I hope I put this in the right forum. You've probably been asked this a million times, but I trawled the web (mostly searching "dan lavry" + "jitter", as I figured he was the "master") as well as his forum, and I couldn't really find my answer, so I asked on his forum (which gave this response: http://www.lavryengineering.com/lavry_foru...f=1&t=7449) (http://www.lavryengineering.com/lavry_forum/viewtopic.php?f=1&t=7449)), and I'll also ask here to hear your opinion:
If we're talking home audio listening (not recording), is jitter then audible, and how does it sound? I'm only talking about playing CDs and digital audio files from either a CD player, or your computer or a streamer through a dac into your amplifier.
Let's also for the time being assume that the analogue to digital conversion during recording/mastering was done well to begin with, and let's assume that the converter you use is of fairly good quality.
I discussed this with an audiophile friend. My own impression is that with 16 bit audio, jitter starts to become audible at -96 dB (without dithering), so with music recorded at sensible levels, it will not be audible, as all jitter is in the noise floor, and the music would mask the jitter. An exception could of course be orchestral music with a huge dynamic range recorded at a very low volume (so the noise floor would be raised). As for how jitter sounds, the video "Digital show & tell" on Xiph.org says quantization noise sounds like tape hiss from analogue tapes, but I might be mixing up quantization noise with jitter (is it the same?). On that website, there's also a file available for download with a 1 kHz tone at -105 dB. When playing that file there's background noise, which I assumed was jitter. In other words, jitter is present in all digital audio, but the amount is so low that you can never hear it except for in those extreme cases mentioned above. I asked Ethan Winer who said:
"Jitter manifests as noise 100+ dB below the music, and is never audible. Nor does it create 'a lack of depth, solidity and a smearing of the stereo image.' You’re thinking of wow and flutter. :->)"
In Ethan Winer's Youtube video "AES Damn lies workshop", he shows the following picture:
(http://s3.postimg.org/rqukd9xqb/AES_Damn_Lies_Jitter.png)
My friend's attitude is that jitter is omnipresent and always audible and smears the stereo image etc. (like the comment above), so although he exclusively listens to digital audio he is starting to think vinyl might be the way to go to get rid of the issue of jitter. Surface noise, pops, clicks, etc. from vinyl can be filtered out by our brains, whereas jitter is an omnipresent 'grating' and unpleasant sound.
I of course understand that when creating converters like Dan Lavry does, it's important to minimise jitter as much as possible in order to come closer to creating the perfect product, as well as creating an A/D converter that will have minimal jitter so the recording artists can raise and lower levels on different tracks as much as they like. But as mentioned, I'm only interested in audibility and listening at home – not in the technical aspect (measuring, etc.).
Jitter doesn't have anything to do with stereo. Its basically just instability of the sampling clock, which causes sidebands around pure tones. Usually its not audible though, since you need to have quite a lot of it before you'll even get the signal above the noise floor of modern music.
It tends to be a much bigger problem in things like RF, since the frequencies are much higher.
My friend's attitude is that jitter is omnipresent and always audible and smears the stereo image etc. (like the comment above), so although he exclusively listens to digital audio he is starting to think vinyl might be the way to go to get rid of the issue of jitter. Surface noise, pops, clicks, etc. from vinyl can be filtered out by our brains, whereas jitter is an omnipresent 'grating' and unpleasant sound.
Your friendly is sadly very misinformed. Jitter in digital audio does manifest itself as noise in the signal. But like you said, it's way below audible levels, unless the recording is done at a ridiculously low level. It can't do anything else, and certainly not "smear" anything at all. To claim that it is omnipresent and "grating" just shows that your friend has bought into the audiophile bullshit machine, big-time.
If your friend is so worried about jitter, he should be absolutely mortified at the amount of wow on vinyl records, even good ones. There can be a significant pitch difference as the record spins if the center hole isn't perfectly aligned. You mostly notice it on long sustained tones like organ music, you get a warbling or pitch rising and faling with the ~1.8 second period of the record's rotation.
Wow on a vinyl record is several orders of magnitudes beyond even very bad digital jitter, and the noise floor is significantly higher.
And that's before you get into other effects such as pre-echo (you can hear a faint copy of loud transients ~1.8 seconds before they hit, because the grooves slightly distort each other during the vinyl production process, or because the master tape has suffered from bleed-through. And of course, everything gets worse the closer you get to the center of the record, as there is less vinyl in each groove per revolution, so the dynamic range and noise floor suffers.
My friend's attitude is that jitter is omnipresent and always audible and smears the stereo image etc. (like the comment above), so although he exclusively listens to digital audio he is starting to think vinyl might be the way to go to get rid of the issue of jitter. Surface noise, pops, clicks, etc. from vinyl can be filtered out by our brains
I think your friend thinks that vinyl is better
because he can hear Surface noise, pops, clicks, etc, then he immediately knows it is vinyl and sound quality is automatically improved in his brain.
When playing that file there's background noise, which I assumed was jitter.
Not a good assumption in any experiment with 20 nuisance variables. Better account for/eliminate them all before ascribing any cause>effect.
Surface noise, pops, clicks, etc. from vinyl can be filtered out by our brains.
Audiophile brains, yes. Music lover no. I attend too much live classical and acoustic music to filter those completely alien artifacts. My references for sound isn't Foghat or other audiophiles stereos.
whereas jitter is an omnipresent 'grating' and unpleasant sound.
That is better remedied by a psychiatrist than a turntable.
cheers,
AJ
If we're talking home audio listening (not recording), is jitter then audible, and how does it sound?
Jitter is FM distortion which has two properties:
(1) The frequency of the modulation, measured in Hz of the modulation frequency. For example due to the 33 1/3 rotation speed of LPs, it is very common of them to have large amounts of jitter at 0.5555 Hz. For another example, due to some geometric effects, it is also very common to also have large amounts of jitter at the fundamental resonant frequency of the tone arm which is usually ab out 10 Hz.
(2) The amount of the modulation which is given in in the deviation in HZ or amount of frequency change due to the jitter.
Jitter was a far greater and more intractable problem with analog recording such as the LP and analog tape. It is also a problem with FM radio reception that is affected by multipath.
I'm only talking about playing CDs and digital audio files from either a CD player, or your computer or a streamer through a dac into your amplifier.
In general this jitter involves modulation frequencies and deviations that are inaudible.
Let's also for the time being assume that the analogue to digital conversion during recording/mastering was done well to begin with, and let's assume that the converter you use is of fairly good quality.
Then there is no audible jitter.
I discussed this with an audiophile friend. My own impression is that with 16 bit audio, jitter starts to become audible at -96 dB (without dithering), so with music recorded at sensible levels, it will not be audible, as all jitter is in the noise floor, and the music would mask the jitter.
That's not true because the jitter is not fully defined. The audibility of jitter is dependent on both the frequency of the modulation which in your example is unspecified, and its deviation which could be discerned mathematical means if the modulation frequency were known.
An exception could of course be orchestral music with a huge dynamic range recorded at a very low volume (so the noise floor would be raised).
That's false because jitter is usually independent of dynamic range. An audio signal can be jittered a certain amount regardless of its level, whether loud or soft.
As for how jitter sounds, the video "Digital show & tell" on Xiph.org says quantization noise sounds like tape hiss from analogue tapes, but I might be mixing up quantization noise with jitter (is it the same?).
Yes, quantization noise and jitter are two very different things and should not be confused with each other.
If you want to hear jitter, the easiest way may be to listen to a LP record or analog tape, particularly a cassette tape. Almost any playback of them will include audible jitter.
Some downloadable files with various amounts of jitter added and a reference file:
Link to post with jitter file attachements (http://www.hydrogenaud.io/forums/index.php?showtopic=107570&view=findpost&p=905631)
Your friend might be more inclined to notice "grating" flaws in recordings (and there are many) when he's listening to the digital version, because analog = more better. The irony is of course that the best analog can't match the noise floor of mediocre digital recording. Another explanation could be dynamic range compression and deliberate clipping, which wasn't really possible until digital manipulation advanced sufficiently. This could certainly be described as an "omnipresent grating sound," though it is in no way the fault of digital to analog conversion. Without having any idea what he's listening to or comparing, it's hard to say if his complaint is purely perceptual bias, or just a confusion of correlation and causation.
Edit: I wouldn't be quick to completely dismiss your friends' perception. Is analog a better medium for sound reproduction? Absolutely not. Could a digital "remaster" be made to sound much worse than the original? Of course.
The noise floor of digital recordings is something draws me to digital devices that definitely have a good noise floor themselves more often than not. I've had my moments where I preferred an analog version over the digital version (usually if the digital version is a bad remaster) but it's rare and even rarer a noisier version of something (old school game consoles (http://www.sega-16.com/forum/showthread.php?7796-GUIDE-Telling-apart-good-Genesis-1s-and-Genesis-2s-from-bad-ones) that didn't have the greatest of sound systems (http://www.sega-16.com/forum/showthread.php?26568-Introducing-the-Mega-Amp-The-universal-Genesis-audio-circuit), usually). Not going to lie there.
My own impression is that with 16 bit audio, jitter starts to become audible at -96 dB (without dithering), so with music recorded at sensible levels, it will not be audible, as all jitter is in the noise floor, and the music would mask the jitter.
I'm curious to learn how you arrived at this impression. Is it due to the diagram in Ethan's video? If so, you would read it wrongly. The diagram has nothing to do with audibility. It tries to answer the question what magnitude the jitter artifacts have in what situation. Whether they are audible is an entirely different question which depends on many more factors, amongst them are masking properties of human hearing.
