Blind ABX test shows difference between 44.1 and 88.2

[Julf]

-there were a couple of all-digital workflows, notably Soundstream's. While it is easy to imagine that editing(*), and even the mixing of a limited number of channels, must have been feasible back then, it is not clear exactly how much processing could be done on these systems, how specialist effects were handled, and what sort of numerical accuracy was maintained. There is an intrigueing lack of documentation about this.

The ones that are sort of documented are the Synclavier "Tapeless Studio" and the Fairlight CMI - the first high-end digital workstations. Both were eventually upgraded to 16 bit / 50 kHz, but with very limited processing power.

It is good to remember that a lot of the iconic records from the 80's were made using 8 and 12 bit systems running at low sample rates...

 

[Werner]

The ones that are sort of documented are the Synclavier "Tapeless Studio" and the Fairlight CMI - the first high-end digital workstations.

I object. The CMI was not a recording system and was not a DAW. It could record bits of sound, it could sequence them, but it had to be used in a studio with a real multitrack environment. No-one in his right mind would make an entire record with just a CMI (*).

It is good to remember that a lot of the iconic records from the 80's were made using 8 and 12 bit systems running at low sample rates...

No. What you probably mean is that some of the sounds on many 80s pop and rock records originated in 8 bit and 12 bit samplers and synths. That is not the same as saying that the records themselves were made at these levels.

And anyway, what is the signal resolution when you send an 8-bit synth through a noise gate, then a plate reverb, then an analogue tape, and finally a 16 bit ADC?



(* Obviously someone will now point out one record made exactly like that. Artists being what they are this is only to be expected.)

 

[merlin]

Not exactly sure what you mean by that.

It was just trying to get an understanding because you are a main poster in threads that discuss the audibility or otherwise of different filters, sample rates and bit depths and I had no idea whether you had a system that was capable of allowing you to hear differences should they exist (I'm not saying they do!).

The truth is, you (like many contributors) simply don't. That's not to say you are wrong, just that it should be taken into consideration when looking at your comments. Please don't take that as trolling or somesuch. It's just a clarification.

 

[mattgbell]

It was just trying to get an understanding because you are a main poster in threads that discuss the audibility or otherwise of different filters, sample rates and bit depths and I had no idea whether you had a system that was capable of allowing you to hear differences should they exist (I'm not saying they do!).

The truth is, you (like many contributors) simply don't. That's not to say you are wrong, just that it should be taken into consideration when looking at your comments. Please don't take that as trolling or somesuch. It's just a clarification.

Do you know, broadly speaking, where the dividing line is between systems that can discriminate between different filters, sample rates etc and systems that can't?

And given that you profess not to know whether such differences (between different filters, sample rates etc) actually exist and a fortiori of what nature they are, how could you conceivably know where the dividing line was?

Not trolling, honestly ...

 

[Julf]

I object. The CMI was not a recording system and was not a DAW. It could record bits of sound, it could sequence them, but it had to be used in a studio with a real multitrack environment. No-one in his right mind would make an entire record with just a CMI (*).

Tell that to Kate Bush, Peter Gabriel, Jan Hammer, Trevor Horn, Art of Noise and the B-52's

But yes, it was the later Synclaviers that were more like modern DAWs.

And anyway, what is the signal resolution when you send an 8-bit synth through a noise gate, then a plate reverb, then an analogue tape, and finally a 16 bit ADC?

"Not very much"?

 

[Werner]

The truth is, you (like many contributors) simply don't.

So all the people that report day and night differences between samples rates do have such discerning systems?

Apparently the following statements all are true:

-sample rates above 44.1k don't matter
-they do, but any differences are very subtle
-they do, and the differences are very obvious.



I think it is time to coin a new logic ...

 

[merlin]

Matt. It's fairly simple.

If your system has an SNR of less that 100db how would you expect to hear the extra bit depth and if your speakers are -3db at 20khz how would you expect to hear any recorded ultrasonic information or noise?

 

[merlin]

So all the people that report day and night differences between samples rates do have such discerning systems?

Of course not Werner. There's rubbish spoken on both sides of all arguments.

But to have a chance of hearing any potential differences you would need equipment that could replay them.

 

[Werner]

Tell that to Kate Bush, Peter Gabriel, Jan Hammer, Trevor Horn, Art of Noise and the B-52's

Yes, they all had CMIs. Does that mean they recorded whole records on it? No. Such was impossible anyway. But if you are prepared to bring Kate over here I'll tell her anyway. Plenty of ueber-nice restaurants for doing so properly, in this corner where I sit.


When I was 17 and we had to give an English language lecture in class one of my subjects was the CMI MkII.

