Fascinating stuff. It’s like in every major post so far, there’s been one little trick I’ve never heard before and plan to try.
2 Likes
Thanks Emma!
Thanks John! Glad you liked it.
Wow, if I’d known you’d have to do so much work I would have played better. Seriously though, since you prefer raw tracks, would there be a benefit to me doing mild processing on the way in to make your life easier? For instance, mild compression for peak leveling, or is it better raw? I’m feeling like some of this could be pre production if you do it all routinely.
1 Like
That’s actually a really difficult question to answer categorically. Sometimes pre-processing can make life easier; sometimes it can tie your hands. A little mild compression and/or peak levelling would probably be fine on vocals and bass. I definitely prefer drums raw. Any added ambience is best to leave to the final mix too.
Part 6 - Guitars
Guitars were probably the easiest part of this mix for me - for 2 reasons:
- Bob is an awesome guitarist, and since 90% of the guitarist’s tone is in his fingers, Bob’s playing and tone are brilliant
- Bob knows how to record excellent guitar tones.
It was simply a matter of me getting out of the way and not messing it up. Having said that, as a guitarist of 41 years myself (many of which I spent being totally tone-obsessed), I have fairly firm ideas about how guitars should sound, and I think it helps that Bob and I are relatively on the same page in that regard.
Incidentally, when I reach the stage in the mix where I’m starting to mix the main supporting instrumental core (in this mix, guitars, drums & bass), I like to fire up the lead vocal and put it in as a “placeholder”. To get the vocal into relatively good shape in short order, I use Izotope’s Nectar Elements plugin, which contains all the processors you need to create a “finished”-sounding vocal in one convenient plugin. It’s just a matter of slapping it on and choosing a preset that will get me in the ballpark of the final sound I’m after. Then later, I return and work on the ‘final’ vocal tone.
It’s much easier to mix around the actual vocal than try to imagine it being there. I can hear clearly any potential frequency-masking problems, and it stops me from making the guitars take up too much space in the mix unnecessarily.
So what kind of shape were the guitars in when they came to me? Very good - here’s an example of one of the main rhythm tracks:
Remembering that I had already inserted Slates Virtual Tape Machine on each track and gain-staged it in analogue style, as I did on this one, I applied my reductive eq first, using the stock ReaEQ. This involved high passing a 105hz, and pulling out some mud and boxiness at 384hz and 572hz respectively. Next, I ran it into VMR for some polishing: I used the Neve preamp emulation first in line. I pushed the preamp input until I heard audible distortion, then rolled it back a way. It just added a nice bit of weight and thickness to the sound. Next into Slates SSL eq emulation, cutting more lows at 80hz, pushing some with a bell filter at 117hz, and a very gentle boost at 200hz. I also pushed the high mids at 6k with a wide, fairly assertive bump of almost 4dB. Next came some compression from Slate’s FG401 compressor - 4:1, slow attack, fast release, with about 2dB of gain reduction. That was followed by Slate’s “Revival”, pushing some low end saturation or “thickness”. Here is the resulting tone:
All in all, there is not a massive change in the guitar tone - it retains the same basic character, just slimmed down to fit within the mix better.
You might wonder why I bother to compress distorted guitars, since by nature they are already highly compressed. The answer is not to further reduce the dynamic range, but rather to accentuate the dynamics…sounds strange I know, but bear with me…With the attack and release settings adjusted correctly (slow-ish attack, faster release) each pick attack jumps up slightly in volume as it sneaks through the compressor’s detection circuit before being clamped down upon, then quickly released in time for the next pick attack. This actually gives the feeling that the guitars are moving forward at the listener and “jumping out of the speakers” - they sound brighter, have more impact, and are more “in your face”. With the right eq, they maintain definition and clarity, despite being often highly distorted.
Without going into all the gory details about the other rhythm tones, suffice to say that I followed a similar method with the other rhythm guitars, of which there were 4 in total. I listened to each tone in turn, eq’d out the undesirable stuff with ReaEQ, then polished them with VMR’s processors. I varied the areas when I boosted and cut, not only to enhance the tone, but to provide complimentary contrasting tones to create greater difference left to right in the stereo field. This technique has the benefit of creating the perception of greater stereo width, which is an excellent additional mix tool to emphasise the impact of new parts.
