demorphus
A new export can change the audio a detector receives, even when the song sounds similar. First compare the actual files and reports. Then isolate encoding, trimming, loudness and processing changes. If the file is unchanged, investigate the analysis conditions instead of blaming the export.
Confirm that this is an export change
Keep both files and compare their hashes, durations and format details. The same SHA-256 hash identifies the same file bytes. Different hashes only establish a file difference: a tag edit can change the hash without changing decoded audio.
Listen at matching timestamps. A project can export the wrong range, leave a track muted or include effects you forgot were active. Check the final file outside the editing session before attributing a score change to a codec.
Inspect one export setting at a time
| Check | What to compare | Next action |
|---|---|---|
| Range and duration | Full mix, selected range, silence and fades | Export matching start and end points |
| Encoding | WAV, FLAC, MP3 and encoder settings | Keep a lossless reference and label encoded copies |
| Signal format | Sample rate, bit depth and channel count | Record conversions rather than changing several at once |
| Processing | EQ, limiting, gain, normalization and master effects | Compare a render with and without the suspected step |
| Detector | Service, date, model version and analysis scope | Use comparable reports before interpreting a difference |
Lossless storage and unchanged audio are different checks
FLAC preserves the audio samples it encodes; it cannot restore a previously compressed source. A lossless export can still contain EQ, gain changes or a different edit if those were applied before saving. Conversely, two different containers may decode to the same samples even though their file hashes differ.
Audacity documents the additional loss from re-encoding MP3. For production, preserve your source and avoid unnecessary encoding generations. For a format comparison, see Suno WAV versus MP3.
If the same file now gets a different score
Confirm that the reports refer to the same input, then compare dates, engine versions and any notices. Check whether the service used a preview or a different analysis route. A displayed song name is not enough to establish that the service fetched the same source.
If the cause remains unclear, keep the file, timestamps and complete responses for support. Do not state that the detector is random or that an encoder removed a watermark without evidence. Published detection research identifies robustness to transformations as an evaluation problem, but it does not diagnose your individual export.
Keep a clean comparison and a final report
- Archive the source and the export that first received a result.
- Return to the same project state and change only the setting you want to investigate.
- Save a separately named file and note the change.
- Compare sound before choosing the preferred candidate.
- Analyze the selected final export and keep its own report.
If the question is when to finish the sound, use the Cleaner and mastering order guide. If the selected output still receives an AI verdict, use the processing checklist. For a fresh baseline, open the free AI music checker.
Sources reviewed October 30, 2026. The comparison procedures are suggested workflows, not new benchmark results.
Prepared with AI assistance. Sources below.