MQA pt II
- Thread starter Jim Audiomisc
- Start date
adamdea
You are not a sound quality evaluation device
Sorry if I am being slow here- but putting together your posts are you saying that you think that tidal MQA files have had a time smearing all pass filter applied to them. How does this compare with a minimum phase filter. I seem to remember that a minimum phase filter can;t be all pass(?) and here the delay in absolute time (not just phase angle) seems to be proportional to frequency.
Jim Audiomisc
pfm Member
Consider a transient sound source that produces a waveform that starts with a 'spike'. In spectral terms that contains a set of frequencies in a specific time/phase relationship.
if you 'disperse' those frequency components using an all-pass filter that adds a delay that varies with frequency you 'smear out' the energy and the result tends not now to be the original spike but a more extended shape at lower level. That may or may not 'sound different' but has a different peak/average level.
IF you can run that though a 'matched' filter it will apply a set of frequency-dependent time-shifts that restore the orginal shape. The question is, can you find such a 'correction' filter? I suspect in this case, yes. But as yet don't know.
Jim Audiomisc
pfm Member
The non-Tidal GO examples also seem to have this applied. What I don't know is who did this, why, etc.
Yes, in this case the time delay is frequency dependent. I had to stitch cycles to show the range correctly. If anyone is interested I can show the way the original behaves but it is a classic 'winding' behaviour.
As yet I don't know if it is fully correctable. Matter of test/experiment I guess.
If people want I can put up the raw output from the transform so they can see the I and Q values, etc, for each frequency. I'd welcome others testing this to pick up any weevils!
adamdea
You are not a sound quality evaluation device
yes please
DimitryZ
pfm Member
Nope. Some of us are curious and are eager to listen, others are not.
DimitryZ
pfm Member
The question remains, why you adamantly question 2L authenticity and are quite accepting of the same in the case of GO?
It is more than curious that the "anomalous" behavior is seen in GOs non-Tidal files. As you say "I don't know who did this and why?"
I have a pretty good guess....but I am a simple engineer who has seen fraud from vendors in test reports..." Though those usually aren't YouTube videos - a long accepted gold standard of scientific enquiry.
Jim Audiomisc
pfm Member
OK.
http://jcgl.orpheusweb.co.uk/temp/IQResults.zip
The zip contains:
a) The basic Trio of the GO 44k impulse patterns.
b) Output0 and Output1 which are the results of the transforms on the 0 (GO source file simple impulse) and 1 (One of the outputs said to be from Tidal) impulse patterns.
c) Impulsive. This is the c source code used to do the transforms. It's crap like all my programs but I just wanted to do a simple FT to see what I got.
The length of the FT is probably too long, so adds in some cruft from afterwards. But I thought that better than accidentially snipping a tail.
The data in the output is: frequency, I, Q, Amplitude, Phase (wrapped)
For the final plots I did I stitches the phases back to be continuous.
The zip was done on RISC OS, but should open OK on other OSs. (RO stores filetypes as an attribute so doesn't need an extension on the file name.)
There may well be mistakes, that's one reason I'm making these things available so they can be found and fixed by better programmers.
BTW Note that the GO source is 24bit but the Tidal files are 16bit. Hence a difference in scaling.
Jim Audiomisc
pfm Member
You keep making false assumptions despite my having already repeatedly redirected you to statements that should clear them up. Others don't seem to share your behaviour.
Your second comment says more about that than it does about me.
I've tried repeatedly to disabuse you of the false assumptions you keep making, so cannae be bothered to do so again. Simpler for you to go and re-read what I've already repeatedly pointed you back to, etc. Others don't seem a puzzled as you.
Beyond that, more info will appear when I finally conclude what I'm actually doing.
You're also welcome to look at the data I generate and laugh at my programming 'skills'. (sic)
DimitryZ
pfm Member
You redirect well.You keep making false assumptions despite my having already repeatedly redirected you to statements that should clear them up. Others don't seem to share your behaviour.
Your second comment says more about that than it does about me.
I've tried repeatedly to disabuse you of the false assumptions you keep making, so cannae be bothered to do so again. Simpler for you to go and re-read what I've already repeatedly pointed you back to, etc. Others don't seem a puzzled as you.
Beyond that, more info will appear when I finally conclude what I'm actually doing.
You're also welcome to look at the data I generate and laugh at my programming 'skills'. (sic)
No answers, other than "others don't share your concerns." I would strongly suggest that that is likely because they share your viewpoint re. MQA, which you have long held before the current phase of your investigation.
And if you accept your "inputs" as correct and provided to you in good faith by other researchers (in this case, "researcher"), when there is little convincing or veryfiable evidence that they are, your entire endeavor becomes highly questionable. This is 101.
I would never laugh at your programming skills, as they are certainly better than my own.
DimitryZ
pfm Member
And where is similar output for what GO claims to be the "Non Tidal Response" but has a curiously similar profile to the Tidal ones? You know, the control one that is supposed to demonstrate differences between TIDAL/MQA and LPCM?OK.
http://jcgl.orpheusweb.co.uk/temp/IQResults.zip
The zip contains:
a) The basic Trio of the GO 44k impulse patterns.
b) Output0 and Output1 which are the results of the transforms on the 0 (GO source file simple impulse) and 1 (One of the outputs said to be from Tidal) impulse patterns.
c) Impulsive. This is the c source code used to do the transforms. It's crap like all my programs but I just wanted to do a simple FT to see what I got.
The length of the FT is probably too long, so adds in some cruft from afterwards. But I thought that better than accidentially snipping a tail.
The data in the output is: frequency, I, Q, Amplitude, Phase (wrapped)
For the final plots I did I stitches the phases back to be continuous.
The zip was done on RISC OS, but should open OK on other OSs. (RO stores filetypes as an attribute so doesn't need an extension on the file name.)
There may well be mistakes, that's one reason I'm making these things available so they can be found and fixed by better programmers.
BTW Note that the GO source is 24bit but the Tidal files are 16bit. Hence a difference in scaling.
How do normals open your plots - what format are they in?
Nope, all of us are curious and eager to listen but you are choosing to be blind and deaf to the simple points that you keep dismissing and twisting, even genuine data, let alone ethics, interesting trait?
DimitryZ
pfm Member
There has been precious little genuine data. And for ethics, I am generally against boycotting products because they may have used deceptive advertising. You can leave your virtue signaling at the door. Perhaps you can start a new thread "Companies that use deceptive advertising and I whose products I am boycotting."
And as I shown before, audiophiles often prefer products even if genuine data is available to show that their choice is technically inferior.
Do you have anything to contribute - interesting technical analysis or valuable listening experience base?
I have shared my experience which can be easier seen, read and unlike yours, is consistent. Oh, and I will not leave my ethics at the door, thanks for the mockery but trumpism is not for me and this is not a shampoo advert. Some of us are not going to eat everything and ask questions, sorry if that keeps escaping you - again may be the same trait?
How about that poll I suggested - are you still collecting courage?
still not? come on
mansr
Objectionist
Correction: Enjoy them while you still can.
DimitryZ
pfm Member
Mansr:"Sky is falling!"
Repeat constantly for 5 years, despite being consistently wrong.
To prove this to yourself, look up. Consistently expanding number of non-MQA hires streaming options as a function of time, with Apple as the last entrant.