The other 2 guitars played parts which I would describe as “colouring rhythms” - not necessarily fundamental to the song’s structure, but parts that enhance and create interest and change in the arrangement. I panned these two parts hard left and right, but then sent them to their own buss, where I automated the stereo width of them from 40% to 100% throughout the course of the song. I also automated the volume on this buss greatly, so that they kept low and in the background during the verses, but came up almost equal volume with the main guitars during choruses. This had the effect making the choruses sound wider and more expansive.
Here is all the guitar parts together, solo’d for a verse and chorus to give you the idea:
You might have also noticed some very subtle ambience on the solo’d rhythm guitar track. This is actually the same “room” ambience I described earlier on the drums. However, rather than send the guitars directly to the room verb, what I like to do to ever so slightly intensify the sense of depth and front to back image in the soundstage is to send my guitar reverb sends to a “guitar room” track. This track has just has Reaper’s stock “ReaDelay” plugin on it, set fully wet, no repeats, with a delay time of about 40ms. This then runs into my “room verb”. Why?
What I noticed when setting up reverb plugins is that perhaps the most powerful control on a reverb is pre-delay. Predelay delays the onset of reverb and separates it from the source sound. Psycho-acoustically, it brings the source sound “closer” to the listener, and “further” away from the boundary surface of the imaginary “room” being suggested by the reverb. Since sound travels at roughly 1 foot per millisecond, it’s easy to imagine a room with the band set up, the drummer at the back, closest to the rear wall, with the keys and background singers just forward of the drummer, perhaps the guitars in front of the keys, and our lead singer right out front…
As it has it’s own self-contained low and high pass filters, I also use ReaDelay to high and low pass the respective “rooms” at different frequencies. Since when sound travels it loses high and low end the further distance it has to travel, it stands to reason that the further our imaginary “boundary” is away from our source sound, the less high and low end the reflected sound will contain.
So this is how I set up my “room”. The “room verb” that all the other “rooms” feed into has no predelay. I have a “keys room” with a predelay of 30ms, “guitar room” at 40ms and a “vocal room” of about 65ms. This is, of course, by no means scientific, just an efficient way to create extra depth while keeping everything in the same “space”, without using a ton of different reverbs.
With the rhythm guitars, I kept them all pretty dry, because they already have some room tone "baked into the recording of them. If you listen in headphones and compare the raw guitar with the processed one, you’ll hear some subtle roominess on the raw file, but you’ll also clearly hear the delayed reverb panned to the opposite side of the stereo spectrum on the processed file. I really like this technique. This “crosstalk” creates a subtle connection between the left and right sides of the mix - it joins everything in a cohesive way that retains great separation at the same time.
What about the solo parts? There were a couple of slide guitar parts that Bob recorded as part of the rhythm guitar “colouring” tracks. These were lower in volume than the rhythm parts, so I needed to volume automate them to stand out in the mix. I actually “multed” the slide on it’s own at the beginning and end so that I could pan it and treat it individually, but the eq and compression settings were essentially the same as the rhythm parts. Since the slide’s role turned out to be a foil for the rollicking piano part that came later in the production, I ended up panning it to the opposite side of the spectrum so they could “play off” each other. As the song progresses, I add delay to the slide parts to increase the sense of excitement:
Here are those beginning parts.
And here is the ending slide with added delay for more vibe:
The lead guitars followed a similar tactic, although they have a fair bit of the same additional ambience as the lead vocal (which I will explain later).
The first lead is a little dryer sounding:
While the second has considerably more delay effect to give it more impact and excitement: The ping pong delay features pretty strongly here:
That’s pretty much it for the guitars - next up: Vocals
2 Likes
Part 7 - Lead Vocals
Ok…Well to the 2 or 3 people who might actually be bothered to read this, I hope you get something beneficial out of it. Certainly, writing about a mix in any meaningful way probably takes about 5 times longer than actually mixing!..
