High End Audio and the Domain of Time

[Stinius]

I am not commenting on stupid videos, sorry. Not interested. I am just saying that ITD 5 - 10 us is possible to detect in a properly made test. I hate oversimplifications based on silly calculations of us to mm and then stating it is impossible. I hate all the oversimplification that ASR is full of. This forum declares itself as scientific, but it is not. It is rather a populistic forum.

Well said, I agree.
Stein

 

[Pdxwayne]

What your hearing is the interference between the 2 sounds in the same freq. range. If you brickwalled 2 sounds so none of there freqs. overlapped I doubt you would hear a difference. So your speaker eg. dosnt do that. And a 1ms delay would be a foot of displacement between the tweeter and woofer!

Hmm? Drum kick vs hi-hat in same freq? Did you even try the test?

 

[danadam]

Except that with modern easily-available software, even amateurs can create test signals that will allow them to determine their thresholds for timing and localization. You could do it, I could do it, lots of other people who aren't working principally in the audio field can do it.

It's a golden age for the curious folks.

Yep. Here are some, starting from 5us: https://hydrogenaud.io/index.php?topic=107570.msg899713#msg899713

 

[sweetsounds]

I do think that his research is relevant. The time domain has been played down and frequency response had become the de-facto criterion in many aspects, arguably, it is much easier to measure. Look at the loudspeaker tests on this site and others, phase errors and delays aren't visible.

Yet, what makes loudspeakers sound so different, even if they have all have flat frequency response?
My guesses: wide vs. narrow beam patterns / omnidirectionality, resonances/decays, diffraction on the bevel edges or on steps on the cabinet and as in this paper the step response / time+phase alignment of the drivers.

On this test you can easily score 10/10 for 1ms in (https://www.audiocheck.net/blindtests_timing_2w.php?time=1) with a desktop speaker.
So you should be able to hear alignment differences between woofer and tweeter as in here: https://audiojudgement.com/loudspeaker-step-response-measurement

1ms equals 42cm and time alignment of multiple drivers on a larger speaker is thus difficult. First of all, they probably would need to be coaxial to avoid seating position errors. Second, you will have comb filtering in the cross-over region(s).

 

[amirm]

As noted, he is famous for saying stuff that is easily shown to be incorrect. Just think of timing in a room. You don't just hear the direct sound but all the reflected ones with their delays. How on earth is phase accuracy mattering in that regard?

That wilson speaker creates a ton of diffraction with those edges of every driver. How is that being faithful to timing? They will all sing a different tune.

 

[CinDyment]

I do not understand much of this but it would be interesting to hear what you think about what being talked about in this video.

He's a quack. Just remember he did an experiment that showed an audible difference between balanced and single ended connections, but never validated either system to ensure they were working correctly.

This video is painful. He makes a bit point out of "timing" but really that was two completely different attacks. Of course it is audible!! Then he takes two drivers and claim a 5usec change is audible by moving two devices. NOT an engineer. If he was even a proper scientists, he would do a frequency, magnitude and phase drawing which would show a difference. He lacks even a rudimentary knowledge of signal processing and does not seem to understand transforms from time to frequency domain. Quack.

 

[CinDyment]

Are you implying that because ITD is a wel established phenomenon, that there is nothing wrong with the video? Or are you referring to the video as amateurish? Production quality definitely is, but the dude has some serious accreditations to show..

Not in audio he doesn't.

 

[CinDyment]

I do think that his research is relevant. The time domain has been played down and frequency response had become the de-facto criterion in many aspects, arguably, it is much easier to measure. Look at the loudspeaker tests on this site and others, phase errors and delays aren't visible.

Yet, what makes loudspeakers sound so different, even if they have all have flat frequency response?
My guesses: wide vs. narrow beam patterns / omnidirectionality, resonances/decays, diffraction on the bevel edges or on steps on the cabinet and as in this paper the step response / time+phase alignment of the drivers.

On this test you can easily score 10/10 for 1ms in (https://www.audiocheck.net/blindtests_timing_2w.php?time=1) with a desktop speaker.
So you should be able to hear alignment differences between woofer and tweeter as in here: https://audiojudgement.com/loudspeaker-step-response-measurement

1ms equals 42cm and time alignment of multiple drivers on a larger speaker is thus difficult. First of all, they probably would need to be coaxial to avoid seating position errors. Second, you will have comb filtering in the cross-over region(s).

Time domain is not played down. It also has never been shown to be highly relevant either. Experiments have shown that relatively massive phase shifts are needed before something is audible. Several experiments over the years on audibility of group delay and you need something fairly substantial before audible.

Distortion, directionality and how it varies by frequency and room interaction are hugely dominant.

 

[SIY]

Time domain is not played down. It also has never been shown to be highly relevant either. Experiments have shown that relatively massive phase shifts are needed before something is audible. Several experiments over the years on audibility of group delay and you need something fairly substantial before audible.

Distortion, directionality and how it varies by frequency and room interaction are hugely dominant.

Moot, if you have frequency domain, you have time domain and vice versa.

 

[voodooless]

Not in audio he doesn't.