 

[Julf]

Yes, they all had CMIs. Does that mean they recorded whole records on it? No. Such was impossible anyway.

Tell that to EBN-QZN.

When I was 17 and we had to give an English language lecture in class one of my subjects was the CMI MkII.

The Series II was still mostly a synth/sampler, while the Series III was much closer to a proper workstation - but I have to admit I am more familiar with the Synclavier line.

 

[adamdea]

It was just trying to get an understanding because you are a main poster in threads that discuss the audibility or otherwise of different filters, sample rates and bit depths and I had no idea whether you had a system that was capable of allowing you to hear differences should they exist (I'm not saying they do!).

The truth is, you (like many contributors) simply don't. That's not to say you are wrong, just that it should be taken into consideration when looking at your comments. Please don't take that as trolling or somesuch. It's just a clarification.

I have no idea what you mean by that. Just so I understand is it the previous genelec/ps audio set up you objected to as not being good enough to hear differences or the current one using only an SBT as dac? If the Genelecs weren't good enough, then I guess they must be inadequate for everyone else who uses them.

Or did you feel that my previous dac wasn't adequate for some reason?

Perhaps you feel that the SBT is currently inadequate to hear more than 16 bits of resolution, but why not? After all the 17th bit halves the quantisation noise level, and if you can't hear that why should the subsequent changes make a noticeable difference. Incidentally the SBT seems to have a reasonable FR to take in the all important octave above 20Khz.

I certainly have not based what I have posted in the past on my experience of the audibility of bit depths, sample rates of filters on the system I am currently listening on. [I have not yet decided whether to borrow a dac before completing the Archimago 24 bit test.]

On the other hand, the fact that I am currently quite content to listen to the system I have is based at least partly on my previous experience.

 

[adamdea]

Of course not Werner. There's rubbish spoken on both sides of all arguments.

But to have a chance of hearing any potential differences you would need equipment that could replay them.

I'm not sure that that's as straightforward as you appear to think and thinking about it systematically just serves to make the idea of resolution over 16/44 seem more bizarre. Let's think about all of the equipment
What is the minimum spec of dac which you think is required?

Presumably the same requirement must be applied to the amps mutatis mutandis. Can we safely exclude anyone whose amps have a SNR of less than 120dB in the audioband? 110? 100?

What about the speakers and room. How loud do you think people would have to play the music? How can anyone hear the difference if they are playing at only 60 db over the noise floor of the room

Oh and what is the mnimum spec of ear? Common sense suggests that we should confine the tests to people who can hear up to 80Khz.

 

[merlin]

I have no idea what you mean by that. Just so I understand is it the previous genelec/ps audio set up you objected to as not being good enough to hear differences or the current one using only an SBT as dac? If the Genelecs weren't good enough, then I guess they must be inadequate for everyone else who uses them.

Both Adam. It's the speakers in both cases that roll off below 20khz - as to be fair most do.

I have far less concern over Dacs personally - even a £50 Dac is capable of performance that will easily eclipse that of all but the very best transducers and amplifiers.

For me the interesting question whether those who do have systems capable of hearing/sensing differences are able to reliably identify them. Sadly they are few and far between and sure don't seem to want to make much noise in threads like these. Probably too busy enjoying their music.

The last part of your second post suggests you are being somewhat facetious which is a shame. If you are going to try to uncover a difference you must at least have equipment that can replay that difference. I'd say that's common sense.

 

[adamdea]

Both Adam. It's the speakers in both cases that roll off below 20khz - as to be fair most do.

I have far less concern over Dacs personally - even a £50 Dac is capable of performance that will easily eclipse that of all but the very best transducers and amplifiers.

For me the interesting question whether those who do have systems capable of hearing/sensing differences are able to reliably identify them. Sadly they are few and far between and sure don't seem to want to make much noise in threads like these. Probably too busy enjoying their music.

The last part of your second post suggests you are being somewhat facetious which is a shame. If you are going to try to uncover a difference you must at least have equipment that can replay that difference. I'd say that's common sense.

Merlin I am being a teeny bit facetious but that does not alter the fact that there is a very very serious point.
1) you are being very selective in using technical requirements to bracket off what is and is not capable of reproducing hi rez. You appear to do so selectively in order to explain away things which do not confrim your beliefs. But in so doing, you are also arguing yourself into a position directyl contradictory to your stance on vinyl and hi rez.

2) yes it's true that dacs have better performance than transducers and amplifiers. But you are ducking the next point which immediately follows- if that is the case then what is the minimum threshold of performance required from them? if an amp doesn't have snr in the audiband below the level of quantisation noise then how can the level of quantisation noise matter?