So…vocals… As I mentioned at the outset, Bob’s vocals are always great - full of soulful tone and powerfully passionate, not to mention pretty much spot-on pitch-wise. The goal was then to simply get them to sound as good as possible, and to emphasise the attitude and emotion in a way that matched the vibe of the music. Here’s a sample of the dry vocal recording I received:
I had already evened out the dynamics manually (see mix prep), so the next step was to listen for any resonances that didn’t sound so great and reduce their impact. Reaper’s stock ReaEQ took care of this task with a high pass around 112hz and narrow dips centred at 270, 439 and 697hz. I then fed the vocals into Slate’s Virtual Tape Machine. I find this to be one of the most effective de-essers around. Since I started using it on lead vocals, I rarely have to de-ess at the end of the chain.
Next came my 2 go-to vocal compressors in series - first the Waves CLA-76 Blue Stripe (1176 emulation) set to 4:1, attack set in the middle, fastest release, with about 5-6dB of gain reduction. Then into the Waves CLA2A (LA2A emulation) set to “compress” mode with about 3dB of gain reduction. I love this 1-2 combination for vocal compression. The combination of these two processors just pin the vocal in place perfectly. The 1176 knocks down the bigger peaks, then the LA2A brings up all the lower level detail.
Here’s what that sounded like:
You might pick up some prominent room sound at the end of the word “raindrop”. That is actually due to the raw file being “gained up” on the “as” word that follows “raindrop” by about 13dB! Probably unnecessary in hindsight, but it worked out fine in the context of the full mix.
Next, time for some “polishing” eq, a little more compression, and some more eq. I used Virtual Mix Rack for this: First the Neve Preamp emulation for a little more attitude and texture, then into the Neve EQ with a little boost at 108hz for some “girth” and a bit at 4k for some “cut-through”. Following that, the Slate 1176 in parallel mode just bringing in a little more detail. Finally, the Slate SSL EQ with a boost around 5k and pulling out a little from 10k up with a gentle shelf, so as not to get too “glossy”.
I still had a hankering for some more grit and bite from the vocal, so I called up Soundtoys’ “Little Radiator” plugin. This can get quite distorted, but the secret is to use the “mix” control to blend in the nastiness until it’s just “nice”. It tends to add a fair bit of noise, so I followed it with a gate to get rid of that when a signal wasn’t present.
One more thing I like to use to get a vocal to sound “larger than life” is to use a little bit of stereo pitch-shifting the spread the vocal out from the centre of the mix. Soundtoys “Little Micro-Shift” is my go-to for this, used very, very subtly.
So, here is the finished, dry vocal tone:
If you compare it with the previous clip, you can hear that the distortion and extra processing just adds that little bit of excitement and “edge” that pushes the vocal right forward and “in-your-face”.
But the insert processing is only part of the story… The ambient effects are the secret to getting the lead vocal to really “nest” into the mix, as well as creating a sense of movement, size, progression and excitement.
I mentioned my “room verb” I set up for the drums and guitars earlier, and how I use multiple busses with different pre-delay times to feed the room verb to provide a very subtle sense of depth… So the first send effect is to the room verb with 65ms of pre-delay. Here’s that sound:
Next send is my “excitement + glue” mono slap delay - set to 166ms. It’s dark, rolled off on the low and high ends, and a little bit distorted. This one just works - it makes quite a difference!..
Another send effect is the “Stereo 8th Delay”. This is another quite dark delay without much low end set to an 8th note repeat on one side, with an additional 10ms on the other side, spread out wide. It’s quite subtle in this example, but I pushed this up for the choruses and at other times for highlights in the mix:
Then we have the plate verb - this really adds some “size” to the vocal. This is a Bricasti plate emulation with the lows rolled off and a massive 150ms of pre-delay:
Finally, we have 2 more “special FX” delays. The first is an 8th note ping pong delay that is used both for a “delay throw” on certain words, and as an “excitement” delay for the chorus. The second is a quarter note ping pong delay that is much wider, very “am radio”-filtered, and only used as a delay throw on certain words.
Here’s the whole section from the verse through to the chorus with all the send effect processing on the lead vocal:
So, that’s about it for the lead vocal. Next up, backing vocals.
1 Like
Just one?!?!? This is really great stuff!! And +1000 on what Emma said as well.
A heartfelt thanks and much gratitude for taking the time to so painstaking put all of this down.