The diagram you found in Ethan's video essentially derives from work done by the late Julian Dunn, who is sometimes quoted by audiophiles in quite misleading ways. You read the diagram like this: If your audio signal is a pure sine wave of the frequency shown on the horizontal axis, then a jitter signal which is also a sine wave will create modulation products of the level shown on the vertical axis, when its magnitude is as written on the diagonal lines. Note that both audio and jitter signals are assumed to be sinusoidal. That's not the case an audiophile is usually interested in. It is the case where jitter effects are most readily noticeable. If the jitter signal is noise, which is the usual case, the effect is orders of magnitude less audible. This has been confirmed by listening tests. Similarly, when the audio signal is music, the effect is also much less audible. The highest sensitivity of human hearing for jitter effects arises, when both jitter and audio signals are sine waves, and the frequency of the audio and jitter signals is chosen such that the modulation products fall into the most sensitive region of hearing, and are outside the masking region. In other words, you can handcraft a scenario that maximizes the chance of hearing it. If you do that, some people can detect jitter in the lower single figure ns range. With ordinary music, you are in the double figure ns. If the jitter signal is noise, you need a jitter magnitude of well beyond 100 ns to have a chance of hearing it.
It is fairly easy to build digital gear with jitter figures of a nanosecond or below, so unless someone has goofed, pretty much any gear should be good enough, jitter-wise. Therefore, it is a non-problem. If you find gear that exceeds a few ns of jitter, consider it defective or misdesigned. Both of those may happen, of course. Competition for ever-lower jitter figures well below a nanosecond may be technically challenging, but is sonically irrelevant.
Note also that jitter is measured and expressed in a variety of different ways which may render two figures uncomparable. Hence to be sure you would have to know more about the measurement conditions than most people are likely to tell you. In audiophile circles, hardly anyone seems to have enough clue of even the basics to deliver a meaningful measurement, so the figures which are claimed for various pieces of gear are likely bunk. Sometimes people buy an oscillator off-the-shelf and use the manufacturer data, without checking if it still holds true on their board, which is far from self-evident, and usually overoptimistic.
And then people confuse interface jitter with sampling jitter, which again leads to complete nonsense.
As for how jitter sounds, the video "Digital show & tell" on Xiph.org says quantization noise sounds like tape hiss from analogue tapes, but I might be mixing up quantization noise with jitter (is it the same?).
No it isn't the same. It has entirely different root causes. Quantization noise would be there regardless of whether there is jitter or not. Quantization noise is amplitude noise, whereas jitter is time-domain noise. Jitter modulates the audio signal, whereas quantization noise simply adds to the signal. The effects of both can be similar if the jitter signal indeed takes the form of random noise, which it may or may not, depending on its origin.
On that website, there's also a file available for download with a 1 kHz tone at -105 dB. When playing that file there's background noise, which I assumed was jitter.
No, it is quantization noise. In fact, it is dither, which is a way of making quantization noise purely random, so that it sound like continuous noise. Without dither, quantization noise may be correlated to the audio signal, which means that the noise coloration changes with the audio signal. The file is supposed to demonstrate that the noise floor is not a limit for audibility. It is possible to hear signals which are below the noise floor. This has to do with the fact that noise is a broadband effect across a whole range of frequencies, whereas the sine wave is a narrow band thing with a narrowly defined frequency. That can be exploited by both the ear and by measurement instruments.
In other words, jitter is present in all digital audio, but the amount is so low that you can never hear it except for in those extreme cases mentioned above.
That's a fair characterization for the vast majority of cases.
My friend's attitude is that jitter is omnipresent and always audible and smears the stereo image etc. (like the comment above)
He almost certainly never checked how much jitter was present in his own trials. And he never checked how much jitter would have to be added in order to make its effects audible. His opinion is pure belief, instilled by "interested parties".
I of course understand that when creating converters like Dan Lavry does, it's important to minimise jitter as much as possible in order to come closer to creating the perfect product, as well as creating an A/D converter that will have minimal jitter so the recording artists can raise and lower levels on different tracks as much as they like. But as mentioned, I'm only interested in audibility and listening at home – not in the technical aspect (measuring, etc.).
Minimizing jitter can be an interesting technical challenge, which can be pursued independently from any audibility considerations. If audibility is the key factor, it is enough to follow ordinary design practices, and check the result with your favorite audio analyzer.
NANOSECONDS? Try picoseconds:
https://www.massdrop.com/buy/centrance-dacp...-dac-combo/talk (https://www.massdrop.com/buy/centrance-dacport-slim-amp-dac-combo/talk)
Thanks for all your responses, everybody, and thanks for the jitter soundclips, Arnie. You answered my question of "how does jitter sound?"
I better state something about my friend before anybody else makes any more comments about him. I thought I made it clear in my original post, but apparently not:
To the best of my knowledge he has never owned a turntable. He's not a vinyl freak. He's been asking me about what gear to buy and what particular vinyl records I could recommend (I was "vinyl only" for 15 years, and have now done a lot of A/B'ing with vinyl and CD as well as various turntables). He's a hi-fi manufacturer, and he's considering buying a turntable purely for his showroom, as many potential clients could come in and ask for a demonstration with a turntable (you know, analogue is always better - we all know that). At one point he said he wouldn't want to switch to vinyl as it was too troublesome (cleaning, the price, etc.).
As for what types of music he listens to, then it's varied, but mostly "nice" modern recordings (Diana Krall and the like) and classical music. Although he might listen to a few compressed albums (like Daft Punk) here and there, it doesn't seem common for him.
And yes, I think he might have been influenced by the audiophile community who's nowadays constantly talking about jitter. 20 years ago it was never an issue, whereas nowadays it's apparently on everybody's lips as the new thing that everybody has to fix. I mean, I'm also asking this question here because I've been reading those sort of articles/comments.
Did anybody read the response I got on Dan Lavry's forum? The link is in my original post. I know that he's somewhat in the "other" camp (audiophile camp), although he seems very, very scientifically minded. Yet, the response from one of his employees seems to be the complete opposite of all your responses: Jitter is always present, and will give a grating sound, unless it's drastically reduced (and still it will always be somewhat audible). I'm still a novice in this field, so I can't concur or contradict anything. I am, of course, aware that Lavry's company is selling a product, and if they state that jitter is an issue, and they then can sell us a product that solves this problem, then it's money in their pockets. That said, I do have a lot of respect for them, and they seems to be some of the most straightforward, no-bullshit manufacturers out there .
And yes, I probably understand Ethan Winer's picture like you explained it, Pelmazo. Also I understood the short entry about Jitter here http://wiki.hydrogenaud.io/index.php?title...Vinyl%29#Jitter (http://wiki.hydrogenaud.io/index.php?title=Myths_%28Vinyl%29#Jitter) as saying jitter was just in the noise floor.
So, anyway, if I understand you all correctly, then jitter will be audible in a recording where the analogue to digital conversion was poorly done, or it will be present in a well-converted recording (A/D) if the digital to analogue conversion (by the DAC in the CD player) is being poorly done. The latter is not so likely nowadays (if it ever really was). The first one might have been a problem in the 80s, or it can still be a problem if poorly designed A/D converters are used. And there might also be other things in the chain that could have gone wrong, so it can be a bit random if jitter is audible or not.
Did anybody read the response I got on Dan Lavry's forum? The link is in my original post.
I got like 10 lines in. Does that count?
I'm still a novice in this field, so I can't concur or contradict anything. I am, of course, aware that Lavry's company is selling a product, and if they state that jitter is an issue, and they then can sell us a product that solves this problem, then it's money in their pockets. That said, I do have a lot of respect for them, and they seems to be some of the most straightforward, no-bullshit manufacturers out there .
Salesmen must love you.
Did anybody read the response I got on Dan Lavry's forum? The link is in my original post. I know that he's somewhat in the "other" camp (audiophile camp), although he seems very, very scientifically minded. Yet, the response from one of his employees seems to be the complete opposite of all your responses: Jitter is always present, and will give a grating sound, unless it's drastically reduced (and still it will always be somewhat audible). I'm still a novice in this field, so I can't concur or contradict anything. I am, of course, aware that Lavry's company is selling a product, and if they state that jitter is an issue, and they then can sell us a product that solves this problem, then it's money in their pockets. That said, I do have a lot of respect for them, and they seems to be some of the most straightforward, no-bullshit manufacturers out there .
Much of the factual information he offers is correct, but you must have noticed that for all the subjective opinions and experiences he gives, he doesn't even attempt to provide any quantitative estimates, or verifications. That's the usual state of affairs for the last decades: People who are convinced of the importance of jitter and of its audibility, have no hard evidence on offer for it. One would wish that particularly those who can give a more or less accurate technical account of what jitter is, would go the extra mile and verify their perceptions with a well designed experiment. You can contemplate why this doesn't happen.
Let me add a few corrections to Brad Johnson.
Strictly speaking, jitter does not count as quantization error. The process of converting analog signals to digital involves two processes: Sampling and quantization. They both occur in the A/D converter, and often even at the same point in the circuit, but it is useful to keep them separate on a conceptual level. Sampling is a time-domain thing, because it involves taking snapshots of the signal at regularly spaced points in time. This can be done in a purely analog fashion with no digital circuitry involved. Quantization is an amplitude-domain thing, as it separates a continuum of analog voltage levels into a fixed number of "bins", with each bin carrying a number. The CD for example, with its 16-bit format, has 65536 bins. Quantization introduces errors, because a range of voltages falls into a single bin, so once quantized they can't be distinguished any more. Note that the error needn't be noise in the common sense. For example, if you have a very low level signal whose amplitude is so small that it falls into a single bin, the resulting digital code is always the same number, which means silence. It takes dither to randomize the error, so that it becomes noise. With dither, the situation is exactly as it would be in analog, where there is always a noise floor, even when there is no signal. Noise floor and signal are effectively independent. Furthermore, if you provide more bins (i.e. more bits in the number), the range for each bin gets smaller, and so the average error gets smaller. The result is a reduction in the level of the noise. One more bit doubles the number of bins, halves the amplitude range that is represented by each bin, and reduces the average error by half. The noise level reduction is approximately 6 dB.
