Every library over a few thousand tracks has duplicates in it. The same record bought twice, an MP3 and a WAV of the same song, a re-download that landed under a slightly different name.
Most people underestimate how many they have, because the way they look for them cannot find them. And the ones who do find them often make things worse, because deleting a duplicate is not as simple as keeping one and binning the other.
Here is why both of those happen, and how to clean up without breaking anything.
Why sorting by filename finds almost nothing
The usual approach is to sort the library by title and scan down the list for repeats.
That only works if the same song was saved under the same name twice, and it almost never was. Consider two files:
Artist - Title.mp3
01 Title (Original Mix).wavSame record. To a computer, nothing about those two strings is similar. They sort into completely different parts of your library and you will never see them side by side.
Now add the rest of the real world: a store that prefixes track numbers, a promo with “(Clean)” on the end, tags that spell the artist differently, a file that lost its tags entirely and is called track01.mp3. Filename matching misses all of it.
So people scan the list, see a handful of obvious repeats, deal with those, and conclude their library is basically fine. It usually isn’t.
Matching by sound instead
The alternative is to ignore the name completely and look at the audio.
An audio fingerprint is generated from the sound itself — a compact signature of what the track actually is. Two files that sound the same produce the same fingerprint, regardless of what they are called, what format they are in, what bitrate they were encoded at, or how badly their tags were filled in.
That catches the cases filename matching cannot touch: the MP3 and the WAV, the re-download, the untagged file, the promo with a different name. This is the only approach that gives you a true count of what is actually duplicated in your library.
The part that actually matters
Finding duplicates is the easy half. Here is the half that costs people their libraries.
The copies in a duplicate group are not interchangeable.
One of them is the file you have been playing for two years. It has your cue points written into it. It is sitting in fourteen crates. It is the one that loads when you pull up that track mid-set.
The other is a stray. You downloaded it twice, it went into your library, and you have never touched it.
They sound identical. They are not equivalent. Delete the wrong one and you have not cleaned your library, you have damaged it: fourteen crates now have a gap where that track was, and the cue points that lived inside that file are gone with it.
This is the single most common way a well-intentioned clean-up goes wrong.
How to decide which copy to keep
Before deleting anything from a duplicate group, work through three questions.
Which copy has my cue points on it? Cue points are stored inside the file itself, so they belong to one specific copy, not to the track in the abstract. Losing that file loses the work.
Which copy is actually in crates? A file sitting in fifteen crates is load-bearing. An identical file that appears in none is not.
Which is genuinely the better file? This is where people over-think it. Yes, prefer the higher-bitrate copy, all else being equal — but all else is rarely equal. A pristine WAV you have never played is worth less to you than the MP3 carrying every cue you have set on that record.
Bitrate is a tiebreaker, not the deciding factor.
Doing it in MLT
Find Duplicates fingerprints the audio, so it catches the copies filename matching misses. Scanning is free and unlimited, on Mac and Windows, so you can find out how bad the situation is without paying anything.
- Scan your whole Serato library, or a single folder if you want to start small.
- Results come back grouped by sound, with the copy it recommends keeping already marked — and a plain-English reason for each one, so you are not guessing at why.
- Review the groups. Nothing is deleted on its own, ever. The tool prepares the list and you decide what is on it.
- Approve, and it can push the list into Serato for you to delete in one sweep.
On a very large library the scan can take a while — it is listening to every file, which is slower than reading filenames. Leave it running overnight if your collection is enormous. It will finish.
Before you start
Take a backup first. Any bulk operation on a library deserves one, and this one is deleting files. If something does go wrong, a backup is the difference between an inconvenience and a disaster. See how to back up your Serato library for what a real one has to include.
If tracks go orange afterwards because something moved during the clean-up, that is recoverable too — see why Serato turns tracks orange.
The short version
Filename sorting will not find your duplicates; fingerprinting the audio will. And once you have found them, the copy worth keeping is the one carrying your cue points and sitting in your crates — not automatically the one with the biggest numbers on it.
Find Duplicates is free to scan in Music Library Tools — Mac and Windows, Serato only. Download MLT and run a scan to see the real number.