AI Music Detector / Reference
Detector changelog
Short version: the engine currently in production is build ensemble-1. This page lists every build, what changed in each one, and which published parameters moved. Detection results are only comparable when they come from the same build, so every measurement on this site is stamped with one.
Last updated August 2026 · engine build historical-ensemble-1
Why a detector needs a changelog
A probability from a detector only means something alongside the version that produced it. If the combination rule, the thresholds or the feature set change, an old score and a new score aren't the same measurement. Tools that publish one accuracy figure with no version history are asking to be trusted on something they can't actually show. Every number on this site carries the build it came from, and this page is where those builds are defined.
service-v1 · 2026-08
Current build. Classification now runs through a specialist third-party AI music detection service.
- Retired the in-browser ensemble engine in favour of a specialist third-party classification service.
- Audio is now uploaded over HTTPS to our server and forwarded for analysis; we never write it to storage.
- Kept the same probability and confidence reporting format, with results keyed by a SHA-256 hash for 14-day caching.
- Raised the file limit to 25 MB and supported duration to 15 minutes.
ensemble-1 · 2026-08
Retired build. Four detectors combined by weighted vote, with confidence reported separately from probability.
- Replaced the single acoustic detector with a four-model ensemble: spectral-artefact (weight 2.2), embedding (1.6), acoustic (1.0) and extended-features (0.9).
- Moved vote weights, thresholds and shrinkage into public/config/ensemble.json so the combination rule can be inspected.
- Added a disagreement penalty: when models disagree by more than 0.20, confidence drops instead of the probability moving.
- Capped the reported probability to 15-85% and pulled it toward 50% by a factor of 0.94.
- Added a measurement-driven explanation panel listing which signals moved the estimate and in which direction.
- 2026-08-24: first measured results published for this build — engine behaviour study 1 (168 detections, 12 reference tracks, 7 encode conditions) at /benchmark. Repeat determinism 0.0 pp, mean encode shift 2.4 pp, median analysis time 406 ms. No accuracy figure is claimed.
single-1 · 2026-07
Historical build (retired). One acoustic/spectral detector running entirely in the browser.
- Client-side analysis via the Web Audio API: spectral rolloff and ceiling, high-frequency energy, stereo correlation, dynamic range, transient statistics and loudness behaviour.
- Reported probability and confidence as separate outputs, with an explicit inconclusive band.
- Generated a downloadable PDF report in the browser from the same values shown on screen.
- No account, no upload and no server-side storage of audio at any point.
What this changelog will never say
- Changelog entries describe behaviour changes, never accuracy claims. An accuracy claim needs a measured benchmark run, and every result cell on /benchmark is still null.
- Because we never write audio to storage, no user audio is collected, so no build in this list was tuned on files submitted through the site.
- When the engine build changes, any published measurement gets re-stamped with the build that produced it rather than carried forward silently.
Where to go next
- Benchmark protocol — the metric definitions and corpus design each build will be evaluated against.
- Limitations — the failure modes no build in this list removes.
- Methodology — what the engine measures and how the signals are combined.
Cite this page
Quotation with attribution is welcome. Please keep the wording of factual claims intact and link back to the source page.
AI Music Detector. “Detector changelog.” Updated August 2026. https://aimusicdetector.com/changelog