Since Jitter is a time-domain thing, it needs an audio signal to produce an effect. If a voltage is constant over time, it doesn't matter how accurately you hit the right points in time when sampling. The resulting numbers would be the same. The error resulting from sampling at the wrong point in time will be largest when the signal goes through a steep slope, i.e. varies rapidly. This happens at high frequencies at high level. Therefore, the best test signal for measuring the effects of jitter is a high pitched sine wave near the upper end of the frequency scale at full volume. That's another reason for being sceptical about the audibility of jitter, since music does not typically contain high frequencies at full volume. (This is, by the way, the reason for the sloped lines in Ethan's diagram.)
Brad Johnson probably knows this, and I don't think he wanted to misrepresent anything. It is a sloppy way of talking about these things that lumps the errors into a common concept. I just happen to believe that it doesn't help understanding, so I prefer to stick to the clearly defined terms.
I also prefer to emphasize the distinction between sampling jitter and interface jitter. Brad Johnson did make this distinction in spirit, but it wasn't very obvious. It is important to clarify this. Sampling jitter is what we are concerned with when we talk about errors at the point of conversion from analog to digital or vice versa. The relevant clock signal for this is the clock that drives the sampler in the A/D converter. This clock is very rarely accessible from the outside of a box. It may be accessible on a circuit board, and if you are equipped with enough knowledge and tools you may be able to measure its jitter directly. In practice, it is much easier to measure the jitter effect in the resulting audio, because that can be done outside of the box.
Interface jitter happens at a digital interface, such as S/P-DIF. Since the signal at this point doesn't directly drive a converter, its effect on the sampling quality is unknown. It may not have any effect at all. Measuring the interface jitter therefore doesn't allow any conclusion about the audio quality. Interface jitter is usually measured with an oscilloscope displaying a so-called eye diagram. The purpose of such measurements is to assess the likelihood of bit errors, i.e. the probability that you get errors in the data recovery at the receiving end. If there are such data recovery errors, the wrong data likely leads to clicks in the audio, which will be fairly apparent and bears no resemblance to the kind of impairment commonly associated with jitter.
There are cases where the sampling clock is derived from the incoming digital signal. In such cases, interface jitter may result in sampling jitter. Whether this happens, and to what extent it happens, depends on the details of the circuit involved. Gear does vary in this respect. This is what Brad described as "jitter immunity". This term describes the relationship between interface jitter and sampling jitter, i.e. how much of the former results in how much of the latter. Jitter immunity matters if your situation involves both of the following:
- You have a significant amount of interface jitter
- The receiver of the digital signal derives the sampling clock from it.
Brad noted a situation when the second point doesn't apply: The source device is synchronized to the receiver, i.e. there is a second connection feeding a synchronization signal from the receiver to the signal source. In this case the receiver can generate the sampling clock locally, and interface jitter has no effect on sampling jitter.
So, anyway, if I understand you all correctly, then jitter will be audible..
No, you don't understand correctly. There is
zero evidence that jitter is, or ever had been
audible in Hi-Fi equipment. Ever. There
is evidence to the contrary.
If you, your friend or Lavry et al could present evidence to the contrary, by all means do so. But it's been more than 30 years. Still waiting for these witch effects (http://www.enjoythemusic.com/magazine/bas/1208/) to show up at dawn during trust your ears only evaluations.
"Omnipresent jitter" is one of the many symptoms of audiophile disorder (a subset of Digital disorder).
I was "vinyl only" for 15 years, and have now done a lot of A/B'ing with vinyl and CD as well as various turntables
How many of those A/Bs were done with vinyl vs a level matched 16/44 rip of that vinyl?
cheers,
AJ
How many of those A/Bs were done with vinyl vs a level matched 16/44 rip of that vinyl?
I do have a question about ABXing vinyl and vinyl rip. A vinyl rip can have very consistent performance during playback because it is already digitized, but can a turntable have enough consistency in each playback? For example, if we rip a vinyl two times and ABX the two rips, can they sound different? Are there any research about this?
So, anyway, if I understand you all correctly, then jitter will be audible..
No, you don't understand correctly. There is zero evidence that jitter is, or ever had been audible in Hi-Fi equipment. Ever. There is evidence to the contrary.
If you, your friend or Lavry et al could present evidence to the contrary, by all means do so. But it's been more than 30 years. Still waiting for these witch effects (http://www.enjoythemusic.com/magazine/bas/1208/) to show up at dawn during trust your ears only evaluations.
"Omnipresent jitter" is one of the many symptoms of audiophile disorder (a subset of Digital disorder).
I was "vinyl only" for 15 years, and have now done a lot of A/B'ing with vinyl and CD as well as various turntables
How many of those A/Bs were done with vinyl vs a level matched 16/44 rip of that vinyl?
cheers,
AJ
Ha, they both sound like crap.
can a turntable have enough consistency in each playback?
It is conceivable that the position of a click may give enough of a hint for identifying the source that is playing. The limitations of vinyl players certainly limit ABX test protocols. You just can't repeat a signal as easily as you can loop a piece of a sound file.
I love the way people casually throw around "A/B"
analogue is always better - we all know that
I hope you're only joking.
I do have a question about ABXing vinyl and vinyl rip. A vinyl rip can have very consistent performance during playback because it is already digitized, but can a turntable have enough consistency in each playback? For example, if we rip a vinyl two times and ABX the two rips, can they sound different? Are there any research about this?
Don't know, but the rip will capture all that lovely vinyl + TT blah, blah sound. I have my doubts that one subsequent replay would be discernible (i wouldn't try it with 100), but if not, it puts those with the disorder at an advantage, in that there would be something to differentiate. So now all we need is one positive "A/B".
It should be good therapy either way, however temporary.
cheers,
AJ
I do have a question about ABXing vinyl and vinyl rip. A vinyl rip can have very consistent performance during playback because it is already digitized, but can a turntable have enough consistency in each playback? For example, if we rip a vinyl two times and ABX the two rips, can they sound different? Are there any research about this?
The most obvious problem here would be static noise and pops, since it is not consistent between playbacks, and could be used to identify the 'live' vinyl record with enough repeated tests.
As I've mentioned here in other threads, the proper way to ABX test direct, live vinyl to a digitized version of it is via a nearly instantaneous (low latency) A>DA>A loop, the exact same methodology used by Meyer and Moran (http://www.aes.org/e-lib/browse.cfm?elib=14195) to compare the live, analog out of an SACD player to a digitized version it via a standalone CD recorder put in "record monitor" mode. Pops, ticks, and clicks won't give away which source is being played because you are hearing the exact same playback session of the vinyl either way.
Funny you should use the word loop.
Great method, thanks mzil, and thanks for the aes link as well. Audiophiles should spend a bit to read the paper rather than spending a lot to "fix" a nonexistent problem.
Did anybody read the response I got on Dan Lavry's forum? The link is in my original post. I know that he's somewhat in the "other" camp (audiophile camp), although he seems very, very scientifically minded.
To the poorly informed, a good charlatan will sound very, very scientific. It doesn't matter which part of science we're talking about, this is how pseudoscience is done for fun and profit.
Yet, the response from one of his employees seems to be the complete opposite of all your responses: Jitter is always present, and will give a grating sound, unless it's drastically reduced (and still it will always be somewhat audible).
That's the usual line of pseudoscientific bull - a mixture of true and false claims.
True statement: Jitter (and every other form of noise and distortion) is always present. (because nothing in the real world is perfect).
False claim: Jitter will give a grating sound unless it's drastically reduced (and still it will always be somewhat audible).
True correction to the above false claim: Jitter can easily be low enough that it has no audible effect.
I'm still a novice in this field, so I can't concur or contradict anything. I am, of course, aware that Lavry's company is selling a product, and if they state that jitter is an issue, and they then can sell us a product that solves this problem, then it's money in their pockets.
Hold those thoughts.
That said, I do have a lot of respect for them, and they seems to be some of the most straightforward, no-bullshit manufacturers out there .
Not at all.
And yes, I probably understand Ethan Winer's picture like you explained it, Pelmazo. Also I understood the short entry about Jitter here http://wiki.hydrogenaud.io/index.php?title...Vinyl%29#Jitter (http://wiki.hydrogenaud.io/index.php?title=Myths_%28Vinyl%29#Jitter) as saying jitter was just in the noise floor.
That's the nature of noises and distortion that are too low to be audible by anybody - they are lost in the noise. In the real world there is always a noise floor.
So, anyway, if I understand you all correctly, then jitter will be audible in a recording where the analogue to digital conversion was poorly done, or it will be present in a well-converted recording (A/D) if the digital to analogue conversion (by the DAC in the CD player) is being poorly done. The latter is not so likely nowadays (if it ever really was). The first one might have been a problem in the 80s, or it can still be a problem if poorly designed A/D converters are used. And there might also be other things in the chain that could have gone wrong, so it can be a bit random if jitter is audible or not.
All recordings that are, will be, or ever were made on analog media (analog tape and LP) are very likely to have audible jitter because one of the reasons that analog media was generally scrapped is because its jitter and other forms of audible noise and distortion could never be reduced below audibility.
If a recording is made digitally, thus avoiding analog media, using audio gear with average or better quality, then it is likely to be free of audible jitter and every other kind of audible flaw. This is easy enough to demonstrate by means of a straight wire bypass test where a back-to-back analog and digital converter is compared to a short piece of wire. It is easy enough to do this by comparing a recording to itself after being re-recorded digitally.
By not directing people to do straight wire bypass tests under bias-controlled conditions, the myth that digital converters and all other kinds of audio gear are not quite good enough is perpetuated for fun and profit.
