For cases of serious impairment, stacking two Crescendos works black magic! It produces a vast soundstage that you could just fall into, without any hint of flutter or scintillation in the faintest sounds. And here’s why…
This graph show a situation for a Bark band with 60 dB of threshold elevation. That’s pretty serious. The green diagonal line is ideal hearing – what comes in is what you perceive. The red curve is the uncorrected hearing recruitment.
By placing a pre-Crescendo ahead of one tuned for vTuning = 60 dB, the same as would be needed for a single stage of correction, but with the pre-Crescendo tuned with vTuning = 44 dB, it produces that orange compression curve acting on the sound field. It lifts the faint end, making it easier for the post-Crescendo to treat. And that is the curve that a perfectly tuned Crescendo would meet in its corrections.
By slightly detuning the post-Crescendo, we can produce that faint red curve between the orange compression curve and ideal hearing.
It is easier for Crescendo to be quite accurate when its correction gains are below 20 dB of boost. We get above that level when we approach the threshold levels of the recruitment hearing. The recruitment curves get very steep, making the slightest fluctuations in the sound, or errors in Bark channel power estimation, become exaggerated. And that is what causes faint sounds to have a fluttery or scintillating character, when you use only a single stage Crescendo for your corrections.
We can optimize the gain in the pre-Crescendo by looking for what vTuning would cause the gains in the two stages to be equal when the presentation level is at the room floor, down there at 30 dBSPL. That turns out to be very close to 44 dB, in the case of 60 dB threshold elevation.
Here is a graph showing the gains in the two Crescendo units versus presentation level:
Both units operate at maximum gain down there at the room noise floor, and they are using nearly equal amounts of boost to bring that sound up to the hearing level needed by 60 dB elevated threshold.
Furthermore, the steep slope in the recruitment curve gets significantly reduced by the actions of the pre-Crescendo, as shown here:
The pre-Crescendo behaves as though it is correcting a recruitment curve at its vTuning threshold elevation, which is much milder for 44 dB elevation, than for 60 dB. That is shown in the green curve.
The effect of the pre-Crescendo is to boost the input signal so that the gains applied by the post-Crescendo are now much lower than they would have been, and the slope for that boosted signal is shown by the red curve.
The pre-Crescendo produces a compressed signal for the post-Crescendo, so its slope (green curve) actually divides the slope of the post-Crescendo, producing that blue curve as the result.
For comparison, the magenta curve shows the recruitment slope used by a single-stage correction Crescendo. It goes through the roof as the presentation level declines, reaching 10:1 at the room floor level of 30 dBSPL. That means that any slight fluctuation in the sound, or minor error in power measurement, gets multiplied to become 10 times larger, in dB space. So a 0.5 dB fluctuation becomes 5 dB, which is clearly audible.
As you can see, the blue curve indicates that fluctuations will be well contained. And listening to stacked Crescendo’s verifies this as fact. The sound stage become so clear and wide that I could almost fall into it…