Digital circuitries don’t work like that, meaning that they don’t go dormant to save power when a low noise level is detected (thereby saving power drain), then they get activated when the signal shows a lot of noise.
Just the sampling rate for hearing aids alone are at 20 KHz, meaning that regardless of whether the input signal is hyperactive or dead quiet as a mouse, the A2D converter CONSTANTLY always samples at 20,000 times per second and converting the analog input signal into its ones and zeros equivalent.
After the input signal has been digitized, it goes through the DSPs (digital signal processors), and these guys are BUSY bees at work analyzing and processing the input signals. The filter banks after the A2D converters convert the input signal from the time domain into 64 channels of equal bandwidth to enable more advanced mathematical processing down the line.
The DNN in the Oticon HA is forever CONSTANTLY busy as well, scanning and analyzing the input sound scene at the rate of 500 times per second to correctly map different sound sources. A quiet sound source doesn’t really shut off the DNN or part of the DNN to save power. It’s more about identification and distribution and processing accordingly, regardless of the content inside.
I can go on and on, but to not bore you with more details, maybe I can give an analogy. Back in the days of analog movies, a 2 hour movie will take up a big reel of film at 60 frames of film per second. Regardless of whether those frames are blank (black or white or a monotone color) instead of a very pretty and colorful picture, it’d still take 60 frames of the monotonous non-moving blank screen displayed on the screen, and the movie projector continues to keep reeling at a constant speed regardless, sucking up the same amount of power from the wall to light up that projector’s light bulb. The content of what’s on each of the frame doesn’t matter. The machine is designed to do its thing the same exact way regardless of the content in the movie.
Of course this is an over generalization, but hopefully you get the picture. Of course it all depends on the design. For digital circuitries, the power drain comes when the transistors inside a logic gate inside a circuit open and close to switch logic states. But digital signal processors operate at a much, much, much higher level than at the transistor level or the logic level. It’s more or less almost at a system level, as each DSP inside the SOC (System On the Chip) is kind of a system by itself, dealing at a much more abstract mathematical level for the most part. So minute tiny control over how to save power at the transistor or logic gate level is non-practical and not really a concern when you’re flying at a 20 thousand foot level. Power saving concern would be focused more on the efficiency of the analog components because they’re much more low lying fruits to reap, and also on the choice of the geometry of the silicon (the lithography scale) which will have to be balanced against the cost of going to lower, more advanced geometry for power saving.
