For the digital portion yes, the digital filters are simply mathematical computations, if you multiply two numbers together you get their product exactly not a different answer depending on their size. I don't think there are chip sets out there anymore with low res integer math which of course might cause an issue.
From the line in the manual I assume you simply say apply eq and give it a file path or something like that. It seems to be part of the setup process.
Reading some very old posts in GS there were some results with only -10dB nulls clearly something is not right.
Well I actually meant the full DA system which includes filters. The physical attempt at realisation of the maths is never perfect? If it was we wouldn't have linearity errors at any amplitude level.
But this is my point - there's no where in Diffmaker that I can find where you can "apply eq file" - it simply isn't possible, unless I'm missing something
Indeed, Diffmaker can produce very non-intuitive results that don't lend themselves to easy analysis of what went wrong
Has anybody found what settings the A/D test results use for their Diffmaker settings?
I didn't know that and I haven't looked at diffmaker since the 1.0 release. It makes no sense to make an eq file unless you are going to use it. Is it possible if you create one (it finds one) it is always used and that is not clear from the instructions? From diffmakers side all computations are probably 64 bit floats and there will be no errors from any of them.
I hope I'm wrong & eq file can be used in Diffmaker but I don't see how!
From the Diffmaker website:
"DiffMaker ... is only trying to help detect whether anything has changed .... It doesn't matter if the difference that DiffMaker finds might not be perfectly reproduced -- only that the difference is left intact enough to hear."
"The sound card used doesn't need to be completely transparent or of highest pedigree. It only needs to be capable of responding to any differences that may occur (even if those differences aren't reproduced perfectly) and of not burying any significant differences in added noise."
Obviously, diffmaker is not intended to evaluate a sound card, or to compare a source file to a looped file in order to try to evaluate a sound card. It's a complete misuse of the software. No wonder the results don't make sense.
Besides being used to make comparisons of changes in a data conversion loop, diffmaker can also be used to compare two files. But, it the latter case it's up to the user to determine that the comparison makes sense and will produces useful results.
"DiffMaker ... is only trying to help detect whether anything has changed .... It doesn't matter if the difference that DiffMaker finds might not be perfectly reproduced -- only that the difference is left intact enough to hear."
"The sound card used doesn't need to be completely transparent or of highest pedigree. It only needs to be capable of responding to any differences that may occur (even if those differences aren't reproduced perfectly) and of not burying any significant differences in added noise."
Obviously, diffmaker is not intended to evaluate a sound card, or to compare a source file to a looped file in order to try to evaluate a sound card. It's a complete misuse of the software. No wonder the results don't make sense.
Besides being used to make comparisons of changes in a data conversion loop, diffmaker can also be used to compare two files. But, it the latter case it's up to the user to determine that the comparison makes sense and will produces useful results.
Last edited:
Could it be that the equalization is automatically done if and when an eq-file is provided?
I think the developer of the Audiodiffmaker is a member of this forum too, why not ask him about it?
No, what I'm saying is the threshold for human cognition for *any* type of distortion is so far above the capability of a good amp.
I'm talking purely about the electrical reproduction. You want what is coming off the wire and into the speaker in your house to be as close to what was coming off the wire and into the speaker in the producers studio (neglecting the speaker for a moment)
After that, there's a host of other issues related to learning about what sounds right. But that's why we value good producers, right? They "teach" us what sounds correct.
But as a first step, all is hopeless if your raw electrical signal isn't very close to the raw electrical signal in the studio.
Distortion of any kind. My assertion is that whatever distortion you can readily identify, a good amp delivers far below that level.
If you are looking for a quantity that cannot be measured or understood, then you might be right. But if something can be measured, ABX is the surest way to identify if you can detect it.
Do you think it's fatal that that we should aim that the signal on the wires most closely matches the signal on the wires in the studio?
There is a button to create an EQ file, and buttons for each file to apply EQ from an acquired EQ file.
See image below.
Attachments
It's well established that simple gain "distortion" will cause people to favor the louder device. It's well established that some peaking in the low end will cause people to favor that device.
I know that a hint of distortion on guitar will make it more pleasing.
The world is full of examples where people tend to favor higher distortions, whether harmonic or amplitude.
Yes, if you cannot detect an added distortion from a zero crossing issue that measures at -40 dB, then you dont' really need to worry if an amp has zero crossing distortions that are at -80 or -100, right?
