No need to introduce what-ifs - billshurv just stated that this 'test' shows "how good ears really are" & yet he stated vinyl noise is at a lower dB level & we all know how vinyl noise is easily heard by everyone.
These simplistic & agenda driven examples show the thread bare logic on display in this & the McGurk example used to similarly try to show some simplistic notion about how hearing
Stopped listening that's a good one, I listen all the time and enjoy it. As I said I have come across many folks that don't give jot about about so called better, realistic sound, and have an incredible depth of knowledge and love of the music. You don't realize how small a community you are.
EDIT - Better and realistic of course in the eyes/ears of some, see for instance the pitch corrections on Robert Johnston's 78 recordings.
Last edited:
Mark, this isn't really that interesting. George shared some undithered down sampling from 24 to 16 bit and the errors were what 30 db above the floor? Throw in a quiet passage and realistically how much are those errors below the 5-10s spot intensity?
Some people can hear distortion down lower, we just don't have the research to prove it to skeptics. Would you like to donate?
Doesn't matter all that much I guess. One can probably devise an experiment to show some humans can't hear much at all, and another experiment to show a few humans can detect seemingly extraordinarily small levels of distortion. There may even be experiments to prove some of the same people show both effects. It would only mean hearing can be quite non-linear in ways we don't fully understand, and maybe that there can be a lot a variability between individuals. Trying to draw conclusions about the general population from limited sample sizes and using less than well-optimized test procedures is silly.
It is probably true that if you test a few people in the exact same ways that was done in the past you will likely get similar results. Don't know how much that means either. Not much in my book.
We need new research and improved protocols if people really want to know who can hear what. Otherwise this is a big time waster.
Doesn't matter all that much I guess. One can probably devise an experiment to show some humans can't hear much at all, and another experiment to show a few humans can detect seemingly extraordinarily small levels of distortion. There may even be experiments to prove some of the same people show both effects. It would only mean hearing can be quite non-linear in ways we don't fully understand, and maybe that there can be a lot a variability between individuals. Trying to draw conclusions about the general population from limited sample sizes and using less than well-optimized test procedures is silly.
It is probably true that if you test a few people in the exact same ways that was done in the past you will likely get similar results. Don't know how much that means either. Not much in my book.
We need new research and improved protocols if people really want to know who can hear what. Otherwise this is a big time waster.
My point is that we can erode down a spot listening test from undithered 16 bit errors very easily to > -40 dB below the 5-10 spot intensity of the track. That's not even the ~ -60 dB in literature, nor the extra masking from Bill Waslo's sousaphone test sitting -50 dB below nominal.
It's not the same thing as claiming 16 bit has 96 db dynamic range and people are hearing errors in LSB of a perfect signal. People hearing undithered errors in 16 bit sources doesn't require new understanding to say, "gee that's very plausible".
It's not the same thing as claiming 16 bit has 96 db dynamic range and people are hearing errors in LSB of a perfect signal. People hearing undithered errors in 16 bit sources doesn't require new understanding to say, "gee that's very plausible".
Last edited:
Dithering doesn't change the file all that much. What's changed is small. What is different is mostly the way brains process the difference into conscious awareness. One thing attracts attention and the other thing doesn't. Same with music mixed in at a low level, brain doesn't latch onto it as something to forward to conscious awareness. However, some trained neurons in a particular brain may latch onto much smaller differences in a 24-bit file encoding a little high-order harmonic distortion as something that does warrant bringing to attention in conscious awareness. Sometimes such things can be smaller than some people believe. Doesn't help and it hurts a lot that untrained audiophiles claim to hear stuff that they don't, most of which by the way are perceptually the same as volume level changes. Small distortion does not perceptually sound like volume level changes at all, as it happens. Still all a time-waster. Don't know why I bother, I should just shut up. My only intention was to point out all this is a time-waster, maybe that wasn't clear.
Nice insult. I will answer in a more measured way that you have.
The evidence is here, you are quite welcome to come around and bring any amplifier that you choose. In fact I welcome it and unlike you I am not seeking a battleground to bolster anybody's ego, yours or mine.
Let the ear be the final arbiter, as imperfect as it might be, I don't see any instrumentation as a better substitute, do you?
And yes, I do think measurements matter, so I have the middle ground. But I have no need to be in the "prove it" business.
Joe
Mark, go back and look at the errors George was generating in his samples when we discussed dither then truncate vs truncate then dither. Or generate them yourself. The propagation errors can be quite large.
Daniel, Way back then I was the one arguing that dither first then truncate is the correct procedure. When you dither properly some of the dither is below the new LSB, and some numerical carrying will occur that affects higher order bits. The incremental difference is still small. I don't see how it could be otherwise.
EDIT: Look, the same thing happens with images that are bit-depth reduced, you get jagged lines that attract attention. Dither makes the jagged edges look blurry instead. Blurry stuff is normal to us and does not attract much attention. There are not large visual errors if you fail to dither, unless you mean something that looks like a straight line for a long time before there is one step.
EDIT: Look, the same thing happens with images that are bit-depth reduced, you get jagged lines that attract attention. Dither makes the jagged edges look blurry instead. Blurry stuff is normal to us and does not attract much attention. There are not large visual errors if you fail to dither, unless you mean something that looks like a straight line for a long time before there is one step.
Last edited:
What was the insult, stating that you showed no evidence that two modern solid state amplifiers with <0.01% distortion and >100KHz +/-0.1dB bandwidth are distinguishable in a double blind test? This, and nothing more or less, was my original claim.
If so, I’ll double down the “insult”. You only showed a preference for an amplifier with zero GNFB vs. a solid state amplifier with GNFB, good for you.
As of coming over, I would be happy to, if somebody would pay for this trip and all the associated costs. Otherwise, I’m afraid I’ll have to pass, my idea of wasting time has a completely different drift.
Hardly, such untrustworthy hearing should never be believed - only measurements will do for billshurv
I spend a fair amount of time measuring, but I spend more time listening. 😀
Here is an interesting two questions re measuring speakers, how many frequency responses does a speaker have and how many power responses does a speaker have?
Anybody want to try answer those, is there even a correct answer?
Joe
This is George's post of interest, which corroborate with what I saw as well, albeit my dB levels were higher and coming out of the 96 dB floor.
John Curl's Blowtorch preamplifier part II
Okay, the magnitude of quantizing noise due to truncation is down around the 16th bit. That should have some limit. So, if the signal level is down near the same level as the quanitzing noise is and you turn up the volume you can hear lots of quantizing noise. As far as I'm concerned that would be a cheat. It is completely unnecessary to do that to hear the absence of dither.
But, if your point is that if a test CD were incorrectly prepared so as to make it possible to cheat then the results of the test would be unreliable, then can't argue with that, I guess.
Last edited:
- Status
- Not open for further replies.