[gregorio]

It looked like just general argumentativeness to me, but I admit, I didn’t read it carefully.

No, it’s specific “argumentativeness”. In my case, an argument specifically against the audiophile myth that there are audible differences that cannot be measured/revealed with measurements. IMO, that’s a particularly good myth to bust because many other audiophile claims and myths rely on it.

So maybe you should read more carefully before responding or just don’t respond if you’re not interested.

G

 

[bigshot]

It went beyond that several posts back.

 

[Light - Man]

Whew! I thought I’d get all the way to the end with no sound.

Whew, that sounds like far too dangerous an experiment to me!

 

[jagwap]

I think this is the issue that audiophile marketers play on and therefore some/many audiophiles fall for: What do you/they mean by “conventional measurements”, the measurements used for published specs, the typical test routines of say an AP55, all the measurements of which an AP55 is capable or all the possible audio measurements? I can certainly buy the idea of audible differences with identical published specs, possibly even with the typical test routines of say an AP55 in pathological cases, but an audible difference that doesn’t show up in ANY measurement, including a Null Test, how’s that even possible?

I am not in marketing. I design the products for a living, and despise almost all marketing nonsense. You relay on this null test, which I have never seen work despite multiple attempts on analogue gear.

I’ve heard of rare situations where it was difficult to find the measurement type for a particular audible difference but not being able to measure an audible difference always ultimately comes down to the incompetence of the person attempting the measurement, albeit with mitigating circumstances on occasion.

Not it doesn't. You need to get involved in the design side. There are plenty of examples, and the people involved are not always incompetent. If they were you wouldn't spend so much money on their equipment.

An audible difference that cannot be measured is an audiophile myth. A myth that defies logic, because digital audio is itself a measurement. So if there were something that digital audio can’t measure, then it can’t be recorded or reproduced.

G

That last point is logical, but in the recording of the difference, if no difference is evident because it is swamped by the phase, amplitude and timing errors of the two devices, then it is moot. It is an unprovable proof.

Right, I should have stuck to that particular case and not generalize to any 2 DAPs. Even I have been troubled by phase or even overall time shifts between gears making inaudible differences look huge in a null test, so we end up with the reversed issue of measured difference we can’t easily judge in term of audibility. But would that be a problem here where you can just swap the casing on the same DAP, record a song twice and subtract them?

That would be interesting. The differences may be low enough to measure something. But none of us were there. Manufacturers rarely divulge all of their working, for obvious reasons.

Would not at least in the case of recorded sound everything audible be measurable since the act of recording is measuring sound in the recording space . So anything not measurable is infidelity

Same as Gregorio's point. But makes the same assumptions. That we can analyse ALL the differences on the recording.

Yes but as you say, that’s the opposite of the claim of audible differences that can’t be measured. BTW, have you tried DeltaWave? It seems very good at auto alignment when nulling two different recordings. Down to an accuracy of 1/1000th of a sample apparently.

...

As I understand it, TIM shouldn’t be an issue with any reasonably competent design and arises in response to a test signal that probably never exists in music recordings. In other words, it’s existence wasn’t known because it probably didn’t exist (in real world music/sound recording and reproduction)! It was discovered by accident about 50 years ago when someone cross wired an amp. I assume you are referring to the fact that TIM is not revealed in a typical THD measurement and before the 1970’s TIM was not specifically measured. However, although the TIM test/signal was not done before the 1970’s it was certainly “doable” and, a null test would reveal TIM (if present). This brings us back to: What is published as specs and what is typically measured as opposed to what can be measured. Obviously, if something isn’t typically measured, that does NOT mean it cannot be measured.

G

I have not heard of DeltaWave. I will look into it, as it looks interesting. How does it remove the phase differences and know they fully removed. What if the phase differences ARE the difference. Don't say we cannot hear them because i know we can. Group delay of 15uS or more between channels at high frequencies, and more than a few mS at low frequences when compared to midband is audible.

As to TIM test, it DID show up issues with designs. Amps were slow back then. Also, what null test was available back then. DeltaWave I suspect wasn't. It was what those teams I was describing was trying to do, and they failed. Some of them were smarter than average, before you insult them again.

