Yesterday’s posting with samples of Crescendo processing got me looking closely at those extreme correction levels. (I have never known anyone to use vTuning levels above 65 dB) So I sat down and wrote a model for hearing exposure.
The model assumes a musical spectrum essentially flat below 1kHz, and falling off at -6 dB/octave above 1kHz. The model also showed that using steeper roll-offs has relatively little effect on the outcome. Even nature’s roll-off of 3 dB/octave produces about the same outcome. Bass heavy music makes essentially no difference because our hearing is so insensitive at those low frequencies.
The musical spectrum is divided into adjacent Bark bands, with the power of each band adding up to the total exposure, taking into account the varying sensitivity of our hearing at each frequency. The model is an overestimate to be sure, since music is not constant. But the model offers some safety guidelines.
This diagram shows the vTuning level used across the horizontal axis, and the effective exposure level after using Crescendo along the vertical axis.
Safety guidelines state that the maximum safe exposure duration declines at the rate of halving the duration for every 3dB above 85 dBA, with 8 hours considered safe at 85 dBA. I have marked these levels out, along with an indication of their safe exposure duration.
These figures are for continuous exposure to broadband noise. Most music is not continuously loud across all frequencies, nor at any particular frequency. So you could scale these durations by anywhere from 3 to 10 times longer.
So that shows that one can comfortably listen with vTuning = 60dB, like me, for more than 8 hours, and for all presentation levels below 77 dB.
Even for people needing 70 dB of vTuning the levels are probably safe for an entire day’s listening. And for people needing 80 dB of processing, very likely safe for a few hours, given that music is not sustained loud over the entire duration of the listening session.
We enforce ear safety in the Crescendo algorithm by explicitly refusing to allow more than 80 dB of vTuning in any of the frequency channels. Had we not done so, then the effect would be as shown in the faint linear extensions of the curves, zooming up to unacceptable exposure levels.
But the internal limiting in Crescendo causes another effect for persons needing extreme levels of correction. You see how the curves begin to bend over for vTuning above about 60 dB, right where the curves depart from those faint linear extensions? The departure causes a gradual loss of the highest frequency restoration. For safety reasons, we simply limit the maximum applied correction to no greater than 80 dB of vTuning in every frequency channel. And this won’t allow sufficient gain to be applied at frequencies where you actually need more than this limit of correction.
The net effect is that you remain ear-safe but with a consequent loss of the highest frequencies. It is doubtful, however, that you could even hear these highest frequencies above 4 kHz if your hearing loss is this great as indicated by your vTuning level at 4 kHz. So the compromise is likely of no consequence for you.
But at extreme levels of hearing correction, the model also shows that there is essentially no benefit to using a lowpass filter to cut away everything above 4 kHz. In other words, most of the sound power that is at damaging levels resides already below 4 kHz.
The chart also shows that, for vTuning levels above 60 dB, attempting to save your hearing by lowering the incoming volume level, really has very little overall effect. It really doesn’t much matter whether you listen at 77 dBSPL or at the much fainter level of 65 dBSPL. With Crescendo making up for your hearing loss, the end result is about the same in all cases.
Keep in mind that this is a simple model, which intentionally overestimates the exposure levels. So reading the chart to any level of accuracy would be a mistake. Use common sense and read the chart as a rough guideline.
For example, even though the model chart shows that 60 dB vTuning is above the safe 85 dBA level, it is only above by a tiny amount, well within the margin of error in the model.
So I take that to mean that 8 hours of listening, plus or minus a bit, would be perfectly fine. In fact, I have been doing so for up to 16 hours per day for the past 10 years. And over that time period, nothing has changed in my hearing levels. But everyone is different, and your hearing could be more, or less, prone to damage than mine.
[ Assuming that you calibrated your audio system for 77 dBA = -23 LUFS, you could place a Loudness meter on the aft side of Crescendo to keep a watch on the levels hitting your ears.
So, at 77 dBA = -23 LUFS set the meter to glow yellow above -15 LUFS. That’s where your 85 dBA level will be in each ear.
My meter isn’t showing anywhere near that much Loudness. I’m reading, typically, down around -17 LUFS while listening to “Massive Attack”. That says my exposure level is around 83 dBA. ]
[ So… I sat down and measured the R128 Loudness in each of the sample tracks I provided yesterday. How does that compare against the model we used?
R128 overestimates the importance of high frequencies. So there is a correspondence, but you can see departures. Those sample tracks will actually come in between the 65 and 77 dBA curves. Which, in turn, means that a Loudness meter will overestimate the actual loudness, and keep you safe. ]