FOR MUSICIANS ONLY Audio Mixing Perspectives -- An Opinion by George Ziemann -- May 7, 2010 Since some readers have found my digital recording/mixing techniques from Hurricane Alley's Category One to be interesting, I thought I'd discuss the evolution of my personal perspective as I've been working on Category Two. Signal Path With live sound, the whole signal path thing can get pretty confusing because whatever it is you're going to use (delays, reverb, other effects and processors) to be hooked up, plugged in, routed somewhere and ready to go before you start and you need to hit all the important spots in real time. When we did Hayden's Wall, I forgot to resist the urge to apply the effects as I heard "Road to Nowhere" being sung. It came out all right, but it locked me into something because I was still thinking live during the recording process. But I was still a ProTools rookie and hadn't really taken advantage of submixes, even though I used them all the time for live sound. Duh! By the time Category One came around, I had added submixes to my templates for starting a new session with the band. Even a single voice sent through a stereo submix will give it more presence and fatten it up a little. So just by sending things through an appropriate submix, the mix takes on a different characteristic. Gain Structure and Compression When we're recording, I just want to make sure everything being recorded has a good, strong signal but stays out of the red, avoiding distortion from clipping. Here's where you have to remember that it is NOT live, despite what everyone will want to hear in their headphones when you're doing it. This is the only purpose of the recording part of the work, in my opinion -- to catch all the sounds being created, as accurately as possible. The exact levels being recorded do not have to be in the proper relation to the other sounds, which is why the faders in ProTools only effect the monitoring level, not the record level. As long as it's close, we can do anything with it. What we do next is where I separate from the rest of the crowd. Everyone uses compression of some sort to level out tracks, vocals especially. How I go about compression is a little unusual, but produces excellent results and there's a solid reason why this is so.

FOR MUSICIANS ONLY... "For Musicians Only" is a category of articles written for the musicians out there in Readerville. Hurricane Alley Category One Prepare Yourself (download) Nothing to Hide (download) More free mp3s and links to buy a copy.

At right is part of a vocal track, as recorded. The vast majority of the track is below -6 dB (the red line) Point A is the loudest spot in the track. Point B is about 2 dB or so above the red line and there are maybe five or six of these minor peaks throughout the track.
After running the track through a compressor, not only have our peaks been reduced, but every other vocal phrase as well.
Returning to the original version, then I took the track through a limiter, which gives about the same result.
My approach is to go through and manually reduce those few peaks so that they fall at or just below the -6 dB line. This lets me bring up the entire track by 6 dB without compressing, limiting or otherwise processing anything that didn't need it. Once we bring the track up, this might present another set of obvious peaks that can be brought down to give the track another 2 or 3 dB of gain. Doing this process, which I call manual limiting, takes much longer to do than running the track through a processor (especially on a kick drum track, for instance), but it preserves more of the original track while still leveling it out. This is not to say that I don't use compressors or limiters. But even when I do, it's not unusual to find a few spikes that happened too fast for the processor to react to. Correcting this is how I started doing manual limiting in the first place. Pre-Mixdown As I go through every track, after I have done whatever it is I'm going to do to it, be it EQ, compression, or simple editing, I run the gain of the track up to about -0.3 dB. This is because it makes much more sense to have strong signals than to try to amplify weak ones. With a hotter signal, you can bring down a track's level in the mix and it is more likely to retain presence, whereas the weak signals have a tendency to get washed out. When all of the individual tracks are ready, it's time to do a little housecleaning. I consolidate each track so that it is a complete, separate file. Next I go to the Edit Window, have it select all the unused files and pieces and parts, then delete them. Make sure every track has a meaningful name, as opposed to Audio 1, Audio 2, Audio 3... Since I had a couple of incremental backups of the session and files before I did the real time-consuming work on all of the tracks (and had gone back to an original version of a track more than once), now is the time when I'd do another one. Mixdown Okay, we've done all of the grunt work. We've got hot, well-compressed tracks that are as clean as possible, except for distortion that was an intentional part