advertisement


MQA pt II

mansr said:
The 'fir' effect can be used to apply an arbitrary filter (up to some maximum length).
Click to expand...

Ah! Thanks. I can now recall having noticed that in the past when reading though the bumf on sox. But had, naturally, totally forgotten! So we could probably define a sox filter to undo the daft added dispersion. Snag being that people would have to capture the DAC output and apply it to that! Not very useful in practical terms. But potentially useful for investigation purposes.

BTW For my current tests I've added a 'tail' to each MQA file. This is for calibration and time-alignement-flagging purposes, mainly. But also lets me check the response curves, coping with 'overs', etc.

In most cases the MQA encoding switches off abruptly when it hits the 'tail'. I then get vanilla dispersion on impulses. But in one case so far the MQA seems to have disengaged 'late' so I may have also caught the impulse function it was applying in addition to any blanket DAC+ADC effects.
 
adamdea said:
snap- I was looking at the Ashihara results the other day and thinking- Would any of the audible tones be "legal" signals according to the MQA encoder?
Click to expand...

Interesting question, innit. 8-]

1/f 'law' meet human hearing. You can't have your kayak and heat it. :)

The Ashihara arrangements seemed particularly interesting to me because in effect they seem to have added the HF tones to pink noise.
 
Jim Audiomisc said:
Interesting question, innit. 8-]

1/f 'law' meet human hearing. You can't have your kayak and heat it. :)

The Ashihara arrangements seemed particularly interesting to me because in effect they seem to have added the HF tones to pink noise.
Click to expand...
Ami I right in thinking that the noise was to mask intermodulation artefacts?
 
Last edited:
IIUC, yes. The nice thing from our current examination POV is that this acts also as a '1/f must be music' sort of background to hearing the presence/absence of HF tones. cf another thread
 
@kaolin
It would be nice to be absolutely sure so Jim can make a direct comparison between MQA and non-MQA versions that come from the same (re-)master file. Perhaps you can tell Jim how you know this.
 
Last edited:
Rosewind said:
kaolin.
It would be nice to be absolutely sure so Jim can make a direct comparison between MQA and non-MQA versions that come from the same (re-)master file. Perhaps you can tell Jim how you know this.
Click to expand...

Provenance is a bit of a mystery for most hi-res material isn't it?
 
Yes. It is tempting just to opsamle redbook to hi res and cash in. MQA deliberately witholds all provenance details to smokescreen any real comparison. Tidal don't tell us either and may even use MQA'd files as the bases for the LPCM Flac files they offer.
 
leroyd said:
I recently purchased this
https://www.highresaudio.com/en/album/view/xhsgfw/rory-gallagher-rory-gallagher-remastered.
It sounds like a bag of spanners, when compared with the red book version.
Click to expand...
A 1971 recording on Scully 8 track from before the Advision Studio got the Neve desk. I doubt there was much there above 20 kHz in the first place.
A proper transfer should fit nicely into Red Book (provided the 8 track tape is in good condition and even exists)
M should be Mix down, because anything two track is not the Master.
 
davidsrsb said:
A 1971 recording on Scully 8 track from before the Advision Studio got the Neve desk. I doubt there was much there above 20 kHz in the first place.
A proper transfer should fit nicely into Red Book (provided the 8 track tape is in good condition and even exists)
M should be Mix down, because anything two track is not the Master.
Click to expand...

well, the label wants to remastered in 24/96kHz space.
 
Jim Audiomisc said:
Interesting question, innit. 8-]

1/f 'law' meet human hearing. You can't have your kayak and heat it. :)...
Click to expand...
What do you think about the idea, which I have seen put forward to support extended frequency response, that the human hearing system doesn't do a linear Fourier analysis so human frequency resolution (and response?) may not be the limiting factor in human temporal perception?

Here for example. It's well outside my field so I have noted the idea but I have no opinion.
 
John Phillips said:
to support extended frequency response, that the human hearing system doesn't do a linear Fourier analysis
Click to expand...

Nobody in the field would claim that the auditory system does do a linear Fourier analysis. What happens in the cochlea is pretty well known.

so human frequency resolution (and response?) may not be the limiting factor in human temporal perception?
Click to expand...

The acuity of inter-aural temporal perception is well established to be a handful of microseconds. This has (luckily) nothing to do with the ear's frequency response, though, and as such digital audio is not a limiting factor here.

Here for example.
Click to expand...

That was a somewhat interesting study, let down by its sensationalist title. The uncertainty principle was not tested: the subjects knew what to listen for. In other words: they had additional information.
 
John Phillips said:
What do you think about the idea, which I have seen put forward to support extended frequency response, that the human hearing system doesn't do a linear Fourier analysis so human frequency resolution (and response?) may not be the limiting factor in human temporal perception?

