Removing vocals from an audio track has long been the desire of musicians and remixers in order to obtain the instrumental version of a song, otherwise available only as a complete mix. The sections below explain how this task can be accomplished and some of the problems that can get in the way.
Is Removing Vocals Possible?
Yes and no. Unfortunately, there is no simple answer because it depends on many factors including what instruments and sound elements make up the non-vocal part of the audio signal and how these and the vocals have been processed and mixed into the final recording. In most cases, you can not expect to get results that are acceptable for professional use. However, in some cases, it is possible to render reasonable output useful in many, non-professional, scenarios. And, once in awhile, it is even possible to get excellent results.
A few software programs are capable of removing vocals from a stereo audio file automatically, eliminating the hassle of the multi-step manual procedure described in the following section. Some of these programs include.
While the automatic approach is easier, the manual approach can be accomplished in most audio editing software. It also provides insight into how the automatic vocal removal feature in most software works.
This method of removing vocals can be accomplished fairly easily, but depends greatly on how the entire mix was recorded. This method works when all or most of the non-vocal audio is mixed in stereo (some more to the left and some to the right) and the vocals are mixed in the center. Fortunately, this is common practice in popular music, making it a fairly reasonable method for many songs.
The resulting stereo audio file should now contain the same mix with all center-mixed audio cancelled from the mix. Hopefully, the entire instrumental contents will still exist, while the vocals will be canceled, or at least much more quite.
Problems Removing Vocals
Due to the complexity of many music and audio tracks, removing vocals using the method described above may result in poor or useless results. For example, when many of the background elements are also mixed in the center of the stereo spectrum and overlap or share the same frequency range with the vocals, the method above will produce distorted results that have lost much more than just vocals.
Additionally, effects, such as echo and reverb are often applied to vocals before they are recorded to the final mix. Unlike the commonly centered vocals, these effects tend not to be centered, making them difficult to remove. Additionally, the effects can distort and alter the vocals in ways that result in audio outside the normal vocal frequency range, making it difficult to "EQ out".
With some trial and error, you'll develop an ear for determining which audio tracks contain vocals that can be successfully removed and which do not. While the method described here does not always work, it is a fairly simple procedure that in many cases can render useful results.