ultrasonic harmonics

[fondone]

Simple question:

Do ultrasonic frequencies affect the way we hear 20hz -20khz????

also

do you consider this when you look at the specs of gear ????

thank you for your time.

[dcollins]

Originally Posted by fondone

Simple question:

Do ultrasonic frequencies affect the way we hear 20hz -20khz????

The answer is not so simple, but "ultra-sonic" means you can't hear it.

In order for >20kHz to become "sonic," some distortion mechanism must be present.

do you consider this when you look at the specs of gear ????

No.

DC

[jamsmith]

There has been some discussion over the years as to affect of harmonic content above the hearing range to what were hear. Some have claimed for instance that some of the nuances are lost in the digital recording of a Stradivarius violin at 44.1kHz because of the loss above 22KHz. This may indeed be hogwash or the result of poor DA conversion. However, to those who are absolutley convinced nothing above 20kHz can have an effect on the human mind might want to read this US Patent.

http://www.raven1.net/5159703.htm

[bkuijt]

Originally Posted by fondone

Simple question:

Do ultrasonic frequencies affect the way we hear 20hz -20khz????

also

do you consider this when you look at the specs of gear ????

thank you for your time.

In my experience the ultrasonic freqs are most important for two issues:

- Keep the signal clean and intact for as long as possible. So no bandwidth limiting until he very end of the chain if possible. Those ultrasonics are not audible if they were separate tones. But they make higher harmonics that contribute to tone, transients and phase information in your recording (if it hasn't been bandlimited earlier in the chain).

- With digital equipment high sample rates greatly improve the stereo image definition.
This has got to do with the interaural time differences we can hear. The smallest shift in stereo location we can distinguish is 3,5 microseconds if I recall correctly. At 44.1 the intersample time gap is 22 microseconds. Check out this article by James Moorer (founder of Sonic Solutions).
It corresponds with my own experience, instruments are better defined as a source and the perspective is more clear.
And again you don't want to throw anything away until the very end.

I couldn't comment on the benefit of ultra high bandwidth in tracking or mixing situations with mono sources.

So the short answer for me would be: yes and yes

Bests, Bastiaan

[UnderTow]

Originally Posted by bkuijt

In my experience the ultrasonic freqs are most important for two issues:

- Keep the signal clean and intact for as long as possible. So no bandwidth limiting until he very end of the chain if possible. Those ultrasonics are not audible if they were separate tones. But they make higher harmonics that contribute to tone, transients and phase information in your recording (if it hasn't been bandlimited earlier in the chain).

Either the harmonics have already interacted and are in the audible range (and thus will be captured at 44.1Khz) or they are not in which case they are inaudible and do not need to be recorded.

- With digital equipment high sample rates greatly improve the stereo image definition.
This has got to do with the interaural time differences we can hear. The smallest shift in stereo location we can distinguish is 3,5 microseconds if I recall correctly. At 44.1 the intersample time gap is 22 microseconds. Check out this article by James Moorer (founder of Sonic Solutions).
It corresponds with my own experience, instruments are better defined as a source and the perspective is more clear.
And again you don't want to throw anything away until the very end.

This whole article is based on these misconceptions: "If you put a pulse into one ear, then a pulse slightly delayed into the other ear, most people can hear a time delay of 15 microseconds or more. Under some circumstances,
some people can hear time delays of 15 microseconds. Note that one sample at 48 kHz is 20.833 microseconds. At 96 kHz, it is 10.4167 microseconds. The minimum inter-aural (across the 2 ears) time delay that most people can hear is less than one sample period at 48 kHz."

Inter channel phase accuracy at 44.1Khz is in the pico second range. WAY beyond what we can perceive. Also, moving your head half an inch will cause time delays between the ears larger than 15 microseconds. So unless listening to music through broken converters with your head in a vice is your idea of a good time, this article is hogwash.

So the short answer for me would be: yes and yes

I would say: no and no.

