Handling Peaking Frequencies During Mixdown.

[lassoharp]

I'm starting this thread for some general discussion on the topic of techniques for handling peaking frequencies of individual tracks during mixdown.  I'm always curious to hear how others approach this problem. 

In case my terminology is different than what others may use, I'm not referring to simple transients like a snare drum hit that's 10db too loud across the band.  What I'm talking about are the relatively narrow bandwidth peaks that are common on vocals, electric guitar, bass and piano & organ.

Some of the tougher problems are when there are multiples on say a lead vocal track.  You get these crazy peaks at maybe 500Hz on one section and then 1.25K on another, 300 etc . .  I've had good results in some cases such as this using a splitband compressor, and automating the thresholds if necessary.  If one of the peaks is very large and narrow I will often automate a dip of slightly less bandwidth and about 1/2 to 1/4 of the amplitude so as not to create too much of a hollowing effect.  I came across an interesting article by Tom Elmhirst describing the use of multiple deep notches on lead vocal.  I have found notch filtering most useful on sources like acoustic and electric guitar, but have found them somewhat limited on lead vocal tracks.

http://www.soundonsound.com/sos/aug07/articles/insidetrack_0807.htm



Rhodes & some Hammond organs are another notorious source of nasty peaks.  I forget the exact note but, starting there and past it the Rhodes produces very sharp and shrill transients that even good players can't always avoid generating.  I've found notching and split band compression very useful for these.

[MHanson]

Interesting read.  Thank you.

My first reaction was I don't think I've ever done that to a vocal but I then realized that I have.... 

I think in the times I've done big notches like in the article I was also compressing the vocal rather hard, bringing the vocal back together....

Like everything else in making music, you do what works ... at that moment in time ...

There are days that notching out something seems like the best and only way to do it.  It requires a really good eq that can do really tight bandwidths though (usually plugins work best for me in this regard). 

Here are the few ways I approach it...

If it's a stuck overtone (something always present at a constant volume), a notch works great.  If it only happens once and a while then I'll either start automating the depth of the notch or switch to frequency dependent compression.  Sometimes it's a combination of both automation and freq-compression.  If you have an actual dynamic eq like the BSS or something, that may work even better.  Lately I've been having fun with Sonnox SuprEsser.  I'm finding that I have great control with the frequency that I want to control and the compression time constants can be very quick and inaudible.

It's always a balance to make sure that everything sounds right when the peaking frequency is present AND when it is gone.  I guess, ultimately, I only want to alter a sound at the moment I have to and no more.  Meaning we're back to ... you do what works ... at that moment in time ...

In terms of your vocal example, I'd probably start with separating the sections of the song a little.  Most likely the 1k areas are in the chorus and the low freq problems are in the verses (I'm guessing).  So now you have 2 vocal tracks, verse and chorus.  Just play with it.  If it's just one word here and one word there I'd most likely just automate an eq for that ONE word.

Michael

[Michael Tibes]

I do notch filters like in the Elmhirst article on everything if it's necessary, including lead vocals. That's mostly with a static EQ. Those resonances can keep things from 'sitting in the mix', but cutting too much will obviously render things hollow and liveless. Especially on vocals mid range it can influence a lot of emotion, so it can be an 'aggression knob' as well. If the performance has a lot of dynamics a dynamic eq might be handy.

I wonder why you feel like it wouldn't work so well on vocals, maybe try another eq or listening environment? Cutting out the resonances in the room or speakers might be a trap to fall into?

Michael

[lassoharp]

I wonder why you feel like it wouldn't work so well on vocals, maybe try another eq or listening environment? Cutting out the resonances in the room or speakers might be a trap to fall into?

The vocal tracks that have had the biggest resonance problems that I've worked on were situations where there apparently were several going on at once and I wasn't able to solve them well enough with the high Q notching technique.  Compared to say a ringing overtone on an acoustic that cleans up quickly and nicely with a single notch.

Ah, room resonances.  Yes, always minor problems there that can show up.  I've had best results on that end by making sure monitor volume always stays in same place.  If that ever fails I switch to another room and speaker set until I've ruled out room effects.

The analyzer on Logic's channel EQ is pretty good at showing where problem areas are.

[ruairioflaherty]

Didn't read the SOS link but so often when I hear of people notching like crazy they are actually notching out resonances in their listening environment and speakers.

A sharp notch is very rarely the solution to any problem IME.  It is rare that the problem is so contained and specific. 

[lassoharp]

Didn't read the SOS link but so often when I hear of people notching like crazy they are actually notching out resonances in their listening environment and speakers.

A sharp notch is very rarely the solution to any problem IME.  It is rare that the problem is so contained and specific.

Good point.  The speaker/room issues are unfortunately many and sometimes hidden until the right combination of circumstances bring them forth - at least for most small or personal studios without the resources and experience of larger facilities.

 

[Gold]

I think simple is best. Try riding the fader first. Then add processing as necessary. Of course the real solution is to notice it as you are recording and fix it with recording technique.

