Artificial intelligence is definitely everywhere! This time it's to extract the vocal part of a song and remove all the instruments.
Acapella Extractor is a tool that is based on the machine learning algorithm to separate the vocals from the instrumental part. The service uses the open source Spleeter library developed by the Deezer research team. We had already talked about this library also used by the Melody.ml tool which does roughly the same thing as Acapella Extractor.
To use this service, nothing could be simpler, just go to the site and upload the song that you want to process. The tool works with MP3 or Wave files smaller than 80MB or 10 minutes.
The service will then grind for a few minutes before offering you a link to download the a cappella version of your song.
Of course the results obtained are not perfect, but frankly the tool works rather well!
For best results, the developer advises using acoustic versions of songs and avoiding vocals that use a lot of processing (autotune, distortion, etc.). Finally, he suggests using a high-pass filter, transforming the audio to mono, and adding a slight compression…
Finally, if you rather need to keep only the musical part by removing the vocal parts, you can take a look at Remove Vowels which works the same way!
Article updated on January 7, 2025 by Byothe