Alistair

[fondone]

so undertow,

in laymens terms why does 24 bit 96khz, 24bit 192khz and dsd all sound better than 24 bit 44khz?????????

thanks for taking the time to answer.

[djwaudio]

Originally Posted by fondone

in laymens terms why does 24 bit 96khz, 24bit 192khz and dsd all sound better than 24 bit 44khz?????????

Are we sure it all sounds better?

[UnderTow]

Originally Posted by fondone

so undertow,

in laymens terms why does 24 bit 96khz, 24bit 192khz and dsd all sound better than 24 bit 44khz?????????

Does a soundblaster at 96 Khz sound better than a Lavry/dCS/Mytek/etc at 44.1 Khz? If it doesn't, then why not? It isn't the sample rate so it must be other things. Those are probably the things worth focusing on, not ultrasonic harmonics.

Alistair

[peeder]

The obvious opportunity for audible effects of ultrasonic frequencies is intermodulation distortion. But I'm not sure IMD is desirable in any situation. So it may actually improve sound quality to filter them out. Even though you might sense that there is "more going on" in a signal that does have non-harmonic IMD. It might not be what you want going on though.

[jamsmith]

Originally Posted by UnderTow

Does a soundblaster at 96 Khz sound better than a Lavry/dCS/Mytek/etc at 44.1 Khz? If it doesn't, then why not? It isn't the sample rate so it must be other things. Those are probably the things worth focusing on, not ultrasonic harmonics.

Alistair

I would hope than when someone is making a specification comparision, it is within a certain grade of equipment. Psychoacoustics is a funny game and only with very carefully controlled experiment can the answers be found. And even then, people will argue. I stepped away from the studio professionally 10 years ago. I still see people arguing issues that I remember double blind tests with noted professionals "settled" back then.

[fondone]

QUOTE:
Does a soundblaster at 96 Khz sound better than a Lavry/dCS/Mytek/etc at 44.1 Khz? If it doesn't, then why not? It isn't the sample rate so it must be other things. Those are probably the things worth focusing on, not ultrasonic harmonics.
Alistair

that wasn't really my point Alistair.

I understand that components, clocks and anti aliasing filters all contribute...

My question is: is DSD , 24/96, 24/192 better in THEORY than 24/44khz???

In other words, assuming we have the same reference converter is 24bit/44khz mathematically all you need????

[Spartacus]

Originally Posted by UnderTow

Does a soundblaster at 96 Khz sound better than a Lavry/dCS/Mytek/etc at 44.1 Khz? If it doesn't, then why not? It isn't the sample rate so it must be other things. Those are probably the things worth focusing on, not ultrasonic harmonics.

Alistair

But Mytek AD for exemple, even for 44,1 khz, is oversampling x128...
so we're far more than 96 Khz..

or maybe you're talking about the DA stage? but a soundblaster also have a DA stage with decimation, even if they are saying they are working at 96Khz sometimes, there is always sometimes they need to go down to 44,1 Hz too..

so indeed we're comparing Mytek to Soundblaster, but at same SR

[bob katz]

Originally Posted by jamsmith

There has been some discussion over the years as to affect of harmonic content above the hearing range to what were hear. Some have claimed for instance that some of the nuances are lost in the digital recording of a Stradivarius violin at 44.1kHz because of the loss above 22KHz. This may indeed be hogwash or the result of poor DA conversion. However, to those who are absolutley convinced nothing above 20kHz can have an effect on the human mind might want to read this US Patent.

http://www.raven1.net/5159703.htm

Sorry, guys, this patent has nothing to do with whether inaudible harmonics of audio affect the perception of the in-band material! This patent strictly demonstrates the use of inaudible material as a carrier.

The mysteries of why high sample rate material sounds better than low have probably already been solved. I did spend a whole chapter of my book going through investigations and conclusions... but I won't spoil the suspense, except to quote our old friend Bob Olhsson:

"The issues of the audibility of bandwidth and the audibility of artifacts caused by limiting bandwidth must be treated separately. Blurring these issues can only lead to endless arguments."

