Since the music recordings are a mix of different tracks of individual instruments (solo voice, drums, bass, piano etc.), it is certainly possible to recover these separate tracks (technically called stems). Extracting the singing from the music is not only a way to create karaoke tracks made only with musical accompaniment, but also a way to better listen to the voice and transcribe it, or study the way in which the individual instruments are played or make remixes. Separating the voice from the music has always been a complex type of activity available only to experts or feasible only with expensive programs. But now, thanks to an innovative tool based on artificial intelligence, you can do it easily from your PC, with any music.
The Spleeter project, released as open source by Deezer, is capable of separating individual tracks or stems so as to be able to extract the voice from the musical accompaniment and also the individual instruments .
While Spleeter is a project published on Github that can be used via the command line (not therefore in a simple way), however, a completely free website has been released which exploiting the power of the algorithm of separation of the audio sources of Deezer, allows you to extract sung and the sound of individual musical instruments from any song.
Going therefore on the Moises.ai site (which we hope will remain online for a long time), it is possible to process any music for free and without limitations, starting from an MP3 file or from a Youtube video .
On the initial page of Moises you can immediately hear an example of a song that has been separated in musical accompaniment and sung. To start with a song of your choice, press the button to process a new song at the bottom and access the site using a Twitter account, Google or via Email. After logging in, you can select an MP3 file from your PC or use the search box to find the title of a song on Youtube. At the bottom you can also select the type of separation, if by extracting the sung from the musical accompaniment or even if you separate 4 stems ( bass, drums, voice and rest of the instruments ) or even 5 stems, to separate the voice and the played with bass, drums, piano and all the rest . The default option is that of the two stems, to extract the sung from musical instruments .
At the end of the processing, which can take a few minutes, you can listen to the separate vocal and accompaniment tracks and also download them to the PC. Everything works incredibly well, to try absolutely and immediately also because, at least for the moment, it is completely free.
Alternatively, in another article we wrote how to remove the voice from a song with Audacity or other programs
The Spleeter project, released as open source by Deezer, is capable of separating individual tracks or stems so as to be able to extract the voice from the musical accompaniment and also the individual instruments .
While Spleeter is a project published on Github that can be used via the command line (not therefore in a simple way), however, a completely free website has been released which exploiting the power of the algorithm of separation of the audio sources of Deezer, allows you to extract sung and the sound of individual musical instruments from any song.
Going therefore on the Moises.ai site (which we hope will remain online for a long time), it is possible to process any music for free and without limitations, starting from an MP3 file or from a Youtube video .
On the initial page of Moises you can immediately hear an example of a song that has been separated in musical accompaniment and sung. To start with a song of your choice, press the button to process a new song at the bottom and access the site using a Twitter account, Google or via Email. After logging in, you can select an MP3 file from your PC or use the search box to find the title of a song on Youtube. At the bottom you can also select the type of separation, if by extracting the sung from the musical accompaniment or even if you separate 4 stems ( bass, drums, voice and rest of the instruments ) or even 5 stems, to separate the voice and the played with bass, drums, piano and all the rest . The default option is that of the two stems, to extract the sung from musical instruments .
At the end of the processing, which can take a few minutes, you can listen to the separate vocal and accompaniment tracks and also download them to the PC. Everything works incredibly well, to try absolutely and immediately also because, at least for the moment, it is completely free.
Alternatively, in another article we wrote how to remove the voice from a song with Audacity or other programs