3) if it matters to you that the speakers have to have a frequecy response which is (anecoically?)flat well above 20Khz, then what's the point using people who can't hear above 20Khz. I admit this bit is a teeny bit facetious but excluding results because the replay chain has to be able to detect particular frequencies carries with it an inexorable logic.

4) if it is impossible for equipment to reproduce the required level of resolution then we can also conclude that all reports of the benefit of hi resolution from people whose equipment does not meet these cirtiera are false positives.

5) When sacd first came out there was a fashion for speakers (and supertweeters) to be brought out in order to take advantage of the benefits of the new hi rez material. So mention was made that this speaker or that was producing output to 40Khz. This all died down. The fact is that there never was any connection between the people who claim to hear the benefits of hi rez and there being either program material over 20KHz or the means of reproducing that material

6) that left the hypothesis advanced furiously by fans of analog and fans of hi rez (like yourself) that its all really something to do with the linear phase brick wall filter on 16/44. Many speakers might not be able to reproduce tones over 20Khz but they don;t have a brick wall linear phase filter built in. And whatever limits they have would surely have the same effect on vinyl. So if DSD sounds like vinyl because its hi rez, and PCM doesn't, this would have to be audible in any system which can reproduce the vinyliness of vinyl.

 

[adamdea]

Matt. It's fairly simple.

If your system has an SNR of less that 100db how would you expect to hear the extra bit depth

If you follow this logic through then surely the only valid results can come from people listening in rooms with a noise floor at least 100 dB below the threshold of pain/ear damage (or lower if the listener is eccentric enough to listen at a quieter level than that.)?

 

[merlin]

1) you are being very selective in using technical requirements to bracket off what is and is not capable of reproducing hi rez. You appear to do so selectively in order to explain away things which do not confrim your beliefs. But in so doing, you are also arguing yourself into a position directyl contradictory to your stance on vinyl and hi rez.

How's that then?

 

[adamdea]

How's that then?

You'll find it in the rest of the post

 

[merlin]

Right I've had a coffee so I'll answer a few of these but no idea where you are trying to go

2) yes it's true that dacs have better performance than transducers and amplifiers. But you are ducking the next point which immediately follows- if that is the case then what is the minimum threshold of performance required from them? if an amp doesn't have snr in the audiband below the level of quantisation noise then how can the level of quantisation noise

it doesn't. If you have a valve amp I would not expect you to hear 24 bit resolution. I have a valve amp.

3) if it matters to you that the speakers have to have a frequecy response which is (anecoically?)flat well above 20Khz, then what's the point using people who can't hear above 20Khz. I admit this bit is a teeny bit facetious but excluding results because the replay chain has to be able to detect particular frequencies carries with it an inexorable logic.

I can't hear a test tone above 13khz. Most people my age can't. I can sense clearly if I remove frequencies above 15khz however. Is that the sonic effect of the filter? Why would that one be audible and the PCM one not be?

4) if it is impossible for equipment to reproduce the required level of resolution then we can also conclude that all reports of the benefit of hi resolution from people whose equipment does not meet these cirtiera are false positives.

Yes. Absolutely. Placebo and the power of suggestion. I'm not claiming one way or another remember, I am merely stating that if you want to see a car passing outside you will at least need a window.

5) When sacd first came out there was a fashion for speakers (and supertweeters) to be brought out in order to take advantage of the benefits of the new hi rez material. So mention was made that this speaker or that was producing output to 40Khz. This all died down. The fact is that there never was any connection between the people who claim to hear the benefits of hi rez and there being either program material over 20KHz or the means of reproducing that material

There you go stating "facts" again. FWIW most major transducer manufacturers have continued to drive FR and resonance upwards. BW, Focal, TAD, JBL, Revel, Scanspeak. All of these companies are at the cutting edge of drive unit design and continue to extend HF response. Look at the FR of many modern HF drive units. I've always wondered why bother if it's inaudible?

6) that left the hypothesis advanced furiously by fans of analog and fans of hi rez (like yourself) that its all really something to do with the linear phase brick wall filter on 16/44. Many speakers might not be able to reproduce tones over 20Khz but they don;t have a brick wall linear phase filter built in. And whatever limits they have would surely have the same effect on vinyl. So if DSD sounds like vinyl because its hi rez, and PCM doesn't, this would have to be audible in any system which can reproduce the vinyliness of vinyl.

DSD 64 doesn't really have a FR that even extends to 20khz so I think you are maybe barking up the wrong tree there.

Look at the FR for the 805's you are using

Do you not think that resonance is going to be as destructive to the signal (and audible) as any filter?

I'm not going to get into some silly circular argument with you here. I think you are getting confused - especially with what I have stated with regards to both vinyl and DSD. Maybe you are getting me confused with someone else. Whatever. If you think my position here IS contradictory then please be so kind as to post and example rather than just inferring it.