So beware of experiments or tests that purport to prove the issue one way or another without separating the above issues! It's a lot harder than you think. That's what this chapter helps cover.

BK

BK

[fondone]

Mr. Katz ,
can you explain why DSD sounds better than the best pulse code modulation converters.????

maybe it doesn't... both sets of converters I listened to were "top of the line". I can only go by my subjective ears

but DSD had more depth.

[bob katz]

Originally Posted by fondone

Mr. Katz ,
can you explain why DSD sounds better than the best pulse code modulation converters.????

maybe it doesn't... both sets of converters I listened to were "top of the line". I can only go by my subjective ears

but DSD had more depth.

It's almost assuredly the filtering. Different filter designs do sound different!

Have you compared apples to apples? Here's an interesting listening test:

The identical D/A converter varying only the final output format. DCS DAC upsampling 44.1 kHz PCM to 192 kHz PCM versus to DSD. There are differences, still, and some unsolved mysteries, like why there seems to be more reproduced depth with the upsampling to DSD.

BK

[fondone]

Thanks for answering,

So it is still down to better mathematics in the anti aliasing.

geez, I thought with child labour and really good calculators we could have better 44khz filters in this day and age.?

I guess that means, theoretically a 24 bit 44.1 khz converter could still be made which has such a level of mathematics in its filtering that it could beat even the laziest of DSD filtering.....

I guess you get what you pay for.

[jamsmith]

Originally Posted by bob katz

Sorry, guys, this patent has nothing to do with whether inaudible harmonics of audio affect the perception of the in-band material!

Somehow I knew that was coming. That's why I hedged my bet and said "effect on the human mind" instead of "hearing"!

Originally Posted by bob katz

I did spend a whole chapter of my book going through investigations and conclusions...

And Amazon said that was supposed to be at my door today! (Maybe I shouldn't have been a cheap bastard and paid the extra 9 bucks to have instant access online.)

[Priscilla]

Originally Posted by bkuijt

In my experience the ultrasonic freqs are most important for two issues:

- Keep the signal clean and intact for as long as possible. So no bandwidth limiting until he very end of the chain if possible. Those ultrasonics are not audible if they were separate tones. But they make higher harmonics that contribute to tone, transients and phase information in your recording (if it hasn't been bandlimited earlier in the chain).

- With digital equipment high sample rates greatly improve the stereo image definition.
This has got to do with the interaural time differences we can hear. The smallest shift in stereo location we can distinguish is 3,5 microseconds if I recall correctly. At 44.1 the intersample time gap is 22 microseconds. Check out this article by James Moorer (founder of Sonic Solutions).
It corresponds with my own experience, instruments are better defined as a source and the perspective is more clear.
And again you don't want to throw anything away until the very end.

I couldn't comment on the benefit of ultra high bandwidth in tracking or mixing situations with mono sources.

So the short answer for me would be: yes and yes

Bests, Bastiaan

First, the time relationship between the signal and the sample pulse train does NOT alter the time delay through the conversion. If you delay the sample train by some amount of time, you will of course get different sample values, but the filtered DA signal will "fill the gaps" between the sample points to yield the original signal, and that is the case for ANY time relationship between the signal and the samples. The delay through the converters at any given sample rate is constant.

Second, I visited Moorer when he was still at Sonic, and we discussed that issue of "stereo time location". He said that he tested some people and found that they can hear 15usec. I said: how can it be? 15usec is 0.18 inch. One would have to keep thier head motionless to 0.18 inch. His answer was 2 fold:
1. The delay change in the experiment was SUDDEN 15usec jump
2. The listeners in the experiment had thier chin held fixed in a vice grip

Clearly, we do not change the delay suddenly by 15usec. And regarding the fixed head position, I do not mind a listening "sweet spot" but that is going a bit too far :-)

Regards
Dan Lavry

[Priscilla]

Originally Posted by fondone

QUOTE:
Does a soundblaster at 96 Khz sound better than a Lavry/dCS/Mytek/etc at 44.1 Khz? If it doesn't, then why not? It isn't the sample rate so it must be other things. Those are probably the things worth focusing on, not ultrasonic harmonics.
Alistair

that wasn't really my point Alistair.

