Tuesday, February 26, 2013

Apparent Loudness

This is a tricky subject. Psycho-acoustics is a deep pond to swim in and I may not have enough air in my tanks to get to the bottom of it, but I'm going to give it a try. The particular area I want to go over today deals with how loud something seems to be. 

You can play an uncompressed sample and then compress it and most people will identify the compressed version as the better sounding one. There can be a lot of reasons for that. A little light compression (with appropriate make up gain) will bring quieter sounds up in a mix and things will sound fuller. The average level increases as well which will make the compressed track sound louder too, even though the peaks aren't going as high as the original. 

So if a little compression is good then more would be better, right? Welcome to the late 1990s and the Loudness Wars. Technology has progressed to where you can just about make a song where the level doesn't stray more than one decibel from a full scale reading for the duration. It even spawned a new term, "hyper-compression". It's really just a continuation of the contest that record companies have been a part of since the 1950s. If your record is a little bit louder than the other records on the radio it will get more notice. But there are better ways.

But there's more than one way to skin a cat. The problem when you go too far with compression is that everything goes flat. Nothing sounds natural. A whispery ballad is the same volume as a brutal dub step drop when they're both coming out of your clock radio in the morning. That's just not natural. What makes things exciting to listen to is dynamic range. The trick is to balance the amount of compression needed to make the mix full with the amount of dynamics that you leave in to keep it interesting.

Take for example a kick drum. Untreated it sounds blah and boring. Start applying compression and it will fatten up. But go too far and it will just get flabby and loose all of its punch. One way to balance is to relax the attack time on the compressor so that the first transient of the beater hitting the drum gets through as a big spike, then the compressor engages and squeezes the resonance out. Or you could split the track and keep one side uncompressed and heavily compress the other one. By blending them together you can keep the interest of the attack and blend in the resonance of the drum.

Another way to create loudness is to compare it to something soft. How many songs have a dramatic pause before things really kick in? And how much more incredible is that break after a measure of silence? Everyone from Led Zeppelin to Skrillex knows it. If you want to make a building look taller, dig a hole next to it.

But tricks aside, with production getting ever denser it's really tough to keep dynamics in a mix. Whether it's a metal track with multiple guitar layers or EDM with a bucket full of crazy synth sounds, eventually you have to get it to come together as a mix and not turn to mush. So let's take it one method at a time.

Stereo width can play a big part. People are nervous about their mixes summing correctly to mono. They should be. But there are plenty of ways to push the pads and subtle rhythms out to the sides to leave room for stuff in the middle that will still work out if a DJ plays it back on a mono system or it gets piped into a department store.

Next is careful management of frequencies. Chances are if you go into a project and tune up each track without listening to the whole thing together it's going to come out awful. Adjustments need to be made while taking into account how they effect the whole mix. Kick drum is a great example. Most people think about 60 to 80 Hz when they talk about kick drum, and probably something up around 4 kHz for that beater sound. But what about 200? Is there room for some 400 in the mix? People spend all kinds of time clearing out the low mids when they're working on guitars and vocals. With good reason, they muddy things up. But if you're not putting anything back in there you're just wasting the space you created. Why not find the resonant frequency of your kick drum, go up a few octaves and enhance some frequencies that will actually play back on laptop speakers?

And lastly, we come back to where we started, dynamic processing. If you're not thinking of your compressors as EQs then you really should try to get your head around it. A comp will make a vocalist sound good if it brings out some nice mids in their voice, but it can also make them sound shrill if the wrong frequencies are emphasized. That's why using side chaining or multi-band compression is a great idea. Make a compressor work on exactly what you want to. It doesn't have to just be a traffic cop watching out for speeders. 

Throw a multi-band compressor on your two mix and work it over. Grab some mid frequencies, shove the ratio right up and drag the threshold way down. Now do it with the lows and the highs. Get so you have a feel how each separate area of your mix responds to it. Even better if you can mute some of the bands while you work on another. Now you can listen to specific frequency segments and figure out what they need. 
Is there room in the lows? Or is that area full up? Is the high hat the only thing hanging out above 4k? Maybe you could incorporate some click from the kick drum and make it even louder without giving up much headroom. How do your vocals interact with other things in the mid range? What happens when the vocals stop? There's a whole world in there to explore, section by section.

Well Brethren of the Knob and Fader, I feel like I've been swimming for hours and I'm still nowhere near the bottom of the pond. This is a deep, deep subject and there's a lot more to get in to. Stay tuned till next time and hit us up with your questions and comments.

Thanks to 4Flexx for prompting me to write this. It was good to think through this stuff and get a lot of different ideas together on the same page.

No comments:

Post a Comment

You're the Scotty to our Kirk