As I've mentioned here in other threads, the proper way to ABX test direct, live vinyl to a digitized version of it is via a nearly instantaneous (low latency) A>DA>A loop, the exact same methodology used by Meyer and Moran (http://www.aes.org/e-lib/browse.cfm?elib=14195) to compare the live, analog out of an SACD player to a digitized version it via a standalone CD recorder put in "record monitor" mode. Pops, ticks, and clicks won't give away which source is being played because you are hearing the exact same playback session of the vinyl either way.
So maybe that's why I don't reach a transcendent state of blissful nirvana and enlightenment when I play my LPs, rather I just enjoy them as I would the same music in any other format.
It's that blasted A>D>A DSP crossover I'm using, surely that must be robbing me of pure audiophile bliss
Good point though, that's obviously the best way to do that ABX test.
Apologies if it was on this very site, I clicked a link and saved the PDF --- and I cannot remember where I saw the link.
No doubt familiar to many here, it looks like a vital piece of education for a recovering audiophile*, so being one (although I have never subscribed to the analogue/digital divide) I read it and saved it...
The Dawn of Commercial DIgital Recording, Thomas Fine, ARSC Journal (https://www.google.co.in/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CB4QFjAAahUKEwiUw_nA7s3HAhVKGI4KHdL_AAQ&url=http%3A%2F%2Fwww.aes.org%2Faeshc%2Fpdf%2Ffine_dawn-of-digital.pdf&ei=3WvhVdTgBcqwuATS_4Mg&usg=AFQjCNEJcKgrO_zJccg4OnTiYMXYIR8XhQ&sig2=9D0oIgFIYI0B9cGX1yGN1A).
Amazing, how so many audiophiles seem to think that digital audio was an assault on the very basis of music, whereas it was actually adopted and further developed by the recording companies to improve their results. "...Denon's stated purpose: "To produce recordings that were not compromised by the weaknesses of magnetic tape recorder." (p3: Pre-Dawn: Denon Introduces Music to Digital)
*or, even, people interested in digital history rather than myth
No doubt familiar to many here, it looks like a vital piece of education for a recovering audiophile*, so being one (although I have never subscribed to the analogue/digital divide) I read it and saved it...
The Dawn of Commercial DIgital Recording, Thomas Fine, ARSC Journal (https://www.google.co.in/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&uact=8&ved=0CB4QFjAAahUKEwiUw_nA7s3HAhVKGI4KHdL_AAQ&url=http%3A%2F%2Fwww.aes.org%2Faeshc%2Fpdf%2Ffine_dawn-of-digital.pdf&ei=3WvhVdTgBcqwuATS_4Mg&usg=AFQjCNEJcKgrO_zJccg4OnTiYMXYIR8XhQ&sig2=9D0oIgFIYI0B9cGX1yGN1A).
Amazing, how so many audiophiles seem to think that digital audio was an assault on the very basis of music, whereas it was actually adopted and further developed by the recording companies to improve their results. "...Denon's stated purpose: "To produce recordings that were not compromised by the weaknesses of magnetic tape recorder." (p3: Pre-Dawn: Denon Introduces Music to Digital)
Digital haters who seem to also tend to be analog fan boys, apparently want to forget about that approximate 10 year period near the end of the heyday of analog, and just before the arrival of consumer digital, where just about any self-respecting recording engineer did as much recording and production on digital gear as he could before the final LP master cutting session.
This listening test strongly relates:
Listening Test of digital production gear for analog media (http://djcarlst.provide.net/abx_digi.htm)
"
The Ampex 16 Bit Digital Delay Line vs. wire comparison was heard in a recording studio control room on time aligned UREI 813 speakers with McIntosh MC-2100 amplifiers. The audio source was a master 2-track 15 IPS tape on a Scully 280. This master tape had been mixed from a 24-track master tape on an Ampex MM-1000. The mixdown and playback was through an API console. The listeners included professional recording engineers with years of experience on major label projects, professional maintenance engineers, and recording engineering students.. For those not familiar with studio equipment, these are some of the most revered pieces of equipment of that day. API consoles are still prized today for their high quality.
"
This was essentially the same test configuration as the Meyer-Moran tests, only about 30+ years earlier (late 1970s-early 80s).
The Ampex 16 Bit Digital Delay Line was composed of stereo back-to-back ADCs and DACs separated by a digital delay line that preserved every digital data bit, but added a user-adjustable amount of delay. For the bulk of the testing the delay was set to zero, so the converters were back-to-back.
The purpose of the Ampex 16 Bit Digital Delay Line may be instructive of some of the technical details of SOTA vinyl master production.
It is always a challenge to fit a useful amount of wide dynamic range audio on a LP side. The louder the track, the larger the cutting pitch or groove spacing that is required to avoid damaging the immediately previously cut groove a few thousands of an inch back.
If the track has a quiet passage then reducing the track's pitch is good practice. Unfortunately this sort of decision making has to be made before the loud passage is cut. Therefore SOTA tape mastering machines of the day had a "Preview head" that played the music back approximately one revolution of the LP in advance of the tape head that played the music back for the purpose of cutting the groove. The preview head fed analog cutting head positioning circuits to optimize the groove pitch before the groove was actually cut. Therefore the tape machine had to have two identical sets of tape heads whose performance needed to be optimized and matched.
The Ampex 16 Bit Digital Delay Line eliminated the need for the second tape head, but its sound quality was of paramount importance because it controlled the sound quality of the final LP master, which needed all the help it could get. If it were indistinguishable from the canonical "Straight Wire With Gain" then it could not be further improved on.
The recording engineers whose work product would be profoundly affected by the sonic performance of this digital device threw everything at it that they could. Even recordings of the classic "Keys jangling" were used. It became clear that the SOTA and carefully optimized high speed wide track mastering tape machines were a little less than sonically perfect, but the Ampex delay line shrugged off the most challenging program material.
For those who claim that the digital audio converters that were used 10 years later could not be up to the task at hand, these test results are very challenging. The unimpeachable sonic performance of this device from a decade earlier tells a different story. In the early days of consumer digital, the existing and past technology was up to the task, and was demonstrably sonically transparent.
Just a general point concerning white papers - IME I have found that though they are often technically correct, they often leave out important context and can be misleading.
IOW:
Statement A: "this can cause noise"
Statement B: "our process ensures that this noise will never cause a problem"
Though both A and B might be 100% true, the part they hope one infers - that without their process the noise will be a horrific problem - is never actually explicitly stated anywhere. Instead there might be careful dancing around the audibility question and hypothetical statements like, "if this distortion is great enough, it will...".
So I read white papers as technical marketing documents and read carefully for signs they are dancing around saying things explicitly and ask myself, "If it's a problem in the real world, why don't they just say that? Why all the dancing?".
Though both A and B might be 100% true, the part they hope one infers - that without their process the noise will be a horrific problem - is never actually explicitly stated anywhere. Instead there might be careful dancing around the audibility question and hypothetical statements like, "if this distortion is great enough, it will...".
There's a positive side to this, which IMHO shouldn't be neglected: By doing this, they try to avoid an outright lie. This may of course be an attempt at minimizing legal complications, but some of them might actually have preserved some professional standards that they are reluctant to give up. This is in contrast to others who have absolutely no inhibition to take the piss out of their customers.
The question is how to act in a situation where you know that your business depends on customers who believe that your product delivers better sound quality than your competition. Even worse, telling people the naked truth would be met with disbelief and even opposition, so you would be in danger of alienating your customer base.
Is it any wonder that even the more sincere amongst the manufacturers try to find a compromise which permits their deluded customers to keep their delusions, and the manufacturers to keep their revenues?
I think that Lavry is one of the more sincere amongst the high-end manufacturers. I tend to respect that, even though the customers who want to keep their delusions will find plenty of pretexts and excuses in Lavry's writings that support them in that. Lavry himself, I am confident, knows which part of it is true and which isn't.
Though both A and B might be 100% true, the part they hope one infers - that without their process the noise will be a horrific problem - is never actually explicitly stated anywhere. Instead there might be careful dancing around the audibility question and hypothetical statements like, "if this distortion is great enough, it will...".
There's a positive side to this, which IMHO shouldn't be neglected: By doing this, they try to avoid an outright lie.
The moral advantage of an intentional deception over an outright lie being exactly what?
I personally favor the outright lie because it may be easier to detect! ;-)
It is sad that so much of audio has come down to this sort of thing. There is real progress yet to be made, but so much energy is wasted.
The moral advantage of an intentional deception over an outright lie being exactly what?
It contains at least a tacit admission that the truth matters to the deceiver, even if he's prepared to obfuscate it for profit. Not very much of an advantage, I know.
I personally favor the outright lie because it may be easier to detect! ;-)
And easier to attack ;-)
It is sad that so much of audio has come down to this sort of thing. There is real progress yet to be made, but so much energy is wasted.
That's still nothing compared to the wastage due to other delusions. Truly sad indeed.
It is sad that so much of audio has come down to this sort of thing. There is real progress yet to be made, but so much energy is wasted.
That's still nothing compared to the wastage due to other delusions. Truly sad indeed.
But the other wastage is often based on the carefully worded deceptions.
Well-intentioned persons often end up contributing to these deceptions. One classic example of this that I will boldly proffer, full knowing of his good technical work and numerous fans: NWAVGUY.
In my opinion he plied the world of two-orders of magnitude overkill for fun and profit (at least profit for others) and raised concerns that were psychoacoustically questionable to say the least.
But high fidelity enthusiasts are concerned with the psychoacoustically questionable.
An enthusiast is not interested in equipment that is "just" good enough to satisfy some clear cut psychoacoustic requirements based on the average listener, except if it is e.g. especially cheap. He/she wants more, probably willing to spend a lot more to gain very little.
It's like the computer enthusiast, that buys a more expensive CPU cooler to achieve 1° lower temperature which effectively makes no other difference.
That's why nwav set the bar(s) higher. If you're cynical you may say that he set them higher such that "inferior" products could not meet them, while his projects do... but even that is awesome considering the prices on both sides.