I understand that components, clocks and anti aliasing filters all contribute...

My question is: is DSD , 24/96, 24/192 better in THEORY than 24/44khz???

In other words, assuming we have the same reference converter is 24bit/44khz mathematically all you need????

To my knowledge, DSD has not taken off commercialy. Sony decieded to stop supporting it a few years back. There are still a few that are trying to keep it going, but I see it as a losing battle, and I do not have intentions of joining those that want to "beat a dead horse".

DSD was an honest attempt at making good audio in the mid and late 90's. DSD offered an advantage of a single bit quantization (one comparator), which means (at least in theory) that the conversion has perfect differential linearity. In contrast, the multi bit PCM converters of the time suffered from imperfect differential linearity, which is a result of "uneven spacing" between quantization steps. The down sides for that DSD benefit was lowered code efficiency (about twice the data then standard CD), and a lot of very high frequency noise (from around 22KH to 1.4MHZ). That noise is inaudible, but it did cause problems, such as when trying to "meter" the audio levels.

DSD is really a one bit noise shaping. At the time DSD was introduced, the technology of noise shaping was sampling at 64fs, with a 5th order filter.
The dynamic range of such system was in the low 100dB's at best, but
I recall some designers thought they would be able to keep raising the filter order to much higher order. They were way too optimistic. At 5th order, you are very near the point of diminishing returns. Doing a 7th order filter does not buy you much. And obviously, going for faster upsampling would yield even less efficient format (much more data for the same music). But at that time, DSD was a serious competitor to PCM

DSD had another issue that made life difficult. Any processing, even a simple addition of channels, or change in volume, immediatly introduce a lot of bits into the process, so the data has to be re noise shaped into 1 bit, which lowers the dynamics and raise distortions. In addition, doing various processing in the DSD world takes huge amount of expensive hardware. In fact, much of the DSD processing in the real world is done by converting to PCM, doing the processing in PCM and then converting back to DSD. That is not a widely known fact.

Well, time moves on, and with it technological solutions. The PCM camp was able to solve the differential linearity issue in multi bit conversion. There are a few concepts that make it possible, the main one is called DEM (dynamic elements matching). As a result, the dynamic range of PCM has improved a lot, and with it the distortions are lowered. The newer multi bit PCM noise shapers far exceed DSD in performance, and all that without the down sides of high frequency noise. The newer multi bit noise shapers typically operate faster 128fs-1024fs is pretty common. That fast "localized sample rate" simplifies filter requirements and does not generate huge data.

In other words, DSD was a serious contender during a short time window when PCM had some limitations. It was not better then PCM on all fronts, but it did offer differential linearity advantage. Such is no longer the case.

Last comments: it is significantly easier to filter an oversampled PCM then it is to filter DSD. A highly oversampled PCM signal is pretty close to the desired signal. In both cases (PCM and DSD) the error signal is the high frequency content that needs to be removed by the anti imaging filter. For an oversampled PCM the energy is lower and further then the audible range. The higher you upsample, the lower the amplitude of the error signal is, and the further it is from the audible range, making an easier filter. But for DSD, the error signal is always very high amplitude (even with no audio signal at all - digital black), the signal is always going between two peak to peak values (an altering train of zeros and ones) and even worse is the fact that the error signal (frequency content) starts rising fast at around 22KHz. In addition, analog filtering at the DA output so close to 22KHz will also introduce phase non linearity. Oversampling PCM enables filtering with insignificant phase non linearity.

Regards
Dan Lavry

[Bob Olhsson]

DSD always assumed that all mixing and signal processing was going to be done analog so any arguments about DSP limitations are just an attempt to discredit the format. No, you can't mix DSD "in the box." So what?

