BACK
Audio Source Separation
Splitting a mixed recording into its individual components, such as vocals, instruments, dialogue, or effects.
Audio source separation is the process of taking a finished, mixed recording and dividing it back into its constituent sources, such as vocals, drums, bass, dialogue, music, or effects. Historically impossible to do cleanly once a mix was committed, modern AI models can separate sources even from recordings that were never multitracked. It is the foundation for stem creation, remixing, remastering, dubbing, and many other workflows.
Related terms
Where this comes up at AudioShake