"Inside-Out" Mixing

[Anderton]

I wrote about this in my Friday Tip for Studio One, but it  applies to any DAW. The basic question is...where, not how, do you start your mix? I'm finding it works best for me if I start the mix in the middle, in the busiest part of the song, and work outward from there to the beginning and end. Of course, there are more details in the linked blog post.

[AROIOS]

41 minutes ago, Anderton said:

I wrote about this in my Friday Tip for Studio One, but it  applies to any DAW. The basic question is...where, not how, do you start your mix? I'm finding it works best for me if I start the mix in the middle, in the busiest part of the song, and work outward from there to the beginning and end. Of course, there are more details in the linked blog post.

Solid content as usual. The guitar bend reversal trick is super cool! I'm gonna try it on my next synth solo track. Thanks for sharing these insights and tips, Craig.

[GovernorSilver]

I've been working through a course for Elektron Syntakt and recently completed a lesson in which the instructor said something similar.   He does not pan the tracks to the sides immediately.  He works out other things first, then starts panning tracks to the sides later.

[Philbo King]

I like it!  I'll have to give it a try on my next song.  

 

I already start the mix in mono, making it easier to use EQ to 'carve space' spectrally so things fit together.

 

The inside-out paradigm (sorry to use corporate-speak, a habit I picked up from my preretirement day job...) anyway, it seems like it has a lot of potential to get mixes done more easily.

[KenElevenShadows]

That sounds very logical and smart as an approach.

 

So I often simply mix, addressing all the stuff that pops up that either bugs me or draws my attention in some manner. When I can listen to a mix without reaching over and grabbing things, I know I'm approaching the finish line.

 

Do I recommend this? Not really.

 

Is it largely fueled by ADHD? Yessir.

 

Does it work for me? Sure.

 

Print this out and paste it right by your DAW. It's free, after all.   

[Anderton]

 

7 hours ago, KenElevenShadows said:

When I can listen to a mix without reaching over and grabbing things, I know I'm approaching the finish line.

 

Same here. I highly recommend that approach to people who say they can't finish mixes

 

The inside-out part is more about what happens after arranging. It's mostly about working fast. Before I got into that kind of mixing, if I made tweaks while tracking or arranging, I often had to redo them when the arrangement was finalized. So, I found that finalizing the arrangement first meant that I had to make only one set of tweaks. 

 

I'll admit it's tempting to edit sounds while tracking or building an arrangement, but I always regret doing so

 

[KenElevenShadows]

I think I mostly edit sounds during the tracking process if there's something that needs fixing, whatever that might be, but otherwise generally leave them alone. I might break from that if I really know that I want something a certain way, though. 

 

During mixing, I often have things that really pull at my attention, and I usually just go with it. So in that regard, I probably mix differently than a lot of people because I don't generally approach things exactly the same way. I might one time really work on getting a groove going before anything else. Another time, I might be really focused on the lead vocal and then sculpt everything around it. So I don't know. I don't know that it's the best way to go about things, but it's what I typically do.

[Anderton]

12 minutes ago, KenElevenShadows said:

I think I mostly edit sounds during the tracking process if there's something that needs fixing

 

You have more patience than I do!

[KenElevenShadows]

2 hours ago, Anderton said:

 

You have more patience than I do!

 

It's not so much patience, but fear of forgetting to fix it later, and having something bug the crap out of me.  This is why I don't recommend that most people mix like me, haha!

[Sundown]

Hi Craig,

 

Good article ...

 

I never thought about it explicitly, but I sort of do the same thing by starting with the section of the song that is the climax or peak. If that's the "big bang", everything other segment needs to take a backseat (from a psychoacoustic perspective and from a basic mathematic one). I too minimize ambience effects at the start of the process and I'll first focus on EQ and balance, and I often do my big EQ cuts in mono. If I can make some space and get the overall frequencies in-line with mono, I'll generally have plenty of sonic space when I work in stereo and start panning with ambience.

 

I would love to say that I arrange and then mix, but invariably I'll find myself with a troubled passage and then I realize it requires an arrangement fix or a different sound. 

 

I know we traded thoughts on buss compression a few months back, but I still use it and I mix into it since it affects the bottom end. If I add an SSL-G compressor or a Vari Mu, it's going to have a sonic imprint that I need to plan ahead for with channel EQ. I only use a touch of buss compression (maybe a half dB) but it helps me glue things together. If I add it after the fact it throws the bass off.

 

As a synth player, I'm also a huge fan of tape emulation as it really helps tame rough frequencies and peaky transients. I'll apply some judicious UAD Studer A800 to mid and high synth parts, typically with the 456 emulation and 15 ips. I don't use my Ampex ATR-102 mastering plugin much as I don't like the head bump on bass and low frequencies, but the Studer is gold for keyboards, particularly for mid and high parts.

 

I almost never EQ in isolation and I don't do the typical 'high-Q--high-boost--sweep' method of finding trouble frequencies. I might do it a bit if a sound is really problematic, but I find that if you boost any frequency with a narrow Q it's going to sound like crap and then you'll wind up cutting everything by mistake.