---
A few ns jitter may not be audible with most music, but we should still strive for lower jitter - that is progress. Jitter is just a random example, but what's greater than all those advancements trickling down into cheaper gear, raising the average fidelity?
It is sad that so much of audio has come down to this sort of thing. There is real progress yet to be made, but so much energy is wasted.
That's still nothing compared to the wastage due to other delusions. Truly sad indeed.
But the other wastage is often based on the carefully worded deceptions.
Well-intentioned persons often end up contributing to these deceptions. One classic example of this that I will boldly proffer, full knowing of his good technical work and numerous fans: NWAVGUY.
In my opinion he plied the world of two-orders of magnitude overkill for fun and profit (at least profit for others) and raised concerns that were psychoacoustically questionable to say the least.
nwavguy wrote an entire article explaining how I am completely wrong for insisting on ABI that you can adequately measure the performance of a 16 bit audio player using RMAA, the pair of headphones you want to use with it and an inexpensive sound card. His perspective was very much in favor of excessive perfectionism, both in terms of hardware and in terms of measurement.
On the other hand, I did enjoy the precision with which he measured things. I don't think it was often necessary, but if you're not doing the work, its great to have a few more decimal places worth of accuracy
But high fidelity enthusiasts are concerned with the psychoacoustically questionable.
An enthusiast is not interested in equipment that is "just" good enough to satisfy some clear cut psychoacoustic requirements based on the average listener, except if it is e.g. especially cheap. He/she wants more, probably willing to spend a lot more to gain very little.
It's like the computer enthusiast, that buys a more expensive CPU cooler to achieve 1° lower temperature which effectively makes no other difference.
That's why nwav set the bar(s) higher. If you're cynical you may say that he set them higher such that "inferior" products could not meet them, while his projects do... but even that is awesome considering the prices on both sides.
---
A few ns jitter may not be audible with most music, but we should still strive for lower jitter - that is progress. Jitter is just a random example, but what's greater than all those advancements trickling down into cheaper gear, raising the average fidelity?
PICOSECONDS! Sheesh...
At 170ps in a <$200 sound card, jitter is approaching the limits of measurement, let alone audibility:
http://www.stereophile.com/content/asus-xo...ds-measurements (http://www.stereophile.com/content/asus-xonar-essence-ststx-soundcards-measurements)
Edit: Naturally, even though it has less usability problems, the the Xonar doesn't hold up to JA's $2500 Ayre QB-9, for some unknown reason.
And I see no problem with products with 2-digit picoseconds jitter for enthusiasts either.
However, there appear to be huge problems in audiophool circles: FUD and focusing just on something like minimizing jitter while ignoring far more important things.
And I see no problem with products with 2-digit picoseconds jitter for enthusiasts either.
However, there appear to be huge problems in audiophool circles: FUD and focusing just on something like minimizing jitter while ignoring far more important things.
Such as?
And I see no problem with products with 2-digit picoseconds jitter for enthusiasts either.
However, there appear to be huge problems in audiophool circles: FUD and focusing just on something like minimizing jitter while ignoring far more important things.
Such as?
Speakers and room acoustics have been far and away the largest source of noise and distortion in any audio system since we moved beyond analog media.
The noise and distortion in rooms and speakers struggles to be as much as 40-50 dB below listening levels.
Compare that with the noise and distortion in even just mediocre amps and dacs - more like 70-90 dB down.
Oh, definitely, though ambient room noise probably isn't as likely as computer onboard audio to trick your brain into expecting an imminent insect attack.
Do any of the "Hifi" rags like Stereophile even talk about room acoustics?
Do any of the "Hifi" rags like Stereophile even talk about room acoustics?
Rarely because there's (almost) nothing to sell.
I've recently made some big improvements to my room by the use of carpeting, drapes, strategic furniture placement, and some modifications to my air ventilation system but how many of these things, I bought, are advertised in rags like Stereophile? None. They cater to their advertisers so pushing overpriced wire, amps, and DACs is their bread and butter, not articles about rugs, drapes, etc. which actually do make a difference.
And I see no problem with products with 2-digit picoseconds jitter for enthusiasts either.
However, there appear to be huge problems in audiophool circles: FUD and focusing just on something like minimizing jitter while ignoring far more important things.
This article claims jitter can be audible at 20ps:
[a href="http://www.madronadigital.com/Library/AudibilityofSmallDistortions.html" rel="nofollow"]http://www.madronadigital.com/Library/AudibilityofSmallDistortions.html[/a]
All you have to do is slow your pulse and respiration, put on a good pair of headphones and listen to John Cage's 4'33" in a sensory deprivation chamber. You're bound to get a spurious 2kHz tone eventually.
I am interested in his claim that mass-market AV receivers can have thousands of picoseconds of jitter over HDMI. Granted, this is also a video connection, which makes you even less likely to notice distortion. Does anyone know how to convert picoseconds of jitter to a decibel level?
Most question already answered here (http://www.hydrogenaud.io/forums/index.php?showtopic=51322&st=25).
Btw, the big HDMI jitter problem has been fixed years ago. Products released over 5 years ago did already manage ~300ps and lower total correlated jitter. Arcam had a product back then (maybe still has the same) with essentially unmeasurable jitter.
This article claims jitter can be audible at 20ps:
If you know about the posting history of that author here and at the AVS forum you'd know that he is nothing more than a pompous, insufferable lobbyist for the grossly overpriced DACs and other "high end" snake oil manufacturers which he sells at his retail store. I'd advise avoiding any material he's written, it's a waste of your time.
I exposed him as a liar in post #3 here (http://www.hydrogenaud.io/forums/index.php?s=&showtopic=108127&view=findpost&p=887328).
And I see no problem with products with 2-digit picoseconds jitter for enthusiasts either.
However, there appear to be huge problems in audiophool circles: FUD and focusing just on something like minimizing jitter while ignoring far more important things.
This article claims jitter can be audible at 20ps:
http://www.madronadigital.com/Library/Audi...istortions.html (http://www.madronadigital.com/Library/AudibilityofSmallDistortions.html)
The author was given the opportunity to back up his claims with a listening test and failed. His claims are based on other works that themselves were not generally based on reliable listening tests.
Jitter is one of the things that people have pontificated on for about 30 years while either not gathering or ignoring evidence that is contrary to their beliefs. The beliefs are a kind of a license to pontificate, so they tend to be self-perpetuating.
All you have to do is slow your pulse and respiration, put on a good pair of headphones and listen to John Cage's 4'33" in a sensory deprivation chamber. You're bound to get a spurious 2kHz tone eventually.
That there is a 2 KHz tone or audible tone at any other frequency due to real world jitter to hear is generallly a misapprehension. Jitter adds sidebands that while they show up as something like tones on a FFT are generally not heard that way.
I am interested in his claim that mass-market AV receivers can have thousands of picoseconds of jitter over HDMI.
I would take that to be something that was a fact at one time in some cases. However just because the jitter could be described with scary numbers does not mean that it was a problem. His proposed solution which was ansynch USB converters have some pretty severe-seeming inherent problems of their own, which are generally obfuscated by the industry. A blogger named Archimalgo has writtten what little there is that has been backed up with even just technical tests.
Remember, no reliable listening test, and you're just dealing with speculation.
Granted, this is also a video connection, which makes you even less likely to notice distortion. Does anyone know how to convert picoseconds of jitter to a decibel level?
You need to know the frequency and amplitude of the jitter and the frequency and amplitude of signal being jittered to do that. If you know those four things then the conversion is one of those things you learn in an advanced undergraduate class in modulation theory.
All you have to do is slow your pulse and respiration, put on a good pair of headphones and listen to John Cage's 4'33" in a sensory deprivation chamber. You're bound to get a spurious 2kHz tone eventually.
Just tried. The frequency of my tinnitus seems higher than 2khz
I am interested in his claim that mass-market AV receivers can have thousands of picoseconds of jitter over HDMI. Granted, this is also a video connection, which makes you even less likely to notice distortion.
They also have DSP and thus buffer audio data which kills jitter.
Isn't it an audiophile fact that electrons have memory? Thus, signals which have been exposed to jitter, must surely be left feeling, and sounding... jittery!
... All you have to do is slow your pulse and respiration, put on a good pair of headphones and listen to John Cage's 4'33" in a sensory deprivation chamber. You're bound to get a spurious 2kHz tone eventually. ...
I'm sure you were just having a little fun there, but some readers don't know that and might take what you said as a fact.
(The sound of jittered silence is... silence.)
I am interested in his claim that mass-market AV receivers can have thousands of picoseconds of jitter over HDMI. Granted, this is also a video connection, which makes you even less likely to notice distortion.
They also have DSP and thus buffer audio data which kills jitter.
The high numbers for jitter that were measured were taken at the analog output of the AVR which were downstream of all of those enhancements. It was real.
The more serious problem is that there is no general knowledge among consumers or technicans about how many (pick a prefix indicating a small division of time)-seconds of what frequency jitter makes an audible difference in which signals at whatever audio frequency.
IME, most consumers can't be relied on to tell you whether a nanosecond was larger or smaller than a picosecond, and by how much. Obviously, the smaller unit of time you choose, the bigger the conventional numbers that describe the actual measurement. Good practice is to pick a unit of time that gives reasonable numbers, but obviously many in the industry obviously ignore those guidelines and pick very small units of time so as to contrive large numbers > 1,000.
Add to that the perceptual differences in how or whether jitter is audible depending on whether it is random or periodic, and the differences in audibility based on the frequency spectral content of both the audio signal and the jitter itself.
How much jitter is audible is thus not an question with a simple single number answer. Consumers generally lack the background and interest to manage just a single number. Reality is that large numbers of numbers are needed to describe the problem.
How much jitter is audible is thus not an question with a simple single number answer. Consumers generally lack the background and interest to manage just a single number. Reality is that large numbers of numbers are needed to describe the problem.