of a guitar or keyboard sound. Drums and Bass The combination of these two is going to be the foundation of the song, especially if it's rock. So I'm going to start the mix by turning everything else off and getting these two together. There are a million ways to mix and EQ the drums, which depends on what you're aiming for. My initial goal is to make them sound like the drum set I recorded. I used to have a dysfunction when I did a drum mix for a record, because I would mix from the drummer's perspective. Never had that problem live, but it was a different visual connection. In the first band I did sound for (in the 70's), Kenny, our drummer would come down after almost every set and ask me, "How does the kick sound? Pillowy or punchy? Is the hi-hat crisp and sizzling? Does the snare snap or just pop?" and things like that. He taught me how to properly listen to a set of drums, and when I'm mixing or recording, I still run through Kenny's list of questions. It keeps the drums from ever seeming like a daunting task, even live, working for people whose names you would actually recognize. Bass is pretty straightforward, panned center with a level that is dependent on drums. Or, you could send it to a stereo submix, which would fatten it and give it more spread, if you need to. Submixes These are just stereo Aux In channels. I always use one for vocals, vocal FX, and drums. I might also add a background vocal submix, and another for its effect. Depending on the song, maybe a guitar or keys submix is a good idea, too. So, for example, the lead vocals get routed to Bus 5-6 instead of the Main Out 1-2. Bus 5-6 is the source for the vocal submix channel and the vocal effects channel. Background Vocals go to Bus 7-8, and so on. They are in separate submixes so that you don't compromise the presence of the vocal to put an effect on it. In fact, even throwing a single vocal into a stereo submix does good things to it. Since I've compressed or limited every track, I really prefer not to add any more of it to either the drum mix or the master output. Sometimes it's unavoidable. Sometimes, it's just a matter or going back to Step One, making sure the bass/drum combination is tight, and try not to bring everything else up far enough to start drowning out things again. Sometimes just a little extra EQ boost at the appropriate frequency will do the trick, too. I find that if the kick begins to be a little weak, and I'm already driving its primary resonant frequency to the max (about 360 Hz in this case), if I add a narrow notch at twice that frequency (720 Hz), it'll only take a couple dB of that to make it stand out again. But it IS a rock song I'm most likely working on, or the drums wouldn't be fighting for presence. Usually bringing down the rest of the rhythm instruments just a tad will take care of this. Sometimes a little compression is the only way to go. But just a little. Try to save some dynamics. Automation This is really the final step, although I'll still find editing tasks when I'm in the process, which sometimes includes adding entirely new tracks -- maybe a harmony, maybe a decision that a Hammond would sound better than the piano. My method is probably bizarre, but it keeps me from getting burnt out on one song. We've got 11 songs for Category Two, all of which are in this final mixdown stage. So I pick a song, record the basic automation, then start listening to it. I'll give it maybe four or five plays, adjusting the automation, fixing little details here and there, tweaking EQs, panning, whatever, then move on to the next song. I don't think there's any special way to do this part. Every song is different. How you hear, what you want to hear, your ability to accomplish that, and your own personal tastes are what will really shape the mix. A reggae act is going to want something completely different than a country act or a punk group. Some people like their vocals upfront; others like to be deeper into the sound of the band. I also spend time listening to the depth of the mix. Is it a flat plane, left to right? Or do things come forward and drop back? I prefer the latter. When I can listen to a song four times in a row and not hear anything that needs to be changed, then it's probably finished (because I'm pretty relentless about it), or at least ready to be judged by my teenager. If it passes her scrutiny, then I let Carl hear it, and he's a little tougher to please, as he should be. I wouldn't want anything less. Attitude The last thing I want to do is to call a song "good enough" just because I'm tired of listening to it. The other thing I really don't want to do is to slack off because I don't expect to sell a million copies. In the past, I've heard, "Well, it's only a demo..." and "It's only to give to club owners..." as reasons something is "good enough." I think that if the goal is only "good enough," it's not even worth doing at all. It takes far too many hours (not to mention the experience and gear required) to make an album to waste time on anything if you're not going to at least shoot for great, whether you actually hit it or not.