Re: 11 Bark channels

Well, it was felt by our board and the investors that the marketplace for Musical Correction for Impaired Listeners would be just too small a market to pursue. They had ideas in mind about developing an IP Portfolio (hence the Patents) and selling to some larger company as a quick exit strategy.

Typical starry eyed bean-counter MBA mentality. For myself, I am an Astrophysicist. What the heck do I know about marketing and sales?

My own interest has always been making musical timbre-correct spectral pre-warping of sounds to present to people like myself with hearing damage. The Telephony applications, while certainly performing as advertised, has wider audience but far less interest from me. I, personally, can’t stand using telephones.

I have always dreamed of a wearable device. But in the early days, even a Pentium computer was unable to perform all the computations necessary. Only power-hungry DSP’s could do the job. Of course now today we have M1 and beyond, and I’m seeing a loading between 3-5% of CPU. But Crescendo has also become more simplified than it once was. 

I now recognize that our sense of loudness perception is nowhere near as acute as our pitch perception. And so you can get away with far fewer bands, and imperfect alignment of Bark channels against the self-organizing bands. 

And gain approximations all sound reasonably good. So you don’t have to have the exact gain difference computed as the difference between the Conductor model for impaired hearing and the EarSpring model for normal hearing. Those two equations are both nonlinear differential equations that takes some horsepower to compute, doing this in stereo, and across 100 channels on each side. 

Turns out to be totally unnecessary. You really can get away with a linear compressor at about 4.5:1 compression ratio, mixed with direct sound. Make the compressor gain at the elevated threshold be 9.8 dB, and have the compressor threshold about 20 dB below the hearing threshold. That works well enough. And do this in each Bark band, and on Left and Right. That is 22 subsystems in parallel, at roughly 300 Hz update rate.

But you do need to do this across more than 3 or 4 bands. So, as long as you are going to do 10-11 bands, you might as well just use the correct analytical solution that I finally obtained.

In those early years I looked at the cubic equation for Sones and decided it would need numerical solving. Not true at all. You really can develop an analytical solution as I show in the writeup. And that solution is a lot faster to compute than a numerical approximation.

So, color me a Scientist, not a business man. While I have been in business for myself for many years, on and off, I really thrive if I can just bury my head into deep thinking and computing.

Economically, my life has been one of jumping from one iceberg to the next. Always managing to thrive, and save up some money. But now, when I enter the phase of life when you stop saving and start spending, I find that the world governments have pissed away all the value in our money. That we are all being slowly drowned by currency devaluation.

But we have more than most people, and not as much as some. And I only have approximately 20 more years at the outside. My concern right now is ensuring that my wife will be able to sustain herself after I’m gone. She is likely to live past 100. She is 76 now, and looks 20 years younger than I do.




> On Nov 4, 2025, at 13:00, Yuri Davidovsky <[email protected]> wrote:
> 
> 
>> On 4 Nov 2025, at 20:08, David McClain <[email protected]> wrote:
>> 
>> The plugin runs on Mac as an AudioUnits plugin. That means you can run it in any AU Host program, any DAW, or in Rogue Amoeba’s SoundSource.
> 
> Hm. Since you actually have an AudioUnit plugin implemented, then you already have the product ready to go. Why didn’t you try to market that at the time instead of trying to chase the phone producers?
> 
> Especially given that, as you said, sound engineering people have asked you about that. Or did you have a specific use case for the algorithm in mind (sort of a wearable device for musicians and the likes) and decided that a software solution would not realise its full potential and won’t be as useful for people as it should?


_______________________________________________
Lisp Hug - the mailing list for LispWorks users
[email protected]
http://www.lispworks.com/support/lisp-hug.html