Most audiophiles would agree with this. The mantra that human hearing is complicated and mysterious and partly unknown, is almost universal amongst audiophiles.
While it is true, the good news is that you can get rid of such complications by taking "safe" limits, which contain enough contingency for all practically occurring variations, and use the resulting simple value as a guideline how to build gear. This has long been done with distortion. THD+N is a very simplistic measure compared to the complications of human hearing, yet it is fairly straightforward to put down a number that is "good enough" for all practical cases. "Good enough" meaning that even a golden ear won't notice a problem. We know from research that in the case of THD+N, the audibility limit is above 0.1%. That's therefore "good enough", and any better won't make any audible difference, with a high degree of confidence.
The same thing of course applies in the case of jitter, which basically means that if the jitter is below a couple of nanoseconds, you're on the safe side, even in the most critical case. Any better than that won't make an audible difference, with a high degree of confidence.
So all those large numbers of numbers can be condensed to a few simple data points, if the goal is merely to be on the safe side. It turns out that with today's technology, neither THD+N nor jitter poses a significant problem anymore, i.e. reaching the safe side is easy and needs no expensive technology. That would be a point that even the layman should be able to understand.
Agreed, and to further add to that it is important to note that except in very rare instances it is impossible to conceal an audibly objectionable level of jitter yet still show acceptable, friendly, low levels of THD+N. I learned this from the brilliant digital engineer Bob Adams of Analog Devices decades ago.
I am interested in his claim that mass-market AV receivers can have thousands of picoseconds of jitter over HDMI. Granted, this is also a video connection, which makes you even less likely to notice distortion.
They also have DSP and thus buffer audio data which kills jitter.
The high numbers for jitter that were measured were taken at the analog output of the AVR which were downstream of all of those enhancements. It was real.
Could you explain this in a little more detail? I'm not understanding why HDMI would influence jitter of the downstream system. HDMI is a packet-based digital protocol, how does the rate at which packets arrive influence the analog performance of the device? Unless the DAC is actually clocked off of the HDMI clock (which would be tough since a lot of different HDMI clocks are possible), the DAC's jitter shouldn't depend on if you fed it with HDMI or any other input. Just like how playing Spotify vs. iTunes doesn't change the jitter in your sound card.
The high numbers for jitter that were measured were taken at the analog output of the AVR which were downstream of all of those enhancements. It was real.
I don't trust anything that king of the snake oil peddlers says. I looked into Paul Miller's measurements on that unit they like to trot out to scare people, a boogieman, the Yamaha RXV3900, reviewed in Feb 2009, and although the unit got a "FAIL" rating, by my reading of the raw numbers it was because it only put out 95 wpc cleanly yet was marketed as 100w/ch or more, I guess. Look at the pages for PCM jitter tests and all I see is "Passing" grades, although it isn't patently obvious, I'm afraid, as to what's HDMI and what's not*. In fact his charts aren't that easy to interpret. It took me forever to deduce the m on the x-axis, on some graphs, is short for "magnitude", i.e. no real units given [but you can usually figure out when they mean dB, etc.]
Sign up for an account to register, if you don't already have one. It is free, easy, and solicitation free:
http://www.milleraudioresearch.com/avtech/index.html (http://www.milleraudioresearch.com/avtech/index.html)
He gives some of the most detailed raw specs out there and I'd suspect you've already seen his site but did you ever actually examine the RXV3900 raw data yourself, rather than trust what others claim about it?
*CORRECTION: I see 96k/24 bit does get a "FAIL" for HDMI jitter, at the very bottom of the LPCM data for that Feb 09 test of the RXV3900. OOps.
Isn't it an audiophile fact that electrons have memory? Thus, signals which have been exposed to jitter, must surely be left feeling, and sounding... jittery!
You have to dilute your jitter to at least 30C.
I'm not understanding why HDMI would influence jitter of the downstream system. HDMI is a packet-based digital protocol, how does the rate at which packets arrive influence the analog performance of the device? Unless the DAC is actually clocked off of the HDMI clock (which would be tough since a lot of different HDMI clocks are possible), the DAC's jitter shouldn't depend on if you fed it with HDMI or any other input.
That AV receivers clocking system must be weak, is it HDMI transport layers jitter at all or just AV Receivers internal crappy clocking jitter?
Isn't it an audiophile fact that electrons have memory? Thus, signals which have been exposed to jitter, must surely be left feeling, and sounding... jittery!
You have to dilute your jitter to at least 30C.
And like cures like? So we pass our signal through a
homeopathic jittered solution to remove jitter?
Isn't it an audiophile fact that electrons have memory? Thus, signals which have been exposed to jitter, must surely be left feeling, and sounding... jittery!
You have to dilute your jitter to at least 30C.
And like cures like? So we pass our signal through a homeopathic jittered solution to remove jitter?
Precisely. Just be careful about grounding.
The high numbers for jitter that were measured were taken at the analog output of the AVR which were downstream of all of those enhancements. It was real.
I don't trust anything that king of the snake oil peddlers says.
I have no disagreement, there. But, I'm citing from the same body of Miller Reasearch as you appear to be, except I'm talking about the relevant HDMI info, and I see through the fog of numbers and understand that without proper interpretation in the light of audibility, that data is just numbers. It is even a pretty good example of cherry picking the units of measurement in order to produce impressive-looking numbers.
I looked into Paul Miller's measurements on that unit they like to trot out to scare people, a boogieman, the Yamaha RXV3900, reviewed in Feb 2009, and although the unit got a "FAIL" rating, by my reading of the raw numbers it was because it only put out 95 wpc cleanly yet was marketed as 100w/ch or more, I guess. Look at the pages for PCM jitter tests and all I see is "Passing" grades,
Obvious deflection because PCM related performance isn't the issue at hand. HDMI performance is.
Sign up for an account to register, if you don't already have one. It is free, easy, and solicitation free:
http://www.milleraudioresearch.com/avtech/index.html (http://www.milleraudioresearch.com/avtech/index.html)
Been there, done that years ago.
He gives some of the most detailed raw specs out there and I'd suspect you've already seen his site but did you ever actually examine the RXV3900 raw data yourself, rather than trust what others claim about it?
Despite your apparent hopes and false claims, I'm not that naive or poorly informed. I find it insulting that someone would assert that I was so foolish.. It also shows ignorance of easily knowable facts. because if one knew anything at all about past AVS-based discussions between Amir and I, they'd know that I had done my homework and studied the Miller Research site long before Amir brought it up.
It is you who may be playing the fool on several grounds. What are those raw specs but numbers? Where is the reliable data that relates to real world audibility? Why are you prattling on about LPCM data when HDMI is the issue at hand?
I'm the original poster. Thanks for all your responses, everybody. I hope this thread will never be removed as it really has cleared up many issues, and I'm sure many others will also find it useful.
I better clear something up:
1: I've never believed jitter was audible or a problem at all. I asked here to clear up my doubts (doubts instilled by the hi-fi industry). Although I misunderstood some things, then my assumptions were correct: Jitter is not audible except for a few exceptions. This point mostly goes to ajinfla (it seemed to me that you believed I claimed jitter was audible).
2: My vinyl vs. CD comparisons were simply taking the vinyl LP and the original CD (not a vinyl-rip) and compare them and see which media I liked the best. Then I would keep/buy that media. I used my own collection for this, so by now I have compared around 700 albums and singles, but that's a different discussion. This experience was one of the reasons why my friend asked me about record players and particularly good vinyl records.
As for whether a vinyl-rip and the original vinyl sounds the same, then I would say they do, if the rip was recorded properly. I'm sure Michael Fremer would say that he could hear a world of difference ;-). Anyway, I took several CDs and recorded them with a regular no-name very long cable from my CD player into the line-in on my computer, matched the levels (my cable does actually lower the volume of one channel) and then did an ABX test in Foobar. I couldn't hear any difference! Yet people pay gazillions for cables and A/D converters to record vinyl records. I downloaded some rips from one setup that used a turntable costing around 18,000 dollars (including an 8,000 dollar cartridge). On top of that he had used a cable from the phono preamp to the A/D converter costing around 2,800 dollar. Yet, my reasonably priced Rega sounded better than this setup!
And yes, the salesmen love me, as Saratoga said ;-). Unfortunately, much of the hi-fi industry is built on speculation, assumptions, reputation, price and subjective reviews. After all, nobody can know everything, and we all tend to listen to experts in certain fields (for instance earthquake experts) rather than go out and investigate for ourselves. So, as someone said, when a salesman or an audiophile on an internet forum tells us some pseudo science we'll believe it as we simply don't know any better. Many of the responses you've given me in this discussion are still too complex for me to fully understand, so in this case I also just listen to the experts (you).
So, a lot of us have been suckered in by aggressive salesmen to buy, or at least listen to things that don't change anything, or then change very little. And there seems to have been an enormous increase in focus on materials. Especially one particular salesman chew my ear off for two hours about how crappy my equipment was, how much crap everybody else was selling, how much CDs sucked, how much better vinyl was, how much better their equipment was than anybody elses, how much of an improvement audiophile grade power cords made, and so on, and if I didn't agree I wasn't serious about audio. As I was borrowing a CD player he also gave me an expensive power cord to try out as well, which another, very nice, seller had also done earlier (different brand), and in both cases I'm sure I could hear what improvements it made: it was exactly 0 %!
And as I sometimes say: If audiophiles had actually bothered to do listening tests of all their claims, there would be no arguing about jitter, 24 bit/16 bit, ultra-high sample rates, aliasing, etc. But alas! Hardly any audiophiles want to put those claims to the test.
Thanks again, everybody. I've been sick of thinking about hi-fi for quite a while, and now that all my stuff is in storage and I've gone three months to almost the other end of the world I can get some peace of mind .