But that just shows there are electrical differences, which we already know. The Gearslutz tests is a poor-man's analysis if they don't own an audio analyzer, isn't it? Stick these same combos on a $10,000 audio analyzer and you'll get an even better understanding of the differences. Right?
Gearslutz has nothing to do with what you can readily hear, does it?
That is what I'd like to get to: What % of the DIYAudio crowd can hear a modest amount of crossover distortion in an ABX test? Or gain distortion, etc? And once you know that, then arguing about a distortion that is 20 dB below the established threshold is academic at best.
Last edited:
This is a very important point. Anyone that has ever looked at room response understands that moving the mic (or your head) 1 inch will result in dramatic changes to room response, far greater than that which a junk amplifier could ever hope to cause.
How do you know that? There is not research I am aware of showing that no human can hear very low level distortion.
If we substitute "mastering session" for "studio," that would be the goal of the artist and record company.
Wrong. Good producers know how to push our buttons and pull our levers, inputs that we largely don't even know exist.
Disagree. I have had the opportunity to compare good amps and they don't sound identical to me. In theory though, sure, ideally an amp should be "wire with gain" transparent.
Actually, there is evidence to the contrary, you just ignore it when it someone posts about it. How about this: http://www.diyaudio.com/forums/digi...ble-difference-whatsoever-90.html#post5265102
Guitar distortion is often quite extreme, but guitarists are often very particular about exactly what the distortion sounds like.
In principle, yes. However, some people can hear much lower than -40dB distortion depending on the exact distortion. THD as a measurement correlates very poorly with human perception, which as been known for many decades. Primarily, it is still used because it is easy to measure, and manufacturers feel consumers expect to see it. Its not clear what distortion metric you are thinking of.
What established threshold? According to Earl Geddes who published some of the research on what people can hear, published research used fairly small sample sizes and probably is about right for 95% of the population. Having done the research, maybe he is more optimistic about how accurate it is than I am. Even if he is spot on, that still leaves 5% of the population that would be expected to fall outside the "limit." That's not such a small number either. Out of all the people here, what percentage do you suppose are in the top 5% of IQ tests? Likely many more than 5% of DIYaudio users, I would say based on the complexities of engineering. Maybe some of them are also well up into the top 5% of listeners in terms of distortion perception. How would you know one way or the other? One can't know exactly because it hasn't been measured. What would you estimate?
Last edited:
@QAMATT - I would love to see where we can find all possible distortions reported for any device?
In relation to Geddes, he established a distortion measurement which he stated was better correlated to subjective audibility & just recently made this statement about it on the speaker thread he is posting on -
And to follow up on this he states that this test he designed he claimed produced far better correlation to subjective listening results than the well known disconnect that exists between THD + N & subjective listening
In relation to Geddes, he established a distortion measurement which he stated was better correlated to subjective audibility & just recently made this statement about it on the speaker thread he is posting on -
And to follow up on this he states that this test he designed he claimed produced far better correlation to subjective listening results than the well known disconnect that exists between THD + N & subjective listening
Last edited:
I think the main misunderstanding is to ignore, that we listen and enjoy music in an emotional state of being. Music consumption is a nonverbal and nonintellectual activity, activating areas in our brain, that are not associated with analyzing and verbalizing. So when we are forced to listen analytically we miss a lot of information. As in an ABX Test.
The quality of music and especially the soundquality is best judged through a longer evaluation period. How does it feel? Can I emotionally connect to it?
This way you can discriminate between very similar sounding devices (like converters) and in my experience it correlates to measured differences that are under the perception threshold. Like often seen in jitter measurements.
As addendum to my last post; Paul Frindle reported that he and his team detected those differences in ABX tests with music samples not test tones or constructed special waveforms.
Paul Frindle, Are We Measuring the Right Things? Artefact Audibility Versus Measurement, UK 12th Conference: The Measure of Audio (MOA) (April 1997), paper no. MOA-05, 45 .
Paul Frindle, Are We Measuring the Right Things? Artefact Audibility Versus Measurement, UK 12th Conference: The Measure of Audio (MOA) (April 1997), paper no. MOA-05, 45 .
What level do you you think could be readily detected by someone skilled?
Not really. The goal of the mastering engineer is to improve the suitability of the final product on a range of playback mediums. Their primary tools are eq and compression.
Agree. But without this training, kids tend to prefer a scooped response: boosted bass and boosted treble.
What level and type of distortion do you think you could readily discern without much trouble?
Not sure what you are asserting here? I believe ABX can detect differences, and I suspect most would surprised to learn what they could and could not hear.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.