Here for example. It's well outside my field so I have noted the idea but I have no opinion.
Click to expand...
aside from the inter-aural time difference issue identified by Werner, I spent some time, as an amateur, trying to identify what was known about temporal resolution of the hearing system a few years ago. I would love to hear from someone expert about it (any lurkers do speak up). My recollection from what I could find was that there was no unique measurement but tests had been done on minimum detectable interval (ie gap) measurement and minimum threshold at which differences in duration could be detected and that these were in the millisecond range. I think there is some learning in the area around age related hearing loss (predictably as this is a field where there is money to be made).
The striking thing amongst all the guff about hi res / time domain etc marketing/ white papers is that it seems not to be based on any concrete evidence as far as I could find. again all non expert on my part but if one recalls the MQA marketing blah about the importance of twigs snapping it seemed to me to have no connection to the extensive research about how distance and direction perception works.
The notion that this is really about cutting edge maths / engineering seems to be a slight misdirection: it would be great to hear from some perceptual scientists
 
This one is always abused with the old staircase DAC myth to imply that Red Book cannot encode time differences finer than the sampling rate, which is just wrong
 
John Phillips said:
What do you think about the idea, which I have seen put forward to support extended frequency response, that the human hearing system doesn't do a linear Fourier analysis so human frequency resolution (and response?) may not be the limiting factor in human temporal perception?

Here for example. It's well outside my field so I have noted the idea but I have no opinion.
Click to expand...

As ever, these areas tend to be infested with misunderstandings and poor experiments. It is the case that human hearing is nonlinear. *As are devices like headphones and loudspeakers.* That's one reason I chose a paper that used pink noise to mask intermod effects leading to a 'hearing' HF due to a non-human nonlinearity.

More generally, the loss of a sense of pitch at *audible* HF means we will tend to simply treat what can he heard as 'added salt' bringing up the 'taste' of a given pitch *defined by lower frequency components*. However that's an argument for liking HF distortion, not MQA being sounding like the 'Master' because of what it hides or fiddling with the ADC's dispersion.

FAQE again... :)

So no, I don't believe that you can have a 1/f-ish spectrum and still hear the stuff > 22 kHz when it *60dB below* the threashold people can measure in tests like the once I cited. I've never seen any measurement that even gives a plausible reason to accept it. At those levels, the ear will be essentially add its own sense of harmonics at a higher level I suspect because of any hearing nonlinearities.

If someone can come up with a *reliable* set of test results to the contrary I'd be interested. But I've read papers on this for decades and not found any that seem plausible. Too many errors and sources of alternative reasons for what was 'heard'.

Note that although some people can hear tones well above 20kHz, they tend to need to be at a relatively high level... which MQA would say isn't 'music'. They can't have this both ways.
 
Jim Audiomisc said:
As ever, these areas tend to be infested with misunderstandings and poor experiments. It is the case that human hearing is nonlinear. *As are devices like headphones and loudspeakers.* That's one reason I chose a paper that used pink noise to mask intermod effects leading to a 'hearing' HF due to a non-human nonlinearity.

.....

Note that although some people can hear tones well above 20kHz, they tend to need to be at a relatively high level... which MQA would say isn't 'music'. They can't have this both ways.
Click to expand...
One additional point, my recollection is that pitch discrimination is very low at high frequencies and that everything in the top octave or bin (at least of the conventional 20-20khz range) sounds the same. This would imply that even if you could hear very high frequency tones, when they were incorporated as harmonics (the only way they are likely to get there in music) they might not add much more than a tiny bit (NB 1/f) of the same high pitch. I accrept that this might not hold good in the >20khz range (since the dramatically declining hearing threshold as frequency rises up to 20khz seems not to apply and maybe this range sounds different)-

Does anyone know whether anyone has done pitch discrimination tests at these Ashihara frequencies? Also whether there is a 20khz+ part of the basilar membrane which responds to these Ashihara frequencies or whether it is the lower freuncy sections (ie confirming or not at the physical level the argument that the subjects can;t be responding to intermodulation?)
 
Jim Audiomisc said:
As ever, these areas tend to be infested with misunderstandings and poor experiments. It is the case that human hearing is nonlinear. *As are devices like headphones and loudspeakers.* That's one reason I chose a paper that used pink noise to mask intermod effects leading to a 'hearing' HF due to a non-human nonlinearity.

More generally, the loss of a sense of pitch at *audible* HF means we will tend to simply treat what can he heard as 'added salt' bringing up the 'taste' of a given pitch *defined by lower frequency components*. However that's an argument for liking HF distortion, not MQA being sounding like the 'Master' because of what it hides or fiddling with the ADC's dispersion.

FAQE again... :)

So no, I don't believe that you can have a 1/f-ish spectrum and still hear the stuff > 22 kHz when it *60dB below* the threashold people can measure in tests like the once I cited. I've never seen any measurement that even gives a plausible reason to accept it. At those levels, the ear will be essentially add its own sense of harmonics at a higher level I suspect because of any hearing nonlinearities.

If someone can come up with a *reliable* set of test results to the contrary I'd be interested. But I've read papers on this for decades and not found any that seem plausible. Too many errors and sources of alternative reasons for what was 'heard'.

Note that although some people can hear tones well above 20kHz, they tend to need to be at a relatively high level... which MQA would say isn't 'music'. They can't have this both ways.
Click to expand...
Yes I had noted the fundamental contradiction. However I was wondering if there was any well-founded reason (outside marketing) for validly asserting human perception of what actually can be carried by MQA and carried better by conventional high-rate audio. I have sometimes seen unproved candidate arguments (as per @Werner's comments above) and sometimes fundamentally wrong arguments (as per @davidsrsb's comments above). Nothing I have seen has yet convinced me of audibility below the levels of an extended audibility curve as per Ashihara. But the marketing continues.
 
You must log in or register to reply here.


advertisement


Back
Top