Just to add:
Someone posted this link:
http://www.hydrogenaud.io/forums/index.php...51322&st=25 (http://www.hydrogenaud.io/forums/index.php?showtopic=51322&st=25)
In there an experiment with both "regular" people and "professionals" were conducted and there were quite clear results:
All participants could distinguish between sounds with and without time
jitter when the jitter size was 9216 ns. A few could when it was 1152 ns. No
one could when it was as small as 576 ns.
There was a question, however, if the result would depend on the listening
environments and the skill of the listeners. That is why we carried on the
second experiment. This second experiment is reported in the paper, the one
that you probably read.
Listeners in the second experiment were all professionals, audio engineers,
recording/mixing engineers, musicians, etc... Sound materials were selected
by the listeners so that each listener could use his (her) familiar
materials. The experimenter (we) visited the listeners' studios or listening
rooms so that we could use listeners' own DAC, amplifiers, loudspeakers and
headphones. The system configurations, therefore, varied among listeners.
They were mostly mid-end or above, I suppose.
As you can find in the paper, some listeners could distinguish the sounds
when time jitter was 500 ns. It could not be detected, however, when the
jitter was as small as 250 ns.
In Ethan Winer's "AES Damn lies" video he mentions that his USB soundcard is spec'ed at 0.5 ns. So what an enormous difference! That's really a number that says something. I'm very glad I saw those numbers. That really helped a lot! I really liked the ending of the portion I quoted parts of above: "I won't
believe someone who says that he can detect time jitter of 100 ps or less in CDs."
..., I'd suspect you've already seen his site
I find it insulting that someone would assert that I was so foolish.. It also shows ignorance of easily knowable facts. because if one knew anything at all about past AVS-based discussions between Amir and I, they'd know that I had done my homework and studied the Miller Research site long before Amir brought it up.
It is you who may be playing the fool on several grounds.
Yikes. Relax. You are attacking me for not being omniscient and knowing which sites you have or haven't examined? Are you serious? I already wrote, quoted above, that my gut feeling is that you probably had, but attacking me for not being 100% sure if you had examined the raw data in person at the Miller Research site (http://www.milleraudioresearch.com/avtech/index.html)? Are you for real?
I'm also at fault for not being intimately aware of
every single AVS-based discussion you've had with him and what exactly was discussed in each one? Isn't it in the
hundreds? Including some jitter related threads that are literally
thousands of posts long and the vast majority of them I've never seen?
Why are you prattling on about LPCM data when HDMI is the issue at hand?
I was trying to be helpful by mentioning that the HDMI data on the RXV3900 can be discovered by
opening the sub section of Miller's report called "LPCM Performance". That wasn't intuitively obvious to me, and I completely missed it at first, so I mentioned it for the benefit of others.
See the screen shot:
http://www.hydrogenaud.io/forums/index.php...st&p=906265 (http://www.hydrogenaud.io/forums/index.php?s=&showtopic=109824&view=findpost&p=906265)
[attachment=8392:Miller_R...for_HDMI.jpg]
I also mentioned the report appeared in Feb 2009, so you or anyone else could find it more quickly, on a site that to the best of my knowledge does not have all the many years of reports alphabetized, so in order to find this one I had to open every year manually and search each one of them individually. I hope you didn't take my mentioning which issue it appeared in as an insult, too.
I admitted that I had made a mistake after writing my post and had misread the Miller Reasearch page:
*CORRECTION: I see 96k/24 bit does get a "FAIL" for HDMI jitter, at the very bottom of the LPCM data for that Feb 09 test of the RXV3900. OOps.
but seeing as my lack of omniscience regarding
every single conversation you've had with him, over hundreds(?), with some of the jitter related threads (where he specifically discusses his jitter article, including the Yamaha RXV3900 data) having over
two three
thousand entries (http://www.avsforum.com/forum/91-audio-theory-setup-chat/1425262-audio-companies-all-involved-huge-conspiracy.html), I no longer wish to discuss it.
My friend's attitude is that jitter is omnipresent and always audible and smears the stereo image etc. (like the comment above), so although he exclusively listens to digital audio he is starting to think vinyl might be the way to go to get rid of the issue of jitter. Surface noise, pops, clicks, etc. from vinyl can be filtered out by our brains, whereas jitter is an omnipresent 'grating' and unpleasant sound.
Your friendly is sadly very misinformed. Jitter in digital audio does manifest itself as noise in the signal. But like you said, it's way below audible levels, unless the recording is done at a ridiculously low level. It can't do anything else, and certainly not "smear" anything at all. To claim that it is omnipresent and "grating" just shows that your friend has bought into the audiophile bullshit machine, big-time.
If your friend is so worried about jitter, he should be absolutely mortified at the amount of wow on vinyl records, even good ones. There can be a significant pitch difference as the record spins if the center hole isn't perfectly aligned. You mostly notice it on long sustained tones like organ music, you get a warbling or pitch rising and faling with the ~1.8 second period of the record's rotation.
Wow on a vinyl record is several orders of magnitudes beyond even very bad digital jitter, and the noise floor is significantly higher.
And that's before you get into other effects such as pre-echo (you can hear a faint copy of loud transients ~1.8 seconds before they hit, because the grooves slightly distort each other during the vinyl production process, or because the master tape has suffered from bleed-through. And of course, everything gets worse the closer you get to the center of the record, as there is less vinyl in each groove per revolution, so the dynamic range and noise floor suffers.
Really good points about the wow (& flutter) of vinyl. Thanks.
However, I think you might be wrong about your very last comment:
everything gets worse the closer you get to the center of the record, as there is less vinyl in each groove per revolution, so the dynamic range and noise floor suffers.
The rate of vinyl playback is constant, like a car going down a highway at 33 & 1/3 miles per hour. The rate of the needle across the groove doesn't slow down or speed up, therefore there isn't a loss of quality (similar to IPS, inches per second, of tape). Rather instead, it's more like the car (needle) is on a road which is curving more and more. The speed of the car over the pavement doesn't change. Similarly, in vinyl, the speed of the needle over the vinyl doesn't change either, so there's no change in medium quality. If there is, it's for different reasons at least. Of coure the needle comes to an end, but that's like a car coming to the end of a road. There's not less medium to represent the waveforms. it's 33 & 1/3 (or 45, or 78) all the way through. Guys, back me up on this.
Anyways, everything else you wrote seems agreeable.
EDIT: I just realised, you might have been talking about the wow/flutter/jitter getting worse at the center of the record. Is this what you meant? and if so, could you please elaborate on this relationship?
I'm not entirely understanding this.
There's not less medium to represent the waveforms. it's 33 & 1/3 (or 45, or 78) all the way through. Guys, back me up on this.
You're wrong there. It is not the angular velocity that matters here, but the tangential velocity. In other words, the closer you get to the center, the smaller the groove length per revolution.
everything gets worse the closer you get to the center of the record, as there is less vinyl in each groove per revolution, so the dynamic range and noise floor suffers.
The rate of vinyl playback is constant, like a car going down a highway at 33 & 1/3 miles per hour. The rate of the needle across the groove doesn't slow down or speed up, therefore there isn't a loss of quality (similar to IPS, inches per second, of tape). Rather instead, it's more like the car (needle) is on a road which is curving more and more. The speed of the car over the pavement doesn't change. Similarly, in vinyl, the speed of the needle over the vinyl doesn't change either, so there's no change in medium quality. If there is, it's for different reasons at least. Of coure the needle comes to an end, but that's like a car coming to the end of a road. There's not less medium to represent the waveforms. it's 33 & 1/3 (or 45, or 78) all the way through. Guys, back me up on this.
The speed of rotation is always the same and does not change between outer groove and inner groove, but the speed of the needle gets slower.
Assuming that the vinyl is has a diameter of 30cm and the label has a diameter of 10cm, the needle passes in the outer groove pi*30cm=94,25cm per circle and in the inner groove pi*10cm=31.42cm per circle. This means that the needle passes in the outer groove 3 times more groove per time-unit than in the inner groove.
Update:
I checked wikipedia. Vinyl with 33 1/3 circles per minute has a speed of the groove of 50cm/sec on the outer groove and 20cm/sec on the inner groove.
The rate of vinyl playback is constant, like a car going down a highway at 33 & 1/3 miles per hour.
An example of confusing linear velocity and rotational velocity which in fact are two very different things.
The speed of cars is given as a linear velocity which makes sense because they generally travel in straight lines.
The relationship between linear velocity and rotational velocity includes the radius. This is also known as CAV or constant angular velocity.
The linear speed of the stylus in the groove is therefore decreasing as the tone arm plays the recording. IOW, the linear speed is proportional to radius.
The rate of the needle across the groove doesn't slow down or speed up, therefore there isn't a loss of quality (similar to IPS, inches per second, of tape).
Completely and utterly false.
The rate of the needle across the groove slows down proportional to decreasing groove radius, and therefore there is a serious loss of quality.
This loss of quality is readily measured and it was always completely and obviously audible to me.
It was well known that the common presence of orchestral crescendos at the end of musical works was a source of challenge and frustration to vinyl mastering engineers.
In Ethan Winer's "AES Damn lies" video he mentions that his USB soundcard is spec'ed at 0.5 ns. So what an enormous difference! That's really a number that says something. I'm very glad I saw those numbers. That really helped a lot! I really liked the ending of the portion I quoted parts of above: "I won't
believe someone who says that he can detect time jitter of 100 ps or less in CDs."
Yes its worth pointing out just how large those values have to be before you notice them. To put that in context, if your 44.1kHz D/A is clocked off an ~10 MHz oscillator at 225x its clock rate, then 9216 ns of jitter would correspond to the 10 MHz oscillator having a bandwidth of about 4 MHz, which is basically impossible. It wouldn't even be oscillating at that point, and any digital logic was hooked up to would have long since crashed
The rate of the needle across the groove slows down proportional to decreasing groove radius, and therefore there is a serious loss of quality.
