Discover similar songs by musical content

Home » Example projects
Sometimes ID3 tag information in audio files is wrong or missing altogether. i-DeClone can examine music by content to discover variations like:

i-DeClone actually "listens" to the beginning of each song, then uses advanced spectral fingerprints to discover similar songs by the way they sound, regardless how they are called, tagged or encoded.

Situation
Multiple MP3 music downloads (also WAV,WMA,FLAC...)

Applies to
Music collections



Step by step instructions:

➀ Connect devices to scan

If you want to scan external disks, connect them, or just scan your PC folder contents. Click on Start scan toolbar button to begin. Then click Start new project to setup scan settings from scratch.

➁ Scan options

First setup the options for a 90% similarity search for Music, scanning your entire PC as seen in the previous example

Then click on Advanced property page and set it up as in the picture to the right. Most things are left at default settings: see how Length (duration) is the only property used for preliminary comparison. Make sure you tick Compare file content, then audio files that are close enough in duration will be opened and compared "musically"

For more thorough coverage, you can (optionally) untick Files must have same extension. This will discover similar songs saved in different formats (e.g. SOME-SONG.MP3 and SAME-SONG.WMA). Otherwise i-DeClone only compares items of the same type (SOME-SONG.MP3 and SAME-SONG.MP3)

All set, click Start scan and wait for the results.
  scan options dialog

➂ Mark and remove duplicates

Use the checkboxes to mark duplicate items for removal, then remove them to clean up space. Use Mark wizard to choose the originals (which will be kept) depending e.g. on their folder location. Finally click Clean-up button to start deleting the marked duplicates. This is a standard procedure explained in detail in the documentation

For varying quality music in particular, mark wizard has options that let you choose which version to keep, so you can go for bigger files (better quality) or for smaller files (not as good quality but taking less space)

There's quite a lot of math and CPU effort for calculating spectral fingerprints of audio files. Therefore only a small portion near the beginning of each song is compared, leaving the door open for false positives (e.g. when two songs have similar beginnings then diverge). Always examine the duplicate groups discovered before deleting files.

Try it out for yourself:   DOWNLOAD FREE TRIAL
©2021-2022 ZABKAT, all rights reserved | Contact | Privacy policy