He teaches 2 courses on acoustics… so there is that…

I wonder if you can pass those by using the same kind logic he uses in his video

 

[CinDyment]

He teaches 2 courses on acoustics… so there is that…

I wonder if you can pass those by using the same kind logic he uses in his video

He teaches 1 acoustics course, and the lab that goes with it .... To answer you question, I believe the answer is yes:

We had a bird course like this at my university, colloquially called "Hi-Fi Sci". Engineers were not allowed to take it (too easy). Shame as you had a 50/50 chance of getting a professor who actually was a world renowned expert.

 

[CinDyment]

Would it be wrong to call the author of this drivel an idiot? ... well other than figuring out how to game the system and keep a job while doing the minimum of "research" to keep it.

Take this paper for instance. This is the work of a tenured professor. It reads more like a 4th year project. In fact, it is far worse than what I would have expected from a 4th year. A source resistance of 1Meg ohm to measure the frequency response of an audio interconnect? No interpretation of the audible impacts (or lack of) when transients decay from reflections in <1usec, not to mention, the reflections are way outside audible limits in the frequency domain. No realistic or valid discussion of the RF noise spectrum (note how low it is). The "paper" is a total joke:

http://boson.physics.sc.edu/~kunchur//papers/Interconnect-cable-measurements--Kunchur.pdf

 

[voodooless]

Would it be wrong to call the author of this drivel an idiot?

Yes

... well other than figuring out how to game the system and keep a job while doing the minimum of "research" to keep it.

So, you proved my point. The man is clearly a genius… an evil one though.

Take this paper for instance. This is the work of a tenured professor. It reads more like a 4th year project. In fact, it is far worse than what I would have expected from a 4th year. A source resistance of 1Meg ohm to measure the frequency response of an audio interconnect? No interpretation of the audible impacts (or lack of) when transients decay from reflections in <1usec, not to mention, the reflections are way outside audible limits in the frequency domain. No realistic or valid discussion of the RF noise spectrum (note how low it is). The "paper" is a total joke:

http://boson.physics.sc.edu/~kunchur//papers/Interconnect-cable-measurements--Kunchur.pdf

All of this is clearly deliberate. I wonder what the actual motives are for this guy to do all of this?

 

[CinDyment]

All of this is clearly deliberate. I wonder what the actual motives are for this guy to do all of this?

Conspiracy nut who has never done a blind test on proper music where the system was validated (unlike his single ended / differential comparison), and/or something is paying him.

 

[SIY]

All of this is clearly deliberate. I wonder what the actual motives are for this guy to do all of this?

"...get him attention, his hobby paid for, and some extra income."

 

[Cbdb2]

Hmm? Drum kick vs hi-hat in same freq? Did you even try the test?

9 out of 10 at 1ms. The difference is in the tone, which usually means they have some overlapping freqs.
"Typical hi-hats are usually between 300-3000 Hz dominant frequencies, and can extend up to 10-17k Hz for crispness, “air” and sparkle. The “shhhhhhhhh” sound is usually found at the 2-3k Hz range."
And the kick drum snap is usually above 1khz.
Theres a thing called the internet where you can learn these things.

 

[Pdxwayne]

9 out of 10 at 1ms. The difference is in the tone, which usually means they have some overlapping freqs.
"Typical hi-hats are usually between 300-3000 Hz dominant frequencies, and can extend up to 10-17k Hz for crispness, “air” and sparkle. The “shhhhhhhhh” sound is usually found at the 2-3k Hz range."
And the kick drum snap is usually above 1khz.
Theres a thing called the internet where you can learn these things.

Wow. Such expert. Do explain to me, why for certain combo at certain timing, there is no tonality change?

See my observations at

DAC and amp combos did not give same clues when running online blind tests. Why? What would be the desired clue?

When running online blind timing tests yesterday, I heard different clues when using multiple combination of dac and amps. I wonder why. I thought, as long as I am using transparent DAC and amp, I can mix and match them and any combo will sound the same. But not this case. Is it okay? More...

www.audiosciencereview.com

 

[KSTR]

Is this generally what you have been referring to?

No. More like that a 1kHz 0dBFS distortion magnitude spectrum is not very relevant for a DACs "signature". Look at signals around -20dBFS and especially look at the time-domain representation of the error signal -- for any kind of stimulus, not just single steady-state sines.

 

[SIY]

No. More like that a 1kHz 0dBFS distortion magnitude spectrum is not very relevant for a DACs "signature". Look at signals around -20dBFS and especially look at the time-domain representation of the error signal -- for any kind of stimulus, not just single steady-state sines.

Can you give an actual example?

 

[Blumlein 88]

No. More like that a 1kHz 0dBFS distortion magnitude spectrum is not very relevant for a DACs "signature". Look at signals around -20dBFS and especially look at the time-domain representation of the error signal -- for any kind of stimulus, not just single steady-state sines.

Can you give us some detailed examples of DACs with a different signature and how we would test them with your -20 dbFS signals to uncover those issues? I'm being serious and genuine in asking. What results would tell us what about the signatures?

SIY beat me to it.