[abbey road d enfer]

Notch is not my to-go tool for vocals, although it is for drums.
When I'm faced with a singer who has nasty frequencies, I would try to make him/her correct that, if possible; many singers today try to emulate the Autotuned vocals they hear on the radio and very often it makes them produce these nasty tones.
If it is not enough, I would EQ them out, but not as drastically as the mentioned article describes; I would leave some of the problem there, because it is part of the personality of the singer.
I think this article exemplifies the consequences of the plethora of available processing: 3 dynamics, and 3 EQ's on the vocals.
The sound they were trying to emulate was done with one compressor, one EQ and one reverb - you may argue thare was also tape compression.

[lassoharp]

the consequences of the plethora of available processing:

And possibly part the reality of modern DAW.  I think the level of accuracy of DAW reveals a good bit more than tape and it seems to make one have to work harder on the front end to iron out problems that tape may have been more forgiving on.  Other things like the singer turning slightly off axis seem much more noticeable on DAW and I've found myself having to do much more level tweaking,  sectional EQing,  and other processing than I did when working on tape and analog console.  I used to be able to shore up slight level differences on tape by buss limiting into something like 1176/78, but in DAW the compressors don't seem to do that in the same way and require intricate fader adjustments and sometimes brief EQing.

[pucho812]

dynamic eq's can work wonders in just such an occasion.

[usekgb]

You could always try very narrow band compression.  I tried this earlier this week on a saxophone, and it did wonders.  Just use a multi-band compressor plugin and set the bandwidth very narrow around the problem frequency.  Adjust the ratio and threshold to taste.

Cheers,
Zach

[dandeurloo]

I think the need for Notching so tight and so much has a lot to do with a number of things.  Cheaper mics, not tracking to tape which probably absorbed some of that info in a musical way as opposed to how a DAW just records what is there.  Also, more tracking happening in less then perfect rooms.  So with all of the above happening to tracks it the requires more drastic corrective processing to clean up the tracks. 

my 2 cents

[abbey road d enfer]

I think the need for Notching so tight and so much has a lot to do with a number of things.  Cheaper mics, not tracking to tape which probably absorbed some of that info in a musical way as opposed to how a DAW just records what is there.  Also, more tracking happening in less then perfect rooms.  So with all of the above happening to tracks it the requires more drastic corrective processing to clean up the tracks. 

my 2 cents

Well, that seems a perfect explanation for undercapitalized productions, but Amy Winehouse?!

[Kingston]

Well, that seems a perfect explanation for undercapitalized productions, but Amy Winehouse?!

I was wondering about that too, but picking the wrong mic for the task is almost too easy. That craft of recording is not exactly commonplace with producers today. Even if your paycheck is six to eight figures. It's still too expensive to find the old beard to place and pick those mics. Then weeks of sessions with too much printed effects. I can imagine also the heavy 1176 type limiting needed for her unstable voice. Hello nasty frequency peaks in all the wrong places! It's almost a school book scenario for the problem lassoharp is talking about.

[edit]

also, how much did that matter in the end. millions were made. just to put into perspective the level of detail we are talking about.

[ruairioflaherty]

I can imagine also the heavy 1176 type limiting needed for her unstable voice. Hello nasty frequency peaks in all the wrong places! It's almost a school book scenario for the problem lassoharp is talking about.

Pretty big assumptions there, I'm guessing you've actually never heard her sing.  She was at the top of her game around the making of that record.

[Kingston]

Pretty big assumptions there, I'm guessing you've actually never heard her sing.  She was at the top of her game around the making of that record.

Yeah I remember that game. It's only been a year. Did you see her live? To say the difference is drastic doesn't quite describe it. I can only imagine the amount of processing and takes they needed to get it to sound decent. Also have you actually heard the records? Technically it's a phase problem city, with the corrections they made easily audible.

But like I already said, none of that stopped it selling millions.

[abbey road d enfer]

I can only imagine the amount of processing and takes they needed to get it to sound decent.

Indeed; you can only guess.

Also have you actually heard the records? Technically it's a phase problem city, with the corrections they made easily audible.

Pray telling me how you hear phase problems?

But like I already said, none of that stopped it selling millions.

As usual, performance is all.

[Kingston]

Pray telling me how you hear phase problems?

Comb filters with deep notches aren't exactly subtle.

[edit]

you learn to recognise these mixing drums with ill matching overheads etc. room mics to each other and the rest of the drum mics. "what happened to my transients". The effect is not far removed from comb filters. Again, you should have hired the old beard before the mixdown phase. I'm sure you knew this already.

[abbey road d enfer]

Pray telling me how you hear phase problems?

Comb filters with deep notches aren't exactly subtle.

A deep notch involves almost 180° phase-shift. No decent equalizer produces that amount of phase-shift. Attributing that to "the corrections they made" doesn't seem correct to me.
Most of the Back to black album has been recorded live in almost uncontrolled conditions (let's get experimental, let's do anything, something great may come out of it), with all mics open in a situation where all musicians were close to each other. It seems it was a deliberate choice of Mark Ronson, who produced the album. I don't know much about him, but I have a feeling he's not too much concerned about the technical problems (spill, relative-phasing of sources) it generates.

[living sounds]

I usually find that heavy compression with nice analog compressors takes care of these problems. It may not sound very natural, but then they should have recorded it properly in the first place, shouldn't they.