 

I prioritize mixing with speakers and I put a lot of emphasis on getting the bottom end right. It doesn't have to be booming or bombastic (my music is anything but), but it does have to be right. It has to fit the music, it has to be sufficiently tight, and it needs to have that warm presence that fills out a song without dominating it. If anything gives away an amateur or subpar mix, it's a bottom end that's out of control and boomy.

 

I start my automation early in the process since channel-level variances can solve a lot of issues that might otherwise call for EQ. If a part is going to fade into the background with level, I don't need to worry so much about EQ'ing any clashes. 

 

I check a lot of different environments before I call something done and for me the car is the ultimate test. If the bottom end is right and details can be heard with road and wind noise (including reverb and delay), I generally know that I'm done. On the opposite end of the spectrum, I'll check an iPhone speaker to see if it gets overwhelmed. 

 

It takes a lot of effort to get a good-sounding track, and at the same time I feel like AI and advanced computing is going to address that challenge in ways we can't imagine. But I still hope that one day my kids will look back on the tracks I completed and published "the old fashioned way" and be proud of their Old Man and his persistence.

 

Todd

 

[Anderton]

2 hours ago, Sundown said:

I almost never EQ in isolation and I don't do the typical 'high-Q--high-boost--sweep' method of finding trouble frequencies. I might do it a bit if a sound is really problematic, but I find that if you boost any frequency with a narrow Q it's going to sound like crap and then you'll wind up cutting everything by mistake.

 

Despite what the internet would lead you to believe 🤣, that method is not for people looking for trouble frequencies, which is something I see a lot ("get a smoother sound by reducing all these frequencies"). It's for people who hear a problematic frequency and want to find exactly where it is. I first became aware of this with acoustic recordings that would pick up a particular room resonance. Sometimes any resonances were extremely narrow, and hard to identify purely by ear - especially on individual tracks. It's only when you mixed a bunch of tracks together that the resonances added up enough to be more obvious. I'd find the frequency by slapping EQ on the master bus. Because the resonance from multiple tracks added up, you could hear where it was more clearly. Then I could go back and reduce it on individual tracks. If there were only acoustic tracks, then the reduction could have happened on the master bus but if there was a mix of direct and acoustic tracks, you couldn't do that.

 

Ditto amp sims. Many times there's a common resonance in different amps/cabs (I have no idea why) that's not obvious on individual tracks, but become obvious when summing together multiple tracks with amp sims. By making it jump out, then I can fix it. I think the amp sim sound is always more "analog" when you find and remove that rogue resonance. 

 

2 hours ago, Sundown said:

I know we traded thoughts on buss compression a few months back, but I still use it and I mix into it since it affects the bottom end. If I add an SSL-G compressor or a Vari Mu, it's going to have a sonic imprint that I need to plan ahead for with channel EQ. I only use a touch of buss compression (maybe a half dB) but it helps me glue things together. If I add it after the fact it throws the bass off.

 

With half a dB, I don't think you're going to get into trouble. And, planning ahead with channel EQ prevents unexpected surprises if you don't use bus compression, and then add it. But bus dynamics processing drives me nuts, because I tend to use big gestures when mixing. So, anything I do to the mix changes the level going into the processor, which changes how the dynamics reacts. That's why I wait until the mix is done. However...

 

It occurred to me that the multiband linear-phase dynamics processing I use during mastering could be just as easily used in the master bus as something you could "mix into." Having the ability to isolate the bass range you want to compress may give the kind of effect you want, but in a predictable way that doesn't affect other frequency ranges. It might be worth a try. Or, use a conventional multiband dynamics, not linear-phase, if the result of phase shifts in different bands is what gives the sound you like (you wouldn't be the first). 

 

2 hours ago, Sundown said:

It takes a lot of effort to get a good-sounding track, and at the same time I feel like AI and advanced computing is going to address that challenge in ways we can't imagine.

 

AI can get good-sounding tracks now, but it's much harder to get tracks that all sound good together because that involves personal preference. That's why you're seeing AI mastering and not much in terms of AI mixing.

 

At the moment, AI's "personal preference" is what has come before, and been judged by the majority of listeners as "good," usually based on sales. I don't think it knows how to think out of the box yet. And no AI ever said "With respect, I just removed four bars of pure self-indulgence from that guitar solo you played starting at measure 36. You're welcome." 

 

Fun stuff, eh?

 

 

 

[Sundown]

On 7/24/2024 at 12:00 AM, Anderton said:

 

Despite what the internet would lead you to believe 🤣, that method is not for people looking for trouble frequencies, which is something I see a lot ("get a smoother sound by reducing all these frequencies"). It's for people who hear a problematic frequency and want to find exactly where it is.

 

Great point, Craig. I hadn’t considered that. My preferred approach when I run into something really problematic is to slap Pro-Q3 on the end of the channel with ten or fifteen neutral bands (and with narrow Qs) and then use the band solo feature to hear what that frequency range is doing. I find band soloing to be one of the best features of digital equalizers for learning.