It has been known, for almost forever, that listening to music at louder levels just sounds better. That is largely because our hearing sensitivity to bass, and to a lesser extent in the treble region, falls away as the loudness level fades.
But knowing that, and actually becoming conscious of it, are two different things. Especially for mixing and mastering tasks, it becomes important to heed this nature.
We don’t want to mix and master for long periods at the target loudness levels. For example, the THX specification has an average loudness level of 83 dBSPL for movie theaters. But that is very loud, probably too loud, for monitoring in a small studio.
Bob Katz has suggested a more comfortable level about 6 dB lower at 77 dBSPL. But even that, for long sessions may be too much for some people. I’m finding for myself that a more comfortable level, for headphone monitoring, is another 6 dB lower to around 71 dBSPL.
But then, if you start adjusting your mixes while monitoring at that lower level, you will find that you end up adding too much bass. So what to do?
I happened to catch a YouTube video the other day, from British Engineer, Dan Worrall, which alerted me to an improved method for mixing and mastering at lower sound levels. He uses an instance of TDR SlickEQ in his monitoring chain, because that EQ has a special metafilter to compensate for diminished bass sensitivity at lower monitoring levels.
[BTW… All of Dan’s videos are worth watching. He is a very knowledgeable and clever mixing engineer, and I applaud his willingness to educate us.]
It turns out, after doing some more research, that you can easily do your own compensation with a simple low-shelf EQ filter. This paper by Fierro, Ramo, and Valimaki provides the background. While the paper recommends using a 1-pole low-shelf compensation filter, I found a decent match using a more common 2nd order low-shelf filter, with best settings corner frequency of 147 Hz, Q = 0.4, and bass boost given by half the amount of the drop in monitoring level from 83 dBSPL. This matches the theoretical compensation to within less than 0.5 dB.
So for instance, if you want to monitor and mix at 71 dBSPL, that is 12 dB lower than the nominal playback level in the THX Standard. So you would apply the low-shelf boost with a 6 dB gain. Simple!
Of course, you also need to lower your monitoring level to that which gives you 71 dBSPL for average 0 dBVU levels. But be sure to do this only in your monitoring channel, not the channel that will be recorded during bounces or final mix.
Once I started doing this for myself, I was amazed at the improved listening quality of all music at this lower monitoring level, and my mixes don’t need to elicit complaints of too much bass.
While doing my research on bass compensation I decided to take the dive and do a really ISO-226:2003 compliant model inside Crescendo. The previous article showed that to overcome sensioneural hearing loss we need to supply adaptive gain to equalize the Sones to what normal hearing senses.
But Sones are derived from Phons, while our measurement devices all show us dBFS or dBSPL. The two are not equivalent. Sones and Phons are perceptual values and cannot be directly measured on commonly available engineering meters. All we can directly monitor are dBFS and dBSPL.
So by fully embracing ISO-226 for its modeling of the equal loudness relation between dBSPL and Phons, I now have Crescendo operating in a more true Sones equalizing manner.
It only makes about 10% improvement in listening quality, over the previous implementation. But the engineering tests also show an improved compression response, and lowered levels of harmonic and intermodulation distortion. These distortion products were already low, and probably indiscernible, but now they are even lower.
Now this is interesting… Audiology exams report back your apparent threshold elevation in units of dBHL versus frequency. But Sones compensation needs to be based on Phons, not dBHL.
My own professionally administered audiology exam reports that I have a threshold elevation of 60 dBHL at 4 kHz. But while listening through Crescendo, I gradually determined that my best vTuning setting was 54 dB. vTuning represents your threshold elevation at 4 kHz.
When you perform the conversion in ISO-226 from 60 dBHL to Phon at 4 kHz, it turns out to be 54 Phon! (Phon are always measured in dB)
So, in the future, we need to be mindful to apply the dBHL to Phon conversion to stated audiology measurements, for determining a proper vTuning setting in Crescendo.