2 Likes
The treatment is very nice, although it’s interesting to hear how much the delay and reverb become less prominent in the mix. It’s also interesting to hear me hit a couple of those notes flat and have the ping-pong delay repeat my errors multiple times. Nice job with the processing, always love the delay in your mixes.
1 Like
Thank you to @ColdRoomStudio and @StylesBitchley for this thread!! This is fantastic reading for all levels/abilities. You guys rock!!! :beerbang:
1 Like
Come on Andrew, the suspense is killing me. What did you do to make the Bobettes sound so good together in the background vocals? Don’t leave me hanging.
1 Like
Yeah, sorry Bob… I’ve been a bit snowed under lately - I’ll try to finish it soon.
1 Like
No problem. Nobody told Da Vinci to hurry up on the Mona Lisa.
I definitely want to keep this thread in the archives. Great info in here for reference!! Thanks again!
Thanks… Sorry I haven’t had time to finish it… Been too busy…recording, actually! 
1 Like
Part 8 - Backing Vocals
Sorry for the delay in posting a follow-up here… been busy…
@StylesBitchley Bob’s backing vocals are a BIG part of his signature sound. In our pre-mix discussions, we determined we were aiming for a sort of early 70’s Rolling Stones/Black Crowes, loose, soulful/gospel vibe for the backing vocals, and Bob really delivered the goods in the performance department. He sent me 4 additional vocal performances to the lead vocal - A vocal double for the choruses, a high harmony, a low harmony and an additional “extra harmony” that was mainly just in the bridge.
Here are the raw harmonies solo’d:
One well-worn thing I like to do to make the most of harmonies like this is to double them. Notwithstanding the fact that Bob hadn’t supplied any doubles of the harmonies, creating doubles from the existing material is almost trivially easy. It’s just a matter of copying other places where the chorus occurs to a new track. The reason? You can then pan the harmonies out and really get a BIG sound with a more rawkus vibe that comes from all those voices rubbing against each other.
Here are the raw harmonies doubled and panned:
I then used reductive eq on the harmony parts, notching out similar frequencies to the lead vocal, but high passing the higher pitched vocals more aggressively than the lower ones. Here’s how that sounded:
On each background vocal channel, I followed that with some compression from Reaper’s ReaEQ, then more console saturation, more EQ and more compression from Slate’s VMR. I also put an instance of Antares Avox Vocal Doubler on the Vocal double track, so I could spread it left and right.
Other processing included sending each vocal to my “room reverb” to put some natural space around it, and finally putting a de-esser and a multiband compressor on the background vocal group track. The combination of vocals tended to resulting in some “peakiness” and buildup happening around 480hz and 2900hz. Compressing these areas on the combined vocal tracks helped tame this issue.
I also added a delay to the vocals overall. This was automated and pretty prominent in the mix at certain points to highlight phrases
Here is the fully processed vocals as they appear in the final mix:
Once I got a basic mix going of the track, but well prior of the finished mix, I sent Bob a sample to see if I was headed in the right direction, and to get some general feedback. Here is the chorus from that sample:
As you can hear, the blend is very different with the lead vocal much more prominent in the chorus mix. The feedback I got from Bob was that he wanted much more prominent harmonies “…like 4 Bob’s standing around a mic, bellowing into it”, or words to that effect. My initial mix also had the BVs way out wide, and Bob wanted them more central. He also decided to add another high harmony that came in just in certain spots to really flavour things nicely.
…So things ended up quite different from my original first blend… and for the better too, I think. One thing I did with the backgrounds was quite a bit of automation as the mix progressed. The vocals started out quite narrow, but subtly widened with each successive chorus. I did this using the excellent “Stereo Pan” function (standard as an option on every channel in Reaper). I also automated the overall volume so that the choruses got more rowdy as the song progressed. Here are the choruses of the final mix played back to back to illustrate the progression:
Next up: Keys and finishing touches.
2 Likes
Thanks so much for this write-up! Really good stuff in here. I do have a question, however.
You wrote: [quote=“ColdRoomStudio, post:36, topic:1184”]
Other processing included sending each vocal to my “room reverb” to put some natural space around it, and finally putting a de-esser and a multiband compressor on the background vocal group track.
