How Science Got Sound Wrong

mahgister,

The timing resolution, within the bandwidth limited system, is orders of magnitude better than the sampling rate at even fairly low SNR. Within a bandwidth limited system, which the "mechanical" auditory system starts as, continuous time can be represented, effectively perfect, by sampled data points. It really is the same under those limits of bandwidth and there is absolutely nothing to indicate our acoustic perception is not bandwidth limited.

There have been experiments done on temporal resolution w.r.t. bandwidth of the signal. At some point, increasing bandwidth stops increasing temporal resolution, and that stop point is within the limits of the audio bandwidth as commonly discussed.

Unfortunately, the author seemed to be writing from a position of "feelings" as opposed to well researched science. There is research, math, even well accepted scientific knowledge that directly refutes his position.

Atdavid
you persist to reduce, to a simple translation in a simulation model, the hearing processings in the body-brain sensors (including more than just the 2 ears sensor) a process more complex, that implicate more synchronized continuous mappings than in one Shannon simulation channel... Then how can i say my point that is invisible to your fixation ?

The successive operation of compression and decompression of a natural object,be it a sound or a nude woman, is possible with a great degree of accuracy, nobody discuss that,neither Softky nor me, but there is a price to pay? Guess what is the price? This is the central point around the Softky argument, and the precise amount of information available via Fourier analysis in a Shannon channel, be it the information amount you point at, or the different amount of information Softky point at, whatever amount of information   there is available will not change the central fact pointed by Softky ’s article , that is to say, the price to pay...


Here is some bits of info that will gives you the beginnings of the answer:


«But variability in the digital world has a very different structure from the “noise” known to science. In one sense, digital variability is lower,having been specifically enriched to appear to our sensory systems as coherent 3D images or sounds rather than as random snow or hiss. In that sense, moment-to-moment digital inputs are designed to seem low noise and clean. But digital sources are hyperdimensional patterns, which (unlike real things) can change discontinuously, thereby violating the continuous natural laws a nervous system expects. The unnatural structure of digital variability can make it appear far more trustworthy and predictable than it actually is.» William Softky

mahgisterAtdavid "you persist to reduce to a simple translation in a simulation model the hearing processings in the body-brain sensors(including more than just the 2 ears) a process more complex, that implicate more synchronized continuous mappings than in one Shannon simulation channel... Then how can i say my point that is invisible to your fixation ?"

There is not much you can do to help him understand the matter is beyond his limited understandings which are acquired mostly from Wikipedia so he can not even ponder, develop, and articulate a coherent, reasoned, logical response but if you point that out he will say something illogical such as "you don't like handicapped people" or "French toast is better than "fritters." 

clearthink,

What spectrum were you talking about in your "on the spectrum" statement? It is quite unclear, I think.

mahgister,

I think you are trying to attach far too much complexity to this topic that is not a factor of the article.  It does not matter if there is 1, 2, or 1000 sensors, there is still just 2 signals, right and left, or perhaps more in a surround system. We only need to get those 2 signals right, no matter how many sensors there are.


A digitized and reconstructed analog signal with a given bandwidth, is no more compressed and decompressed than an analog signal that is stored and played back through a system of limited bandwidth. Arguable, modern digital systems "compress" far less, where "compress" is meant to mean throwing away data.

Yes, it really is as simple (or not so simple) as Shannon-Nyquist. Shannon-Nyquist even predicts that you can hear tones/signals that are lower in amplitude than the signal to noise ratio, both for analog and digital.  For all the talk about micro-timing, etc., there is no evidence that our auditory systems is anything but bandwidth limited. In fact, the experiments clearly show that these "micro-timing" "events", i.e. detecting time of arrival within <5 microseconds, is not improved by increased audio bandwidth. 


This statement below by Sofky is just lazy gobblygook.  Digital "variability" is neither lower nor higher, but digital implementations are lower in noise typically. If you don't like that, then simply add some noise back at playback. A reconstructed digital signal in the analog domain is absolutely as continuous as any other bandwidth limited signal. "Continuous natural laws"? ... huh?  There is nothing "unnatural" about the structure of digital "variability". That is well, a really weird statement, that I can only see someone who has a distorted understanding of bandwidth limited systems making. Hyperdimensional ... that is some hyperbole on his part. No, same dimension as the "real" world, just storing it in a different and more accurate fashion.

But variability in the digital world has a very different structure from the “noise” known to science. In one sense, digital variability is lower,having been specifically enriched to appear to our sensory systems as coherent 3D images or sounds rather than as random snow or hiss. In that sense, moment-to-moment digital inputs are designed to seem low noise and clean. But digital sources are hyperdimensional patterns, which (unlike real things) can change discontinuously, thereby violating the continuous natural laws a nervous system expects. The unnatural structure of digital variability can make it appear far more trustworthy and predictable than it actually is. (Sofky)