Arguments about format are silly. The name of the game is implementation and not any technology's potential for perfection. Only when perfect implementations become common will there finally be something to discuss.

There were a lot of attempts to discredit DSD because all of the fundamental patents involved have expired and this threatened the patent income of many in the theoretical PCM crowd in addition to the whole DSP crowd.

[Dan Lavry]

Originally Posted by Bob Olhsson

DSD always assumed that all mixing and signal processing was going to be done analog so any arguments about DSP limitations are just an attempt to discredit the format. No, you can't mix DSD "in the box." So what?

Hi Bob,

This is completely incorrect. Sony has put a lot into DSD processing, and Sonic solutions did as well. They showed it in the AES convension year after year, and I saw it in both facilities (Sony New York and Sonics in San Rafael California).
Sony had some serious mixing, EQ, compressors... The amount of processing was huge (thus very costly), because of the sample rate (64fs). At Sonics, Moorer was having issues with having to reshape back to DSD. When you changed gain by say .1dB (or did any processing at all), you had to cut in an additional digital noise shaper, and you could hear that clearly. As I mentioned, there are few DSD processors out there that convert to PCM and back to DSD.

My response was completly technical, about facts, not emotions.

Regards
Dan Lavry

[fondone]

Excellent
I thank you both for your information.

so.... I basically want to know both (DAN & BOB) your subjective opinions on the best, most life like converters that money can buy???

I hope you can understand that when I heard DSD it was the best sound I have ever heard....so naturally I am a little curious about the holy grail of converters....

[Riccardo]

Originally Posted by fondone

Excellent
I thank you both for your information.

so.... I basically want to know both (DAN & BOB) your subjective opinions on the best, most life like converters that money can buy???

I hope you can understand that when I heard DSD it was the best sound I have ever heard....so naturally I am a little curious about the holy grail of converters....

If you do a search you'll find some recurring names. You'll probably also notice that Mr Lavry is the genius behind some among the best.

[peeder]

Originally Posted by fondone

Excellent
I thank you both for your information.

so.... I basically want to know both (DAN & BOB) your subjective opinions on the best, most life like converters that money can buy???

I hope you can understand that when I heard DSD it was the best sound I have ever heard....so naturally I am a little curious about the holy grail of converters....

I think Mr. Lavry will have an opinion on that!

Lavry Engineering - Unsurpassed Excellence

[jamsmith]

Originally Posted by peeder

I think Mr. Lavry will have an opinion on that!

Totally unfair to ask a businessman to give his unbiased opinion on his competitors product! Though I have always wanted to ask competitors/peers "Who, other than yourself, do you consider the best" Jimi Hendrix once was asked that. He answered "Terry Kath" (Chicago). And know he never said Phil Keaggy.

[fondone]

QUOTE:
" so.... I basically want to know both (DAN & BOB) your subjective opinions on the best, most life like converters that money can buy??? "


were all adults here. please don't be ashamed to vote for your own converter.

Please, Bob, Dan, or ANYONE for that matter, I am curious about what you think is the best, most life like converters that money can buy????

for me personally its the tascam dvra in dsd mode followed by prism dream at 96khz.... but i have yet to hear lavry gold & Weiss.

[Factory101]

I understand Sony have taken out patents on possible applications of ultrasound although to my knowledge these are more associated with communication/control devices similar to the patent mentioned in the OP.

I studied a bit of neuroscience at university - ultrasound may be a mechanism related to thought so it's inclusion in music/audio may be very important...perhaps adding to the seeming 'authenticity' of a sound. IMO it's not directly related to the musical content - more a general dynamic effect in the sense of the character of a piece of equipment of a certain era.

Perhaps the question should not be about the effect of ultrasonic frequencies on what we can hear, but rather the effect of audible frequencies on what we cannot hear.

I once camped on an island...there was constant sound from the wind and waves, etc. - on return to land there was an 'audible' silence...I can still feel it now by remembering it. Hope that's not too mystic...