You exported a track from Suno. The MP3 or M4A you just downloaded contains the audio you generated, plus a metadata block that names Suno, the model version, and the date — cryptographically signed. Spotify, Apple Music, TikTok and Tidal all read that block at upload. (Why this triggers AI labels and reach throttling.)
Stripping it takes 30 seconds. Here's the workflow.
What Suno Embeds and Why Platforms Read It
A Suno export contains, depending on the format you chose:
- C2PA manifest in a JUMBF box, signed by Suno's certificate, declaring the model and timestamp
- ID3 frames (MP3):
TENC(encoder) names Suno, sometimesCOMM(comment) carries the prompt - iTunes-style atoms in the M4A
udtabox: software string, encoder, sometimes the Suno track ID - Embedded artwork: usually Suno's default art, which carries its own metadata
Platforms read the C2PA assertion first because it's the highest-confidence signal — the AI tool literally signed a statement that this is its output. The ID3/udta fields are read as a fallback. Stripping all of them removes the easy answer; what's left is the audio classifier, which is far less reliable.
Three Things a Metadata Strip Closes (And One It Doesn't)
Closes:
- The C2PA manifest, completely
- ID3 / iTunes-style identifiers naming Suno or the model
- Any prompt or session ID written into comment fields
- Embedded artwork metadata if you also strip the artwork
Does not close:
- Audio watermarks. Some Suno model versions embed a near-inaudible watermark layer in the audio itself. That's not metadata. Stripping the file does not remove it. Removing it requires a re-encode through a DAW — typically a transparent compression pass, sometimes a notch filter at the watermark's frequency band.
- Your distributor's DDEX flag. If you check "contains AI-generated content" at DistroKid, CD Baby, Tunecore, etc., that flag goes to Spotify in a separate database feed. Strip the file and leave the distributor flag unchecked, or you've done half the job.
The watermark and the DDEX flag are the two things that defeat people who think file stripping is the whole answer. Be honest with yourself about which one is in your way.
How to Strip Suno Metadata in the Browser — Step by Step
- Open metadatacleaner.app in any browser. No login. No account.
- Drag your Suno track into the drop zone. Both MP3 and M4A work, plus WAV and FLAC if you exported lossless.
- Click Clean. The file gets processed entirely on your device. Nothing is uploaded. No server sees the audio.
- Click Download. You get back a file with byte-identical audio — same waveform, same length, same loudness — and an empty metadata wrapper.
- Verify if you want: drop the cleaned file into metadata2go.com alongside the original. The before-and-after should show the C2PA manifest gone, the ID3 fields empty, the udta box minimal.
- Upload the cleaned file to your distributor.
- At your distributor's upload form, leave the Contains AI-generated content checkbox unchecked.
That's it. The audio that hits Spotify is identical to the audio that came out of Suno. The wrapper around it is empty.
The free tier handles one file at a time. If you're processing an EP or an album, Pro is $4.99 a month for unlimited batch and ZIP downloads.
What Stays After the Strip: The Audio Watermark Situation
The watermark question deserves a paragraph of its own because it's the thing that makes creators think the strip didn't work.
Some Suno output (especially v3.x, intermittently in v4.0) includes an embedded watermark — a frequency-domain pattern, often in the 19 kHz range, designed to be inaudible but detectable. Platforms with audio classifiers can pick it up.
If your reach is being throttled and you've already stripped the file and unchecked the distributor's AI box, the watermark is the most likely remaining cause. The fix isn't a metadata tool; it's an audio-processing tool. Run the track through your DAW, apply a transparent re-encode (no destructive processing — just decode and re-export), or in stubborn cases apply a high-shelf cut above 18 kHz. There are open-source options for this; it's a separate problem from metadata, and the SEO plan calls for a future post on audio watermark removal specifically.
FAQ
Does stripping change the audio quality?
No. The audio bytes are not modified. Image stripping requires a re-encode through the canvas API (so the image is technically a fresh file), but audio stripping is in-place byte manipulation on the metadata containers — the audio stream itself is untouched. A spectrogram before and after will be identical.
Does this work for Suno's stems / multi-track exports?
Yes. Each stem is its own audio file with its own metadata block. Process each one. The file format dictates the strip approach — MP3 vs WAV vs FLAC — and the tool handles all of them.
Will the cleaned file still play in Suno's player?
It'll play anywhere — DAWs, music players, distributor previews, the original Suno interface. Removing the metadata doesn't affect playback in any way. Playback uses the audio data, not the metadata.
Should I also strip the artwork?
If you uploaded custom artwork, your cleaned audio will include it — but the artwork itself can have its own metadata (especially if it's an AI-generated image). For a paranoid pass: strip the artwork separately as an image file, then re-attach it to your cleaned audio in your DAW or distributor's upload flow. For most cases, just deleting the embedded art and uploading the bare audio is sufficient.
Does this work for Udio, Riffusion, and ElevenLabs too?
Yes. The same metadata containers apply (C2PA in JUMBF, ID3, udta) — different AI tools, same wrapper structure. The Metadata Cleaner workflow is identical regardless of which AI tool generated the audio. (What's actually inside any AI-generated file.)
Why doesn't Suno give me an option to export without C2PA?
Suno (and Udio, ElevenLabs, and the rest) are under increasing pressure from C2PA's industry consortium and from platform partnerships to ship signed manifests by default. There's no realistic near-term path to AI tools shipping unmarked exports — the incentive structure pulls the other direction. Stripping after export is the practical answer.
Strip the file, uncheck the box, deal with the watermark separately if it shows up. Open metadatacleaner.app and drop your track.