BUT STILL we would have been able to reveal/measure it with a Null Test (which had been around for many decades before TIM was discovered)!

Where is this null test? I've over 30 years in the industry, and I have never seen it work, with the exception of the ones I mention, where they didn't really.

A null test will reveal any and all (“everything”) differences between 2 audio signals, regardless of whether we know what’s causing it or whether or not we have a way of measuring it in isolation.

Yes, but with the possible exception the DeltaWave you speak of, I have not seen a successful Null Test of analogue gear.

 

[Dogmatrix]

The question does not stretch to analysis it simply illuminates the logical connection between measured and measurable
All things measured , known or unknown are by definition measurable

 

[bigshot]

All things measured , known or unknown are by definition measurable

I can’t wait to see the argument against that one!

 

[sonitus mirus]

I can’t wait to see the argument against that one!

I'm hoping for a measured response.

 

[gregorio]

You relay on this null test, which I have never seen work despite multiple attempts on analogue gear.

Yes, I rely on it quite a bit, I probably do a couple a week on average but have days when I do 10 or more, which is similar to other audio engineers in my line of work. Usually I’m doing a null test on digital files but I have occasionally done it on analogue gear and it always works, although it is easier to screw up. I suppose it depends on what you mean by “works”. If you mean achieve a perfect null, you’re never get one with analogue gear because you’ll always have (at least) the slight difference random/thermal noise.

That last point is logical, but in the recording of the difference, if no difference is evident because it is swamped by the phase, amplitude and timing errors of the two devices, then it is moot. It is an unprovable proof.

If an audible difference is so swamped by phase, amplitude and timing errors that it’s not evident in the difference file, then it wouldn’t be audible. However, this is pretty much a hypothetical scenario because to swamp an audible difference would need relatively massive amounts of phase, amplitude or timing errors. Many times more than you’d find even in cheap gear.

How does it remove the phase differences and know they fully removed. What if the phase differences ARE the difference.

It doesn’t remove phase differences, it just aligns the two test files. So if for example there’s clock drift, that will still affect the difference file. However, it can be adjusted and disabled. The developer is very open to discussion and questions.

Same as Gregorio's point. But makes the same assumptions. That we can analyse ALL the differences on the recording.

Hang on, I’m not making that assumption! The contents of a digital audio file (including of course the contents of a digital null test difference file) can be analysed in terms of amplitude because a measurement of amplitude is of course what a digital audio file is, and it can be analysed in terms of frequency content with a Fourier transform but it cannot necessarily be analysed in terms of anything else, what is causing that difference for example. In other words, we can analyse the combination of ALL the differences but not necessarily analyse any of the individual differences that make up that combination.

As to TIM test, it DID show up issues with designs. Amps were slow back then. Also, what null test was available back then. DeltaWave I suspect wasn't.

Obviously deltawave wasn’t available back then. I’ve been in the business nearly 30 years and have virtually always done null tests using digital recordings. Although I have done it in the analogue domain a few times and seen older generation engineers do it by simply outputting the difference result to a main studio monitor and listening for anything audible or hooking it up to an oscilloscope. Also, using the null test principle was at one time a technique for feeding a cue mix to a recording artist, although I’ve never used it myself.

G

 

[Dogmatrix]

I can’t wait to see the argument against that one!

To be fair this kind of logic grenade does not mean much it is more a debating tactic
In our context a debate concerning differences between equipment used for recorded audio reproduction it means the term unmeasurable is an automatic fail

 

[Dogmatrix]

On reflection my previous post is bad , it opens the context unnecessarily and invites attack
I would have been safer and more correct to limit the context to the recorded data

 

[jagwap]

Yes, I rely on it quite a bit, I probably do a couple a week on average but have days when I do 10 or more, which is similar to other audio engineers in my line of work. Usually I’m doing a null test on digital files but I have occasionally done it on analogue gear and it always works, although it is easier to screw up. I suppose it depends on what you mean by “works”. If you mean achieve a perfect null, you’re never get one with analogue gear because you’ll always have (at least) the slight difference random/thermal noise.

