Hi - I've recorded a 4-part vocal bit. Two vocalists - one doing the high part, the other doing the other 3.

Understanding the topic is worthy of a doctoral thesis, I'm wondering what some nice hints would be to separate the vocal parts from each other.

The low voice (not that low, around A 220Hz) and the high voice (C 512 Hz or so down to A 440 Hz) are distinguishable, but I have two voices that move in parallel between middle C and the F just above it that are awfully hard to separate audibly. Part of the problem I know is my brain/ears (I have the hardest time hearing George Harrison's harmonies in the old 3-part Beatle recordings), but I think part of it may be that the same person is singing the two parts in a very close range, so it's going to be a hard task anyway.

I've been thinking about treating each to a different EQ - cut out different frequencies from each so they sound different. Is this a good idea? And if so, I haven't the slightest idea how to figure out the best frequencies to cut (I've been doing it randomly, little better than a monkey with a typewriter ... I'd like to develop better skills than that!). Is there a way to analyze each voice to see what the best EQ dip would be?

Thanks for any help ... another newbie question



--------------------
Alexis -Cubase 6.5.0/SX3.1.1.944, XP SP2, 4GB RAM (1GB not accessible, but used just to balance the computer so it doesn't tip over); Delta 66 in Omni i/O Studio; Motif8; UAD-1

You're overthinking this.

The best way to EQ (and other processing) harmonised BVs is as a whole so they sit as part of the mix.

--------------------
An Eagle for an Emperor, A Kestrel for a Knave.

Quote The Elf:

---

You're overthinking this.

The best way to EQ (and other processing) harmonised BVs is as a whole so they sit as part of the mix.

---

That makes sense, thanks The Elf. But you don't eq the individual voices to sit better with each other as you would, say, two guitars, etc.?

Thanks again!

--------------------
Alexis -Cubase 6.5.0/SX3.1.1.944, XP SP2, 4GB RAM (1GB not accessible, but used just to balance the computer so it doesn't tip over); Delta 66 in Omni i/O Studio; Motif8; UAD-1

If I were working on close harmonies for folk, or barbershop, then I may consider separation, but it's just about about using your ears to suit each voice, not calculating frequency ranges.

For harmonies designed to support a full pop/rock mix, or the like, the fact that the BVs blend and homogenize is more often an advantage - in that case I wouldn't be looking for separation.

With harmonies the separation is primarily going to come from the voices and the harmony arrangement. The EQ is of little consequence.

All IME and IMO, or course!

--------------------
An Eagle for an Emperor, A Kestrel for a Knave.

Quote The Elf:

---

If I were working on close harmonies for folk, or barbershop, then I may consider separation, but it's just about about using your ears to suit each voice, not calculating frequency ranges.

For harmonies designed to support a full pop/rock mix, or the like, the fact that the BVs blend and homogenize is more often an advantage - in that case I wouldn't be looking for separation.

With harmonies the separation is primarily going to come from the voices and the harmony arrangement. The EQ is of little consequence.

All IME and IMO, or course!

---

Awesome The Elf, you are a Prince as well - thank you!

This is actually a folk/CSN&Y/almost barbershop-py break. As who is singing is not a variable I can control, and differential EQ of the individual voices likely not expected to help much, I guess unless I change the arrangement it is what it is.

I will revisit it, but not extremely soon - I burned out 1/4 of my brain cells just trying to write the four voices, then recording and then playing around at the DAW for what seems like geologic time (I wouldn't demean the term mixing to use it to describe what I do after tracking ) and my brain is now having a Roberto Duran moment, "No mas!".

Thanks again for sharing your perspective and experience!

--------------------
Alexis -Cubase 6.5.0/SX3.1.1.944, XP SP2, 4GB RAM (1GB not accessible, but used just to balance the computer so it doesn't tip over); Delta 66 in Omni i/O Studio; Motif8; UAD-1

I'm slightly wondering why separation is actually the goal here. I'd have thought the ideal would be to have all the parts blend as seamlessly as possible?

I believe CSNY recorded their harmonies all together around a single mic, so not much scope for separate EQ or dynamics treatment there!

If the aim is to hear four separate harmonies as distinct performance elements, then you need four different voices with different constyructions of fundamentals, formants and so on --sources that sound recognisably distinct.

The tonal complexity of the human singing voice is such that simple EQ isn't capable of making any significant difference to the 'recognisability' For that, you need a proper vocal processor that is capable of messing with the formants and so forth -- but without making the voice sound artificial (as simple pitch shifters do).

However, the normal role of backing vocals is not to be audible as distinct performances, but to gel together to create a pleasant effect that supports the lead instruments/vocals -- in much the same way that you don't hear (or want to hear) individual violins in a string section.

H

--------------------
Technical Editor, Sound On Sound