the frequency range in a song which only consists of vocals

[ritu]

I am an electronics engineering student and i don't know much about sound engineering.i am designing a vocal remover and i need some information.is there a frequency range (Common to all songs) which only consists of vocals ?? i require this frequency range because once i eliminate it from a song i would be left with the vocals which are in the same frequency range as that of the instruments.and i have a circuit designed to eliminate the vocals which are in the same frequency range as that of the instruments.

i am having trouble with finding this particular frequency range.i couldn't find any papers or any statistical data.if you know anyway information which you think might be helpful please do share it with me.

Also , is there any software which would let me to select a frequency range in a song and analyse it?to find out the range with no instruments and only vocals in it? i know it must be some kind of an equalizer.is any such software available in the net?? the basic idea is to find this frequency range for a number of songs and see if the results coincide .

 

[240v]

Here's a good resource to start off with. General Acoustics

Also, to find where vocals sit, you could try using an equalizer (forgot the type name) to strip out all but a certain range and go through all the frequencies I guess.

 

[Steenamaroo]

There is so much overlap that you'll never be able to strip everything except vocals from a song.

You might get close using cancelling techniques to leave any info that's common to L+R, but still, there's no foolproof method that I know of.

On top of that, Barry white is going to have a different range from Barry Gibb, for example.

 

[Massive Master]

On top of that, Barry white is going to have a different range from Barry Gibb, for example.

Barry Gibb was actually Barry White's fourth harmonic for a while if I recall.

 

[ritu]

@240v
thank you for the link .it gives the basic idea.and since its a project i need accurate data.and yeah ,like you mentioned ,the equalizer which would let me go through all the frequencies is the one i am looking for.and some people say that this range won't be common to all songs .and some say that you can't just analyse the frequencies of a song because you would get it as tracks..am not sure if that's the right term.well is it true??

 

[ritu]

There is so much overlap that you'll never be able to strip everything except vocals from a song.

You might get close using cancelling techniques to leave any info that's common to L+R, but still, there's no foolproof method that I know of.

On top of that, Barry white is going to have a different range from Barry Gibb, for example.

yeah i know about this cancelling technique . and yes its not perfect.am using a different technique altogether.and like i said i don't know much about sound engineering. so could you please explain your last point?

 

[Bristol Posse]

The point about Barrys Gibb vs. White is that since one is a bass and the other singing in a falsetto, even if they were singing the same song they would sing it in different registers and thus different fundamental frequencies

Aside from that every note has a different frequency and set of harmonics so even if you found a way to remove all of a sung A at 440 hz and all of it's harmonics, when the singer sings a different note then it's a whole different set of frequencies.

Some singers have a very narrow range some real rarity have a range of four octaves or more so you'd potentially have to scoop out a large amount of frequencies also used by the rest of the elements of the song

I'd be interested to see if you can pull this off though, sounds cool

 

[ritu]

The point about Barrys Gibb vs. White is that since one is a bass and the other singing in a falsetto, even if they were singing the same song they would sing it in different registers and thus different fundamental frequencies

Aside from that every note has a different frequency and set of harmonics so even if you found a way to remove all of a sung A at 440 hz and all of it's harmonics, when the singer sings a different note then it's a whole different set of frequencies.

Some singers have a very narrow range some real rarity have a range of four octaves or more so you'd potentially have to scoop out a large amount of frequencies also used by the rest of the elements of the song

I'd be interested to see if you can pull this off though, sounds cool

The point about Barrys Gibb vs. White is that since one is a bass and the other singing in a falsetto, even if they were singing the same song they would sing it in different registers and thus different fundamental frequencies

Aside from that every note has a different frequency and set of harmonics so even if you found a way to remove all of a sung A at 440 hz and all of it's harmonics, when the singer sings a different note then it's a whole different set of frequencies.

Some singers have a very narrow range some real rarity have a range of four octaves or more so you'd potentially have to scoop out a large amount of frequencies also used by the rest of the elements of the song

I'd be interested to see if you can pull this off though, sounds cool

i really hope i do .think about this.what if i could eliminate a particular frequency at that particular time without eliminating that frequency altogether? so the frequency getting eliminated would vary with the frequency of the vocals.so if the singer sings at different notes in the same song the frequency getting eliminated would vary accordingly each time.
The only chunk of frequency i would be eliminating is the frequency range that consists only vocals and no instruments.and that's what am looking for (with no luck )

 

[Bristol Posse]

The only chunk of frequency i would be eliminating is the frequency range that consists only vocals and no instruments.and that's what am looking for (with no luck )

I've yet to find a song that wasn't an acapella that had frequencies that are only vocals. that's a problem we run into when mixing, you have to find how to bring the elements together so the frequencies that cross over don't mask what is most important (usually the vocal) but without killing the overal sound
At any given moment in a typical song you might find bass, guitars, keys, vocals, Drums, percussion all with some representation at the fundamental frequency of the note that singer is singing
If you remove that frequency you just leave a gaping hole in everything that is represented there. If you turn that frequency down everything sounds wierd and the vocal is still there

Also consider that it's not just the fundamental frequencies you need to remove but the harmonics of that frequency too. Consider a telephone. Much of the fundamental frequency of many voices is not represented at all, the only thing that is heard or transmitted are the harmonics and yet we can still understand the voice and recognize the voice of the person speaking

 

[mixsit]

Except that even if the signal (a vocal) were one frequency, you filter it it removes any content at that frequency.
There is also a very good (typical) chance ther will be content in the form of echo/delay/reverbs spreading the siganl in both time and in the stereo field.

 

[Steenamaroo]

Barry Gibb was actually Barry White's fourth harmonic for a while if I recall.

Fact. ^


OP. Think about this the other way.
Imagine you're trying to 'watermark' some demo audio samples, so you stick a sine wave through them.

Now, I come along, smart ass that I am, and just cancel your sine wave out in Protools; Your samples are now mine.

So how would you stop me doing that? Maybe you use a variable effect like a chorus of flanger on that sine wav?
That's going to be much harder to cancel, and here's the reason.
It's complex and unpredictable, unless I know what tools/algorithms you used.

Assuming that someone somewhere can do that, let's go one step further.

Adding an intermittent vocal to your samples could be considered the ultimate security.
I'm never gona be able to cancel or surgically remove that. I could never work out the algorithm, cos there isn't one.
Instruments and vocals are incredibly complex that way, and unpredictable.

 

[240v]

There is so much overlap that you'll never be able to strip everything except vocals from a song.

That's why those frequency stripped crappy instrumentals on YouTube sound so bad! XD lol

 

[Jekan]

You could try and use a stereo-extracting plugin (e.g. Waves Center). Particularly in older songs the vocals and placed in such a way (stereo/mono) that you can eliminate them quite good. In combination with an equalizer it could get better but in most songs you're best off recording new vocals

Again as said, you have to be lucky to just pick that song on which these tricks work a bit. But in most cases you won't get the result you're looking for I'm afraid