If an audible difference is so swamped by phase, amplitude and timing errors that it’s not evident in the difference file, then it wouldn’t be audible. However, this is pretty much a hypothetical scenario because to swamp an audible difference would need relatively massive amounts of phase, amplitude or timing errors. Many times more than you’d find even in cheap gear.

I'm not saying it swamped the audio differences in that the audio differences were tiny. I'm saying it made the good analysis of the audio differences impossible.

Really good gear will have enough phase shift to cause large signals in the difference signal. You need DC <> light to get rid of it for sure. 1Hz <>100kHz is no where near enough. How do you differentiate from audible phase shift and inaudible? Like in crossovers? Only a limited amount of work has been done on that. Hawksford, Harman...

It doesn’t remove phase differences, it just aligns the two test files. So if for example there’s clock drift, that will still affect the difference file. However, it can be adjusted and disabled. The developer is very open to discussion and questions.

I found the thread on audiosciencereview.com and I will wade through it. A brilliant tool if it works. There seem to be some rather less effective applications before. It is non-trivial to do, as generations of audio design engineer has found. However if it doesn't align the phase differences, they could swamp and comparison of different analogue gear in the time domain. It will also add those of the ADC. Which are audible, which are artefacts?

Hang on, I’m not making that assumption! The contents of a digital audio file (including of course the contents of a digital null test difference file) can be analysed in terms of amplitude because a measurement of amplitude is of course what a digital audio file is, and it can be analysed in terms of frequency content with a Fourier transform but it cannot necessarily be analysed in terms of anything else, what is causing that difference for example. In other words, we can analyse the combination of ALL the differences but not necessarily analyse any of the individual differences that make up that combination.

Obviously deltawave wasn’t available back then. I’ve been in the business nearly 30 years and have virtually always done null tests using digital recordings. Although I have done it in the analogue domain a few times and seen older generation engineers do it by simply outputting the difference result to a main studio monitor and listening for anything audible or hooking it up to an oscilloscope. Also, using the null test principle was at one time a technique for feeding a cue mix to a recording artist, although I’ve never used it myself.

Interesting. Listening in analogue to the difference is a useful tool. I know people who listen to jitter error too. However you do know that is a subjective listening test, and NOT a measurement. So it does not support your argument well. As to measuring the difference signal, we are back to the problem I described.

G

So there may be a null measurement tool that started in 2019 I didn't know about. Before that these things eluded us in measurement form, and until I get a bit more info on Deltawave, we'll see if it is the blanket cure all to all audio measurement.

 

[71 dB]

What I am interested of is why audio engineers need to do null tests on weekly basis? I have not worked as audio engineer so I wouldn't know. What exactly are audio engineers comparing with null tests? Why is there B to be compared against A with null test? Why isn't there just A? If there are A and B, isn't it because there was a need to have different versions? Meaning it is KNOWN and DESIRED that A and B are different? For example different masters of a recording for different streaming platforms.

 

[gregorio]

What I am interested of is why audio engineers need to do null tests on weekly basis? I have not worked as audio engineer so I wouldn't know.

Many audio engineers will do a null test when comparing equipment or as part of a problem solving sequence. That’s relatively infrequent though, probably less (or a lot less) than once a month. The most common use is probably by Dialogue Editors. At the start of audio post production (for TV and Film) the Production Sound Mixer will supply the Dialogue Editor with a polywav for every “take”. These polywavs typically contain 5-16 mono files, each of which is either a different mic (EG. A boom mic, different actors’ lav mics or a plant mic), a mix track (a mix of one or all of the mics) or some sort of safety track (a straight duplicate or a duplicate with reduced gain). Therefore, some of these tracks sound obviously different, some are different but sound very similar and some are identical. A null test is often the quickest and surest way to separate the very similar from the identical, so we may do dozens of them a day at certain points in the editing process.

Another common use of a null test is to identify the phase difference between two recordings we know are from different mics/sources and this applies to music recording (as well as TV/Film sound). It’s quite common to have two different mono mics or sources for the same instrument. For example, it’s common to mic a snare drum with a mic on top (the batter head) and one underneath (the snare head), or to record an electric guitar both from the DI input and with a mic next to the cab. Particularly with the snare drum, there is always an audible phase difference between the two mics. So, we do a null test and slide one of the files in time until we find the quietest (greatest null) location and that’s the phase offset. We’ll usually just do this by ear but of course it’s easy to meter/measure the amplitude of the difference file to be sure.

