the original download and use the original WAV when available.
  2. Process a working copy with Cleaner.
  3. Listen to the returned render before committing to further work.
  4. Apply the mastering decisions the song needs, preserving the intermediate file.
  5. Export, listen through the final master and analyze that export if a detection result is required.

The advantage of this order is practical: you retain a finishing stage for the processed sound. The possible cost is another revision if mastering changes a result you wanted to retain. We have not published a paired study establishing this order as the better performer.

Workflow B: master, then compare a cleaned version

  1. Keep both the unmastered source and the approved master.
  2. Send a copy of the master to Cleaner.
  3. Compare the output with the approved master at similar playback volume.
  4. Review vocals, cymbals, bass, attacks and the ending.
  5. If the output needs another edit, save that as a separate version and assess it again.

This is useful when the available starting point is an existing master. It does not preserve every mastering decision automatically. A lower AI score is not enough if the treatment changes the voice or balance beyond what you will accept.

Compare the two routes without mixing up their effects

For an exploratory comparison, branch both routes from the same source. Record each treatment and export setting. If you use the same mastering settings in both routes, label that a fixed-settings test. If you adjust each master by ear, label it a comparison of finished workflows: more than the order has changed.

Keep separately named files for the source, the Cleaner intermediate, the mastered intermediate and each final candidate. Use the quality comparison sheet before opening the detector reports. Write down the reason for choosing a version, including when neither is an improvement.

Test the delivered file, not an earlier intermediate

If you export another codec, trim the ending or change loudness after analysis, the earlier report identifies an earlier file. Use the export troubleshooting guide when a score changes. Audacity warns that repeated MP3 encoding adds loss; preserve a lossless working copy rather than cycling through MP3 drafts.

Keep listening and classification as two acceptance criteria. Compare Demorphus plans to choose processing access, or use the free checker when you only need to assess an existing export.

Sources reviewed October 30, 2026. The comparison procedures are suggested workflows, not new benchmark results.

Prepared with AI assistance. Sources below.

Sources and documentation