Pass CD Baby's AI Check

CD Baby distributes to Spotify, Apple Music, Amazon and dozens of other stores, and those stores have grown strict about AI-generated audio — from fraudulent uploads to undisclosed synthetic tracks flooding catalogs. To protect its relationship with those platforms, CD Baby runs its own AI screening at intake, before your release ever reaches a store. The goal isn't to punish producers who use AI tools; it's to keep the pipeline clean and make sure anything that needs disclosing is disclosed.

The practical effect for you is that a track carrying a strong AI signature can be caught the moment you submit it. The scanner doesn't care whether your song is a polished, radio-ready record or a rough sketch — it's reacting to how the audio was built, not to how good it is. A well-arranged track generated or heavily processed by a neural model can score just as high as a throwaway one, because the tell is in the synthesis itself, not the songwriting.

That's the frustrating part for legitimate artists. You may have written the topline, directed the arrangement and mixed the record yourself, but if an AI model touched the audio in a way that left a measurable fingerprint, the intake scan can still flag it. Understanding what the check actually evaluates is the first step to getting a clean release through.

CD Baby's screening — like every serious AI-music detector — is a statistical classifier, not a human listening for a "robotic" sound. It has been trained on large sets of human-made and machine-made audio and learned the features that separate them. When your file goes through intake, the system extracts those features and returns an AI-probability score: a confidence level, usually expressed as a percentage, that the audio was AI-generated. It's not a simple yes/no; it's a number, and the distributor sets its own threshold for what counts as high-risk.

What is that fingerprint made of? It lives in several places at once, and none of them are audible in a normal listen:

• Spectral balance. Neural vocoders distribute energy across frequency bands in a subtly regular way, especially in the high mids and top end, where real recordings are noisier and less predictable.
• Phase relationships. Generated audio tends to have unnaturally coherent or unnaturally smeared phase between channels and harmonics — hard to hear, easy to measure.
• Micro-timing and transients. Note onsets and the tiny timing jitter of a real performance are reconstructed rather than recorded, so they carry a machine-regular signature.
• Noise-floor texture. The "air" between notes — room tone, breath, hiss — is synthesized, and its texture differs statistically from a real capture.

Because these traits are baked into the samples, they're what the intake scanner scores on. Getting under the threshold means changing those specific characteristics, which is exactly what the AI Cleaner is built to do.

Getting flagged is not the same as getting banned, but it's not nothing either. Depending on the case and CD Baby's current policy, a flagged submission can be held for manual review, rejected outright, or returned with a request to disclose AI involvement before it moves forward. A held release means delay — sometimes a costly one if you had a release date lined up with playlists or press.

More seriously, repeatedly submitting tracks that trip a distributor's AI or fraud filters can put your account at risk. Distributors take a dim view of anyone who looks like they're trying to slip synthetic or infringing audio past screening at scale, and account-level action is a real possibility for repeat issues. That's why it's far better to resolve a flag before you submit than to argue about it afterward — you keep your standing, and you keep control of your release schedule.

The most common mistake is assuming a fresh render will fix it. It won't. Re-exporting the track, bouncing it to a new file, converting WAV to MP3 and back, normalising the level, or slapping a limiter on the master all change the surface of the audio — but they leave the deep features the classifier relies on almost entirely intact. You're repackaging the same synthesized waveform, so the probability score barely moves.

The same goes for the usual "tricks" people trade online: adding a whisper of background noise, nudging the pitch a few cents, or layering a subtle reverb. These introduce surface changes that a robust detector was trained to see through, and some of them can even read as more suspicious. To actually lower the score you have to alter the specific spectral, phase and timing characteristics the model measures, and do it cleanly enough that you don't create new artifacts in the process. That's a targeted-processing problem, not a "render it again" problem — and it's the reason a purpose-built cleaner beats DAW guesswork. You can watch exactly where your track sits on the probability scale, before and after, with the free AI Checker.

artefactFX gives you the two tools that matter, in the right order. First, the free AI Checker reads your track and returns a clear AI-probability score with no sign-up, so you're never guessing whether CD Baby will flag it. If the score is already low, you're done — don't fix what isn't broken. If it's high, the AI Cleaner processes the file to break up the statistical fingerprint the intake scanner reads, targeting the artifacts without wrecking your sound.

You get the result back as a studio-quality 24-bit WAV — not a re-compressed lossy file — ready for mastering, plus a before/after score so you can confirm the fingerprint actually dropped before you upload anything. That before/after number is the whole point: it turns "I hope this passes" into evidence you can see. Compare plans anytime on pricing; you can start on the free plan and only scale up when you're shipping more releases.

The order you work in matters. The most reliable route from a finished track to a CD Baby submission that clears intake looks like this:

• Check. Run the file through the free AI Checker first. Start from the highest-quality source you have — a lossless WAV or FLAC, not a 128kbps MP3 — for both a more accurate reading and a cleaner result.
• Clean. If the score is high, send it through the AI Cleaner. If you have stems, clean those rather than the flat mix (more on that below).
• Re-check. Run the cleaned file back through the checker and compare the before/after score. Most tracks drop well below the high-risk line here; a minority stay higher depending on the source.
• Master, then metadata. Do your EQ, compression and loudness after cleaning, then fill in accurate titles, credits and ISRCs — mismatched metadata causes its own review holds.
• Disclose where required. If CD Baby's policy asks you to declare AI involvement, do it. Cleaning removes artifacts; it does not remove your obligation to be honest.

If you want to line up key and tempo for the master or a remix, the free BPM & Key finder reads both straight from the file.

Cleaning a finished stereo mix works, but cleaning stems works better. When every element is baked into one file, the processing has to treat vocals, drums, bass and synths together. Split them out and each part can be handled on its own terms — which matters because artifacts aren't spread evenly. Vocals, in particular, often carry the strongest AI signature of any element, so isolating and cleaning them individually usually moves the score the most.

If you have stems, upload them as a ZIP and you'll get each stem back cleaned, ready to re-balance into your own mix and master. If you only have the flat mix, cleaning that is still effective — it's just a coarser tool. For the full breakdown of the trade-offs, read stems vs the full mix.

One honest note, because it matters. artefactFX removes the acoustic artifacts that automated detectors score on — it does not remove any obligation you have to CD Baby or to the stores it feeds. This is a tool for fixing a synthesis fingerprint that's getting a legitimate release held or rejected, not a tool for committing fraud or evading a policy you agreed to. Where CD Baby requires you to disclose that a track uses AI, disclose it. Keep your use of any generator within that generator's terms. Cleaning changes what a scanner measures; it doesn't change the rules.

Used that way, it does exactly what a working artist needs: it stops an inaudible machine signature from blocking a release you have every right to distribute, while you stay on the right side of CD Baby's policy. If your issue is with a different distributor, the same approach applies — see our guides to DistroKid AI detection and TuneCore AI check, or the broader walkthrough on distributor AI checks.

artefactFX was built by people shipping real releases, not a generic audio utility. Detection uses professional AI analysis, cleaning targets the hidden fingerprint without wrecking your sound, and every result comes with a before/after score so you are never guessing whether CD Baby will flag your upload. Check for free, clean only when you need to, and submit with confidence.

It's also honest about its limits. We won't tell you every track will magically pass CD Baby's screening — most drop well below the high-risk line after cleaning, a minority stay higher depending on the source, and mastering afterwards lowers the risk further. You see the real numbers at every step, on your own files, and you decide when it's ready to submit. Compare plans anytime on pricing.