G

 

[71 dB]

Many audio engineers will do a null test when comparing equipment or as part of a problem solving sequence. That’s relatively infrequent though, probably less (or a lot less) than once a month.

That makes sense.

The most common use is probably by Dialogue Editors. At the start of audio post production (for TV and Film) the Production Sound Mixer will supply the Dialogue Editor with a polywav for every “take”. These polywavs typically contain 5-16 mono files, each of which is either a different mic (EG. A boom mic, different actors’ lav mics or a plant mic), a mix track (a mix of one or all of the mics) or some sort of safety track (a straight duplicate or a duplicate with reduced gain). Therefore, some of these tracks sound obviously different, some are different but sound very similar and some are identical. A null test is often the quickest and surest way to separate the very similar from the identical, so we may do dozens of them a day at certain points in the editing process.

Thanks for explaining these things to us who don't know hardly anything!

Wouldn't it be easier to name the mono files according to what they are? For example Take_03_Boom_mic.wav and Take_03_Boom_mic_duplicate.wav? To me it looks messy to have 5-16 files without direct indication (filename) of what they are so that Dialogue Editors need to do null tests to figure out what is what.

Another common use of a null test is to identify the phase difference between two recordings we know are from different mics/sources and this applies to music recording (as well as TV/Film sound). It’s quite common to have two different mono mics or sources for the same instrument. For example, it’s common to mic a snare drum with a mic on top (the batter head) and one underneath (the snare head), or to record an electric guitar both from the DI input and with a mic next to the cab. Particularly with the snare drum, there is always an audible phase difference between the two mics. So, we do a null test and slide one of the files in time until we find the quietest (greatest null) location and that’s the phase offset. We’ll usually just do this by ear but of course it’s easy to meter/measure the amplitude of the difference file to be sure.

G

I actually know about this and I had to address this when I was in the mixing course. However, instead of null tests, what we did was process the signals of each mic until they could be mixed together getting the desired sound. In the simplest form you change the polarity of one of the mics and test which polarity gives louder sound. Of course you can slide the signals to be exactly aligned, but whether one needs to do that to achieve desired sound is another thing. Certain comb-filtering effects can be beneficial for sound with "character."

 

[old tech]

I am not in marketing. I design the products for a living, and despise almost all marketing nonsense. You relay on this null test, which I have never seen work despite multiple attempts on analogue gear.


Not it doesn't. You need to get involved in the design side. There are plenty of examples, and the people involved are not always incompetent. If they were you wouldn't spend so much money on their equipment.

That last point is logical, but in the recording of the difference, if no difference is evident because it is swamped by the phase, amplitude and timing errors of the two devices, then it is moot. It is an unprovable proof.

That would be interesting. The differences may be low enough to measure something. But none of us were there. Manufacturers rarely divulge all of their working, for obvious reasons.

Same as Gregorio's point. But makes the same assumptions. That we can analyse ALL the differences on the recording.

I have not heard of DeltaWave. I will look into it, as it looks interesting. How does it remove the phase differences and know they fully removed. What if the phase differences ARE the difference. Don't say we cannot hear them because i know we can. Group delay of 15uS or more between channels at high frequencies, and more than a few mS at low frequences when compared to midband is audible.

As to TIM test, it DID show up issues with designs. Amps were slow back then. Also, what null test was available back then. DeltaWave I suspect wasn't. It was what those teams I was describing was trying to do, and they failed. Some of them were smarter than average, before you insult them again.

Where is this null test? I've over 30 years in the industry, and I have never seen it work, with the exception of the ones I mention, where they didn't really.


Yes, but with the possible exception the DeltaWave you speak of, I have not seen a successful Null Test of analogue gear.

Apology if I am reading your post incorrectly, but null testing on analogue gear has been part of a designer's basic tool kit for generations.

As an example, null testing was crucial for the Carver challenge, where Carver used null testing to make different analogue amps sound identical - including between tube and solid state amps. It would have been impossible (or very improbable) for him to tune the different amps without utilising null testing.