Freesound Forums

Hello,
I have been producing for around 9 months now (been drumming and playing piano for 13 years) and I cannot figure out how to separate the vocals from a song so that I might use the vocals in a remix. I am currently using LMMS as my DAW. I have spent hours searching different phrases to on google to find the answer, but am unable to resolve my issue. This is probably because I do not know the proper vocabulary. Any advice on software (preferably free because i am a broke collage student)or techniques to properly extract vocals would be greatly appreciated. I am looking for any information regarding this issue, even if it is just the proper vocabulary to search and find the answer myself.
Thank you for your time! I look forward to all ideas and comments. If I can return the favor please let me know I am always happy to help where I can.

PS If anyone wants to check out my soundcloud its http://soundcloud.com/m-c-franchise
Any criticism is appreciated, especially on how to improve my general sound quality.
Collaborations welcome, if anyone wants a sound I have used or to download a track for remixing purposes I will email it to you for free.

Generally you can't extract vocals from audio material, because it would require arttificial intelligence that recognizes vocals in audio stream as a separate content (like you distinguish vocals from other sounds so to speak).

On the other hand, thre are some practical situations in which you can partially get what you need. Everything depends on how the vocals were recorded in relation to surrounding sound environment. Usually, vocal is recorded in a way, that the "common part" (the same for left and right channel, makes strong, pointed presence) is greater than stereo part (different for left and right - makes the spatial effect). At the same time, instruments have greater difference between left and right chanmnel (to make them more spatial or locate farther), and thus - the common part of instrumentals is much smaller (so, the instruments do not go against with vocals, but support them). In pro recordings, solo vocals usually have strong central presence, because they are recorded via single microphones. Instruments and choirs are usually recorded via multi-microphone setups.

Simple audio plugins that extract vocals - look for mid/common part and reject the side part, or do the opposite when removing vocals from audio.

More advanced plugins - additionally try to figure out "what the vocal is" - according to harmonics (vocal is made of a fundamental and harmonics to it), spectra and dynamics - but as far I remember it does not works perfect (maybe something changed since I checked on that subject).

This is simple explanation.
Maybe something change since then.

Thank you soooo much!!! It seems like I will be able to accomplish my task with audacity. now I just have to download that and learn how to use it. lol

Depending on the original sample you may be able to extract vocal samples with a combination of noise removal and having filter/equalization. This rarely works effectively, but:

- If your original sample is just vocals over a bass line or instrumental in the lower register, you can virtually remove the bass from your sample with an EQ/Filter leaving only the higher mids and treble (where most vocal frequencies are)

- Using a noise removal effect, select a portion of the instrumental WITHOUT vocals as the 'noise profile' (if your audio editor has this capability) and then select the section with vocals and 'remove the noise' (in other words, remove the instrumental component)

Once again, these techniques only work in specific circumstances but can be very effective. In combination with some the other techniques already mentioned, you may be able to isolate the pure vocal component and with a little further optimization have sounding close to the original.

Hope this helps!