Why do you need full set of measurements that fully characterizes a DAC? Can you or anyone hear full character of DAC?
If one sounds different from another due to volume level discrepancies, then it would easily show up in measurements. If due to much higher distortion, then that will show up in measurements as well.
Basic "in" vs "out" signal comparisons, be it sine wave or square wave just to name couple. Take a look at Stereophile magazine review section that shows measurements.
You keep misunderstanding basic English and now complaining for getting called out. If you want to debate properly, get the proper tools checked first or you will be ignored.
More like other reviewers., no one technician can do measurement of feelings and impressions. It is job for other specialists..
You've never participated in one, have you.
@Evenharmonic - I can see you are not up to the task of answering questions, providing any worthwhile technical detail or debating in a sensible manner& even learning anything new is beyond you - I'll not waste my time any more (some probably wonder why I even bothered answering you the first time?) - hopefully some readers got something out of the exchanges, you certainly didn't
Last edited:
You do realize that the frequency spectrum at the eardrum is different for everybody due to different pinna & ear canals shapes causing different resonances?
You would do well to read the research of David Greisinger on binaural reproduction?
Have a look at "Accurate reproduction of non-individual binaural recordings without head tracking through individual headphone equalization"
Look at slides 11 onwards to see the how different are the frequency spectrum measurements taken at the eardrums of different 'normal hearing' participants!
Yes, we all have a different frequency spectrum at the eardrums & our analytic engine (auditory processing) equalizes this difference so that we coalesce to a similar relationship with the auditory world.
You originally posted that "there are some pretty good dummy heads out there that capture a reasonable binaural effect" & planet10 replied "Capture, but do not analyze. And the analyzer has ear lobes it has been trained to use. The "ear lobes” on a dummy haed will always be a compromize.". His statement is correct & Greisinger's research shows this is true for individual pinna & ear canal resonances.
So now you are trying a different tack - 'look, Greisinger is using measurements' - who ever said the frequency spectrum at the tympanic membrane couldn't be measured?
Last edited:
@Scott wurcer, I mentioned your use of multitone test signals on DAC modules & if I remember correctly, your statement that it sorted out the wheat from the chaff.
Care to explain some more how a multitone test signal differs from single tone or two tone test signals in regards to the extra information that is gleaned from this sort of measurement?
Care to explain some more how a multitone test signal differs from single tone or two tone test signals in regards to the extra information that is gleaned from this sort of measurement?
You can create multitone signals with statistics that are closer to those of music in terms of using the entire audio spectrum and having a high crest factor. In fact they often sound like a huge 6-handed organ chord. Using synchronous FFT's you can observe all the distortion components in the "noise" floor.
You can vary the crest factor and even create the so called asymmetric waveforms. I have some posted on the Linear Audio site as .wavs. You might want a basic understanding of Fourier math with respect to sampling rates and FFT lengths.
As for Dave's paper it says right in the title NON-individual binural recordings (also look at slide 8). As see this as an engineer using their expertise to solve a problem, making his dummy head recordings sound "right" to a wider audience. The answer was a not too difficult equalization for each user. How many times have folks here "hit the roof" at the very suggestion that two reproduction chains might sound (more) the same with simple equalization?
The Smyth Realizer uses a similar technique to a different end and the "effect" is dramatic.
So in conclusion I would think a 1/3 octave equalization as Dave did would be a requirement for any listening test.
Last edited:
Gleaning wheat
Sorry, it was actually the sheep from the goats, not the wheat from the chaff
"The multitone really separates the sheep from the goats. I'm using 30 1/3 octave tones at about 12db crest factor. Artifacts show up on even the best boards."
Thanks!
Right, between the tone peaks can be seen the non-linear distortion components which is more revealing of how a DUT behaves when processing music. In other words we can get a better view into the DUT's behavior when playing music than if we relied on single tone or two tone tests
So, let's investigate this further - the distortion products seen in the 'noise' are as a result of the additive IMD from each tone. In fact the distortion products are spread across the spectrum like a broadband noise floor. This test is using a stationary set of multitones, no changes in frequency or amplitude during the test which is like a snapshot taken of a precise moment in a music signal.
But a music signal changes moment to moment so the next snapshot will be a different set of multitones with different set of distortion components in other words this looks like a dynamically changing noise floor which is loosely correlated to the signal being processed i.e. it modulates both in spectral shape & amplitude. So rather than it being perceived as a background fixed noise ground, it begins to be perceived as another foreground signal with temporal fine structure (spectral shape) & envelope waveform (amplitude modulation)
This is my analysis of what is the cause of planet10's DDR which he defines as:
"Downward Dynamic Range. More crudely a lower noise floor and the ability to reproduce very low signals in the presence of larger ones. The characteristics that put flesh on a voice, tell you that it is this kind of violin instead of that kind, or how the guy is picking the guitar, and provides crtitcal information to improve imaging/soundstage"
So again, I go back to the basic principle that rather than extrapolate from a simple test tone measurement, the closer we get to using a test signal that better matches music, the better we are able to predict what might be happening with the DUT when handling a real music signal.
Right & have you observed the changes in the inter peak distortion products as a result of varying crest factor/number of tones?
Yes, individual equalization can be performed to adjust the dummy head binaural recording & adjust its spectral shape to better match the individual
Last edited:
In general once speakers are included in the tests the results are a mess.
Well the same could be said about BV's? or Evenharmonic's? high-fidelity criteria - the fidelity of input to output signal (signal from the speakers)!!
Last edited:
Instead of audiphile reviewer terms like detail, inner detail, resolution and a host of other non-technical words?
It is a term to describe a particular part of accurate reproduction.
For more info someone who worked closely with Allen (sadly went far to young), i was only his IT guy. I learned a huge amount from him.
dave
Unfortunately it turns out not to be a simple task… noone has yet cracked it.
Acoustics is mentioned, but that is not the only thing that can produce small, subtle details.
DDR is about how much information is lost as a signal passes thru a DUT. The less information lost, the greater the DDR.
dave
While sometimes fooled, sometimes easily, this is true, and implicit in the statement of Floyd’s i have posted.
For instance it has been shown that we can perceive signals well into the noise. No machine i know of can do that.
dave
I would say that rather than signal lost, it may be more a case of the non-linear transfer function of the various components in the repay chain causing added non-linear distortions which introduces signal ambiguity as far as our auditory perception is concerned. So low level signal is perceptually being buried under this added distortion as the distortion has spectral & amplitude modulation characteristics that mimic actual music signals.
It seems Kippel has no problems in using multitone testing on speakers
https://www.klippel.de/fileadmin/kl...lication_Notes/AN_16_Multitone_Distortion.pdf
And Neumann similary
Intermodulation Distortion
I’ve read this several time and it doesn’t compile. The forum rules require to debate on topic; while an analysis on your statement could be funny and perhaps interesting, I’ll cut it short since so far any previous attempt lead only to another load of bull chips. Short version: ill defined concepts, misunderstandings of basic signal processing principles, zero knowledge and understanding of the Fourier transform. Even shorter: a laughing stock for any EE undergraduate student.
Make yourself a favour and do disclose if you have any vesting interest in promoting this pile of; everybody would understand you have to make a living.
- Status
- Not open for further replies.