 

[davidsrsb]

If you have a tweeter with a resonance like that, and many are, the presence or absence of ultrasonic noise is likely to affect the tweeter behaviour below 20 kHz
Which is better is up to you

 

[adamdea]

Right I've had a coffee so I'll answer a few of these but no idea where you are trying to go

if you need DSD to capture the loveliness of vinyl, and people can hear the loveliness of vinyl on any old speakers then you should be able to hear the loveliness of DSD/unloveliness of pcm on any old speakers.

Perhaps I have misunderstood you, because I was under the impression that you had said that the special quality of vinyl could only be reproduced by dsd and not by PCM.

Is that the special quality of vinyl which ordinary people hear on their quad ESL or Tannoys or whatever or the special quality of vinyl which only people whose speakers go out to 100Khz can hear?

it doesn't. If you have a valve amp I would not expect you to hear 24 bit resolution. I have a valve amp.

righto

I can't hear a test tone above 13khz. Most people my age can't. I can sense clearly if I remove frequencies above 15khz however. Is that the sonic effect of the filter? Why would that one be audible and the PCM one not be?

the problem is that once you accept that you can't hear the material over 13Khz you are left with having to produce a theory about what is going on. If the theory is that PCM filters create audible artefacts even if filtering out inaudible sounds then that leaves the question "why are the artefacts audible?". If the PCM filter has created audible artefacts ie artefacts lying within the range of your hearing then the speakers would only have to reproduce the sounds within the range of your hearing in order to show the shortcomings of PCM.

It does not follow that you need flat frequency response well beyond your hearing range to hear the artefacts caused by the 16/44 anti alias filter.

Yes. Absolutely. Placebo and the power of suggestion. I'm not claiming one way or another remember, I am merely stating that if you want to see a car passing outside you will at least need a window.

Yes but the problem is that it is not clear what is being tested. If it is the hypothesis that people can hear program material over 20Khz then clearly you need to be able to reproduce those sounds. If you are testing the hypothesis that a steep linear phase anti alias filter with a transition band around 20-22Khz is audible to ordinary folk, maybe not.

If you are testing the hypothesis that it is the presence of that filter which means that the properties of vinyl which are audible to all cannot be captured at 16/44, again not really.

if you are testing the hypothesis that for some reason or other 24/96 recordings sound better than 16/44, again maybe not.

There you go stating "facts" again. FWIW most major transducer manufacturers have continued to drive FR and resonance upwards. BW, Focal, TAD, JBL, Revel, Scanspeak. All of these companies are at the cutting edge of drive unit design and continue to extend HF response. Look at the FR of many modern HF drive units. I've always wondered why bother if it's inaudible?

I can think of lots of reasons: why do dac manufacturers produce 24 bit dacs?

In any event what speakers, in your analysis, can test the difference between dsd and 24/192 or upwards. So to test the hypothesis that DXD processing affects DSD files you need to have speakers which are -3dB at what frequency? Are such speakers common? Do they exist? Should we therefore ignore anyone who claims to be able to hear this?

DSD 64 doesn't really have a FR that even extends to 20khz so I think you are maybe barking up the wrong tree there.

Really? DSD does not have a brick wall filter, although it has a relatively shallow filter starting somewhere or other. The presence of frequencies above 20Khz and the absence of the brickwall filter are two sides of the same coin (after all what is 16/44 but dsd plus a brick wall filter and some decimation).

So what FR do you require for your speakers
a) to hear the difference between 16/44 and dsd64
b) to hear the difference between 16/44 and 24/96
c) to hear the difference between 24/96 and dsd
d) between 24/192 and dsd?

Look at the FR for the 805's you are using

Do you not think that resonance is going to be as destructive to the signal (and audible) as any filter?

actually I would have expected that any speaker would have peaks and troughs of that level in room at your ears because of comb filtering. They will have peaks and troughs like that even below 20Khz.
Anyway I thought my speakers were supposed to be rolling off not peaking.

I'm not going to get into some silly circular argument with you here. I think you are getting confused - especially with what I have stated with regards to both vinyl and DSD. Maybe you are getting me confused with someone else. Whatever. If you think my position here IS contradictory then please be so kind as to post and example rather than just inferring it.

Its simple- if your position is that in order to hear the effect of an AA filter you need flat FR well beyond the transition band of that filter, then frankly 16/44 is bound to be adequate for most people (ie the effect of the PCM filter is unlikely to be a problem and the chances that anyone could hear the difference between 24/96 and dsd would be incredibly slight. Equally it would follow that 16/44 needledrops of vinyl would sound the same as PCM or dsd needledrops.