Implications for Recording

The fundamental insight has to be this:

Unless there is no change in sound pressure level, there is no such thing as flat frequency response for recording/playback!

This is bad, bad news. When you change level, you change frequency response. No way out of it.

If you are making acoustic recordings, the control room monitor level has to be the same as the recorded level for there to be any reasonable semblance of spectral accuracy. If you are making multitrack recordings, particularly with samplers ‘n synths, you are absolutely out of luck, because those various original source levels were all different, and as soon as you mix them and listen to them together, some of the levels have to be spectrally inaccurate just so that others can be accurate.

Interestingly, when I installed my new flat-response wide-dispersion prototype monitors about a year ago, I really noticed this effect. Particularly in mono, recordings sounded spectrally deficient until I get them to levels that pretty much matched normal acoustic levels for such sounds. The speakers were flat enough that absolute level became a significant feature of the accuracy of the musical illusion.

The fact that less accurate speakers tend to reduce this phenomenon isn’t a saving grace – now you’re flailing about with two levels of confusion, absolute level and bogus speaker response. This might help explain to you why making good recordings is so damned hard!

Implications for Mixing

When you change level, you change equalization signficantly enough to overshadow many other aspects of sound quality.

When you mix, you work at some given level or levels. The level at which you approve your mixes and say, “Ahhh, that’s really great! Folks, we have a hit on our hands!” is the only level at which your mix can be heard accurately.

So, you’d better pick that level carefully! Think about it a lot. Get really grooved in to it.

Implications for Playback

This is where the audiophiles have to take a serious hit. Unless they/you/we know (a) the original source level and/or (b) the mix level, there’s no way to play back what the producer intended. And that data isn’t included in the liner notes. In the film industry, there is some standardization (85 dB SPL is supposed to be equal to the nominal level for one speaker), but in fact film playback isn’t all that well standardized and, further, even if it were, the difference in levels throughout any given theater would make the standardization pretty much useless except for a small zone of seats.

Things you can do to survive

The most important thing, first off, is to get a Sound Pressure Level meter and determine the level(s) at which you work. Once you get that knowledge in place, the next thing to do is to stabilize your working level. For example, I use 85 dB SPL (C weighting) at the mix position with pink noise on one speaker at –14 dBFS (RMS, not peak indicating).

Groove in to this level. Learn all of your favorite recordings at this level, over your reference monitor system. Become as familiar and stabilized with this reference as you possibly can. Do all of your “serious” listening at this level. Stabilize. Measure. Practice. Listen. Repeat. Repeat again.

When you are mixing, make all of your decisions and changes while working at your reference level. Do not make changes while listening at any other level.

However, after you get this reference level under your skin, you must start checking how things sound at other levels. I particularly like -20 dB (65 dB SPL) because it sounds lousy and also represents a soft “easy listening” kind of level. Don’t “work” at this -20 level, just listen. And listen a lot. Get very used to how recordings sound different at different levels.

Once you get these levels in your ears, then you can start taking advantage of the “virtual loudness” inherent in the equal loudness contours. Instead of mixing at 100 dB SPL because “it sounds better,” try applying the ”100vs.90” response EQ curve to your 90 dB SPL mix. Now, without significantly increasing the overall level, you can add the sonic signature of the 10 dB louder level, with all of its relative desirable impact. Not bad. Nice trick, if done with care. Warning: it can be overdone!

Things you should keep in mind as you work

We’re in a tricky business. Among other things, there is no truth, let alone any single right answer. Further, the whole thing about Really Accurate Recordings is more than a little oversold, and the Equal Loudness Contours and problems they represent are just one glaring example of why that is true.

Your real goal is to make recordings that have as much impact and musicality as possible. A big part of your craft is to manage levels. Another part is to carefully listen to your recordings, under a wide range of conditions, including other levels. Now that you’ve taken a look at these Hearing Response curves and response difference curves, plug in that information to go along with everything else you know, to help predict and explain why your recordings sound various ways at various levels.

Then, target your mixes to work at the levels you expect your end-users to listen at. Think of it this way:

Your job isn’t done until you’ve made your mix work at all of the relevant levels on all of the available speakers.

It ain’t easy. Be prepared to take some time at it. May the Force be with you!

Happy Phons.