[/quote]
I know you probably add the de-essing to the group for simplicity, but, in general, wouldn’t you want to de-ess BEFORE and reverb/delay and other processing so you don’t replicate the sibilant sound throughout the FX?
Also, I’ve wondered if there is a “rule of thumb” for dB reduction and Q size for the complimentary EQing? How much and how wide would you typically notch out of the BG vocals for the lead VOX? Is it the same relative amount for guitar parts too? (i.e. rhythm vs. solo)
1 Like
I would also point out that Andrew probably could have done an entire page on the automation for the backgrounds. I am by no means a trained singer, so mic technique becomes a big part of the story, but it doesn’t come close to getting the needed blend. This is mostly due to the fact that I have to sing louder to hit some of the higher harmonies, and really back down on the lower ones to keep them in pitch. So when Andrew says he did a little automation, he really means he was about to quit the business and become a Tibetan monk after mixing this stuff.
2 Likes
I don’t completely have the same problem, but when I gotta go real high, I have to go very loud as well. I can imagine the pain of automating 
I guess letting a vocal rider do halve of the work isn’t a bad idea
1 Like
V[quote=“miked, post:37, topic:1184, full:true”]
Thanks so much for this write-up! Really good stuff in here. I do have a question, however.
[/quote]
Thanks! Hope it was helpful.
I know you probably add the de-essing to the group for simplicity, but, in general, wouldn’t you want to de-ess BEFORE and reverb/delay and other processing so you don’t replicate the sibilant sound throughout the FX?
Great question!.. And you are correct: De-essing before the reverb/delay is also my preferred way of doing things, and I am actually doing it that way in this case too… I probably didn’t explain it as well as I could have…
All the time-based processing (reverb and delay) are on auxiliary sends, so effectively, they are processed in parallel, not in series on the actual tracks. In the case of the “room” reverb, each vocal track has a send to it (those sends are also panned). When I said the vocals have delay as a group, that is accomplished by sending to the delay from the “background vocal” group track. Does that make sense?
Also, I’ve wondered if there is a “rule of thumb” for dB reduction and Q size for the complimentary EQing? How much and how wide would you typically notch out of the BG vocals for the lead VOX? Is it the same relative amount for guitar parts too? (i.e. rhythm vs. solo)
Generally, complimentary eq requires a wider Q with a fairly conservative gain reduction. The core tone of lead vocal might be centred at around 2k, so the reductive eq on the BVs might be centred at 2k, but the curve might start reducing around 3k at the top end and down to 1k at the bottom… So usually fairly wide, and maybe 1-3dB at most, depending on requirements, but it really depends on if it gets the right result or not.
As frustrating as this answer sounds, I’m very wary of recommending “rule of thumbs”, because in mixing, there simply are none. Everything is context-dependent. Just start with the basic principles and find a way of working that brings a result that sounds good to you.
Ah, thanks for chiming in Bob! You just flagged something I intended to say in the last instalment, but forgot.
Actually, surprisingly, I didn’t automate each background vocal track in the same detailed way I did the lead vocal. In fact, I don’t thing I did any automation on the individual tracks. I certainly automated them as a group, though, but that was simply to increase their presence in the mix as a whole, rather than to control their dynamics.
The dynamics of the individual vocal tracks were controlled adequately with 2 instances of compression on the individual tracks. Normally, in a mix where I wanted the backing vocals to be really slick and smooth, I would put and additional compressor on the group vocal track, but I deliberately avoided that tactic here. That results in a very homogenionous, almost “pad-like” sound that works well for certain styles of music, but it was the wrong choice here IMO.
I mentioned before, the “loose gospel/soul” backing vocal vibe, and what I’ve noticed with that style of presentation is that things DO “poke out”. The levels of the individual singers are uneven… But that’s what MAKES it sound so great and passionate.
A good example of that is The Rolling Stones’ “Gimme Shelter”. The female backing vocalist is absolutely shredding at the top of her range, and she sounds way louder than Mick in the choruses, but it’s absolutely perfect for the shambolic, apocalyptic vibe of the track.
So I deliberately erred on the side of less dynamic control, or “just enough” to maintain the vibe.
1 Like