We all know it's hard to argue with an audiophile claiming they can hear [fill in the blank]...
When I was at Intel, we did lots of work on waveform analysis. Power supplies responding to transients, memory sticks responding to address/data on digital lines - all looking at the quality of the waveform with an oscilloscope to suss out some distinguishing characteristic tha would predict trouble.
These of course were operating well above human hearing FR, from hundreds of kHz up into hundreds of MHz. We paid for some very expensive scopes to be able to see them, and then discern with our eyes the qualities to watch out for. A simple one was the non-monotonic rise time; nothing like getting two clocks out of one edge.
What if you transformed (via dsp) the frequency content of said waveform from whatever multi-megahertz frequencies down into the range of human hearing. Would you be able to reliably tell "this waveform has a non-monotonic edge - that one doesnt"? Or any other waveshape anomaly? One would think so, because;
The human ear must be pretty darn good, if it can hear differences in audio waveforms caused by stuff like, what wire, what tube, what capacitor, what rectifier, what driver - you probably got it.
Maybe better than an oscilloscope. Now if we could just get that 8-10 bit factor out of the way (they dont make 24 bit converters that run at 10 GHz) by a fully analog means to do the frequency shift - there you go - the full potential of the human ear as an analysis tool.
For things it wasnt designed to hear.
Comments?
When I was at Intel, we did lots of work on waveform analysis. Power supplies responding to transients, memory sticks responding to address/data on digital lines - all looking at the quality of the waveform with an oscilloscope to suss out some distinguishing characteristic tha would predict trouble.
These of course were operating well above human hearing FR, from hundreds of kHz up into hundreds of MHz. We paid for some very expensive scopes to be able to see them, and then discern with our eyes the qualities to watch out for. A simple one was the non-monotonic rise time; nothing like getting two clocks out of one edge.
What if you transformed (via dsp) the frequency content of said waveform from whatever multi-megahertz frequencies down into the range of human hearing. Would you be able to reliably tell "this waveform has a non-monotonic edge - that one doesnt"? Or any other waveshape anomaly? One would think so, because;
The human ear must be pretty darn good, if it can hear differences in audio waveforms caused by stuff like, what wire, what tube, what capacitor, what rectifier, what driver - you probably got it.
Maybe better than an oscilloscope. Now if we could just get that 8-10 bit factor out of the way (they dont make 24 bit converters that run at 10 GHz) by a fully analog means to do the frequency shift - there you go - the full potential of the human ear as an analysis tool.
For things it wasnt designed to hear.
Comments?
Why is anything in need to be transformed down to audible range?
Just create the signals you wish to analyse and have a go at it. Make it blind, keep statistics. Show your results here. And at these frequencies you can measure 20 bit at 200Khz with a low cost sound card - is that enough do you think?
//
Just create the signals you wish to analyse and have a go at it. Make it blind, keep statistics. Show your results here. And at these frequencies you can measure 20 bit at 200Khz with a low cost sound card - is that enough do you think?
//
Three separate 'environments' are in the mix now: the source of modulated air, the instrument to pick up this, and the unit to process it.
The first two can be described technically and analysis can predict the outcome. The last one has a variaty of nearly 8.000.000.000 possibilities currently.
The discussion will remain open.
The first two can be described technically and analysis can predict the outcome. The last one has a variaty of nearly 8.000.000.000 possibilities currently.
The discussion will remain open.
Perhaps the loss in signal integrity at high clock frequencies, where the incoming square wave emerges with significant ringing, could be considered analogous to the waveform degradation caused by our amplifiers and speakers at audio frequencies. Audio frequency ringing is visible on a 'scope, but is it audible?
I believe it is...... I just mentioned it yesterday in the bt thread.
Pre-ringing comes across as a odd chirp in the leading edge transients.....found this when playing around with phase timing in ypao (Yamaha dsp) . Problem is there’s no way of telling the parameters of adjustment, and I don’t have any test equipment.
So at this point it’s just conjecture.....take it fwiw.
"Audio frequency ringing is visible on a 'scope, but is it audible?"
"I believe it is...... I just mentioned it yesterday in the bt thread"
Clearly, if one does an FFT display on such a waveform, the spectrum will be different than that for such a waveform w/o it.
Regarding amplifier sound, people describe audible difference between amps with "monotonic decay" of (distortion expressed as) a harmonic series, even "exponential decay vs linear" can be perceived - according to some people's ears.
So the idea is, as an alternative to looking at a waveform display, listen to it. But the signal is in the MHz region - so shift the spectrum down into the audible range, turn it back into a time domain waveform (via inverse FFT or some analog transform) and listen to it. That's the idea.
Valid?
"I believe it is...... I just mentioned it yesterday in the bt thread"
Clearly, if one does an FFT display on such a waveform, the spectrum will be different than that for such a waveform w/o it.
Regarding amplifier sound, people describe audible difference between amps with "monotonic decay" of (distortion expressed as) a harmonic series, even "exponential decay vs linear" can be perceived - according to some people's ears.
So the idea is, as an alternative to looking at a waveform display, listen to it. But the signal is in the MHz region - so shift the spectrum down into the audible range, turn it back into a time domain waveform (via inverse FFT or some analog transform) and listen to it. That's the idea.
Valid?
Last edited:
Indeed!
From a biological perspective: primates were shifted out by their ability to distinguish the comfortable rushing sounds of long grass on the plain (our ticking clock in the living room) from the silent thump of a claw of a hungry predator (suddenly the clock stops and you notice).
So what we hear is different from what we think we hear.
So what we say is different from what we think we say.
So what we think is different from what we think we think.
This marvallous processing unit.
Anyone heard of the Rumsfeld quadrant? We're only aware of 0.1% of what is andor can be known.
Rumsfeld Quadrant (coined by learning):
Consider a piece of paper (A4 / Letter).
There is a top header.
There is a left border.
The crossing of both is your actual current knowledge (the known known)
The rest of the top header are your forgotten memories (the unknown known).
The larger rest of the left border is what you can imagine is out of your scope (the known unknown).
The rest is the void (the unknown unknown).
Given 90% of the human brain capacity is unused.
The void is this 90%.
The known unknown is 1/10 of the rest save the rest: 9%
The unknown known is 1/10 of the rest save the rest: 0.9%
The known known is the rest: 0.1%
Descisions are based upon the actual current knowledge (the rest).
In general, humans are utter stupid lazy opportunistic idiots.
And in their failure able to express compassion.
I love it!
Consider a piece of paper (A4 / Letter).
There is a top header.
There is a left border.
The crossing of both is your actual current knowledge (the known known)
The rest of the top header are your forgotten memories (the unknown known).
The larger rest of the left border is what you can imagine is out of your scope (the known unknown).
The rest is the void (the unknown unknown).
Given 90% of the human brain capacity is unused.
The void is this 90%.
The known unknown is 1/10 of the rest save the rest: 9%
The unknown known is 1/10 of the rest save the rest: 0.9%
The known known is the rest: 0.1%
Descisions are based upon the actual current knowledge (the rest).
In general, humans are utter stupid lazy opportunistic idiots.
And in their failure able to express compassion.
I love it!
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.