This loss of quality is readily measured and it was always completely and obviously audible to me.
I actually haven't played with vinyl records since my single digits. If you were to put the same recording at the start and end of one side, are there fairly audible speed/pitch differences?
I actually haven't played with vinyl records since my single digits. If you were to put the same recording at the start and end of one side, are there fairly audible speed/pitch differences?
No, the difference is not audible as either speed or pitch since the angular velocity is constant both in recording and playback.
The difference will be mostly distortion. Good mastering techs knew how to counteract this by subtly changing the EQ etc. towards the inner grooves and other tricks.
It can also be slightly counteracted by choosing a different alignment for the stylus on a standard radial-arm turntable. The most common "Baerwald" alignment tries to keep the tracking error and distortion as low as possible across all the grooves, but it does see a spike in distortion in the inner grooves. The alternative "Stevenson" alignment tries to counteract the inner groove distortion, at the cost of slightly higher distortion further out. Proponents of this alignment method claim that focusing the most accuracy on the inner grooves gives the most even distortion levels across the grooves, as the outer grooves are naturally less distorted.
But there's only so much you can do, with less material per second, the inner grooves will always have inferior sound to the outer grooves, given the same source material.
The rate of the needle across the groove slows down proportional to decreasing groove radius, and therefore there is a serious loss of quality.
This loss of quality is readily measured and it was always completely and obviously audible to me.
If you were to put the same recording at the start and end of one side, are there fairly audible speed/pitch differences?
No speed or pitch differences are expected or generally found.
That is not the nature of the difficulties that the decreasing linear speed causes.
The problem is a loss of treble extension and an increase in treble harshness. It can be audible, depending on both the playback gear and the recording.
For example, a LP recording that is quiet and lacks treble is naturally not going to lose treble extension and clarity as profoundly as one that is loud and bright.
No matter the quality and state of perfection of the playback gear, these problems are there. The degree varies, but across a reasonable selection of recordings, critical listeners are going to hear it at least some of the time.
I've also read vinyl cutting engineers say that when vinyl was the dominant media, often the more mellow songs would be put at the end of each side, as they would be less affected by distortion. They also mentioned that nowadays, they often get recordings with powerful songs at the end of each side, which caused problems for them.
Two common records you might have or can get a hold of that would show this inner-groove distortion would be "Rubber Soul" by The Beatles and "Shotting rubberbands at the stars" by Edie Brickell & New Bohemians - at least the European copies I have, and the Beatles one has the most audible distortion.
That must have been a pretty high quality vinyl rip if the CD version didn't sound any better. Background hiss should be pretty obvious in almost any quiet or silent passage. Audible differences absolutely exist in cartridges, turntables, etc. I think the takeaway from this thread is that a <$100 digital source will almost certainly have higher fidelity than the most expensive record player.
Casting aside JJ's "subjective preference is inviolate" clause, the only valid objective grounds on which one can base this otherwise vinyl equals or trumps CD nonsense is increased dynamic range of old vinyl over newly remastered CD (IOW, apples vs. oranges). More often than not, it seems claims that new vinyl is from a more dynamic master than the CD counterpart are unable to hold water when placed under scrutiny.
Please check your bogus TTDR measurements (http://www.hydrogenaud.io/forums/index.php?showtopic=102895) at the door.
There's nothing that prevents modern dynamic-range compression from being pressed to vinyl, right? I'd like to give the benefit of the doubt and think that complaints about DRC form the basis of this whole hipster vinyl mythology. For myself, the points made in this thread have dissuaded me from even dabbling with a turntable.
Could you explain this in a little more detail? I'm not understanding why HDMI would influence jitter of the downstream system. HDMI is a packet-based digital protocol, how does the rate at which packets arrive influence the analog performance of the device?
The rate at which the packets arrive is not the only relevant variable.
For example SP/DIF packets arrive in a fairly continuous stream while HDMI packets generally arrive with big time delays between them.
Audio is a fairly small percentage of a HDMI stream and the packets are relatively sparse.
Unless the DAC is actually clocked off of the HDMI clock (which would be tough since a lot of different HDMI clocks are possible), the DAC's jitter shouldn't depend on if you fed it with HDMI or any other input. Just like how playing Spotify vs. iTunes doesn't change the jitter in your sound card.
You're right - the jitter that was observed with some AVRs and HDMI inputs should not have happened. It is probably a classic case of the squeaky gear. I seem to recall comparing Miller's jitter numbers to the standards in:
Eric Benjamin and Benjamin Gannon, "Theoretical
and Audible Effects of Jitter on Digital Audio
Quality", AES Preprint 4826, presented at the AES
105th Convention, San Francisco, September 1998.
and Miller's data was too incomplete for a comparison.
I'm under the impression that by the time your audio card sees the audio data stream from these services, it has been cleaned up quite a bit.
Most DAC chips are fairly susceptible to jitter all by themselves, but the buffering that drives them (such as in CD players) has generally been pretty good.
Could you explain this in a little more detail? I'm not understanding why HDMI would influence jitter of the downstream system. HDMI is a packet-based digital protocol, how does the rate at which packets arrive influence the analog performance of the device?
The rate at which the packets arrive is not the only relevant variable.
For example SP/DIF packets arrive in a fairly continuous stream while HDMI packets generally arrive with big time delays between them.
Audio is a fairly small percentage of a HDMI stream and the packets are relatively sparse.
But it shouldn't be relevant
at all. It should not matter how sparse the audio packets are or aren't, or how the stream works. Jitter is just about how well an analog waveform representing an ideal sampling clock approximates that ideal clock. The data that is ultimately encoded onto that clock doesn't change anything, which is why the HDMI point is so strange. Something else must be going on.
Unless the DAC is actually clocked off of the HDMI clock (which would be tough since a lot of different HDMI clocks are possible), the DAC's jitter shouldn't depend on if you fed it with HDMI or any other input. Just like how playing Spotify vs. iTunes doesn't change the jitter in your sound card.
You're right - the jitter that was observed with some AVRs and HDMI inputs should not have happened. It is probably a classic case of the squeaky gear. I seem to recall comparing Miller's jitter numbers to the standards in:
Eric Benjamin and Benjamin Gannon, "Theoretical
and Audible Effects of Jitter on Digital Audio
Quality", AES Preprint 4826, presented at the AES
105th Convention, San Francisco, September 1998.
and Miller's data was too incomplete for a comparison.
I'm under the impression that by the time your audio card sees the audio data stream from these services, it has been cleaned up quite a bit.
Most DAC chips are fairly susceptible to jitter all by themselves, but the buffering that drives them (such as in CD players) has generally been pretty good.
Yeah something more is happening. You have to get HDMI data to a DAC at a rate that matches the DAC's clock on average. There are two different ways you could do that, use your own clock which you speed up and slow down slightly so that its average sampling rate matches the HDMI's packet rate or run the DAC async from the rest of the receiver off of a clock you derive from the HDMI. If you do the former, the pitch wobbles some tiny, tiny amount at a small fraction of a Hz but you don't care about the HDMI clock. I doubt this is done much, and the wobble is probably much too low frequency to be detected. If you do the latter, the obvious way to do this would be to take the HDMI clock and feed it into a PLL which would generate the clock you wanted while stabilizing it, making the HDMI clock irrelevant as well.
If this problem really only happened over HDMI (as opposed to merely having a broken DAC), I'd guess they did something crazy like try to divide down the ~160 MHz HDMI clock to 48k without a PLL. That would certainly give you some jitter. It would probably also give you a pitch error since I doubt the HDMI clock is always an integer multiple of 48,000 . . .
Yeah something more is happening. You have to get HDMI data to a DAC at a rate that matches the DAC's clock on average. There are two different ways you could do that, use your own clock which you speed up and slow down slightly so that its average sampling rate matches the HDMI's packet rate or run the DAC async from the rest of the receiver off of a clock you derive from the HDMI. If you do the former, the pitch wobbles some tiny, tiny amount at a small fraction of a Hz but you don't care about the HDMI clock. I doubt this is done much, and the wobble is probably much too low frequency to be detected. If you do the latter, the obvious way to do this would be to take the HDMI clock and feed it into a PLL which would generate the clock you wanted while stabilizing it, making the HDMI clock irrelevant as well.
If this problem really only happened over HDMI (as opposed to merely having a broken DAC), I'd guess they did something crazy like try to divide down the ~160 MHz HDMI clock to 48k without a PLL. That would certainly give you some jitter. It would probably also give you a pitch error since I doubt the HDMI clock is always an integer multiple of 48,000 . . .
The HDMI spec contains a chapter that describes the clock regeneration in a receiver. While an implementation may choose any method it sees fit, the overall process is pretty clear and straightforward. However, there are two modes, depending on whether the video and audio clocks are coherent or not. You can't meaningfully discuss the situation when you don't know which mode is active. Furthermore, the signal source has a lot of influence on the process because it determines the mode and supplies the division/multiplication factors to the receiving device. It makes it harder to determine the guilty party when the result turns out to be wanting.
There is no need for the resulting clock to be jittery. If both sender and receiver are working well, the resulting clock can be just fine. People seem to extrapolate the problems some devices have to the HDMI interface in general.
People seem to extrapolate the problems some devices have to the HDMI interface in general.
Are there any recorded samples that are readily available which can be used to demonstrate an audible problem with poorly designed devices?
Are there any recorded samples that are readily available which can be used to demonstrate an audible problem with poorly designed devices?
I didn't want to imply they are audible. Some people seem to have measured poor jitter performance with some HDMI setups. I can't say how serious they are in practice. I'm also sceptical regarding the ability of some people to troubleshoot their measurements.
It would be necessary to have a proper investigation into the source of the poor jitter measurements some have quoted, in order to find out what went wrong. It might have been poor design of the unit under test, but that is not the only possible explanation.
I didn't want to imply they are audible.
I'm sorry to imply that I thought you did. I guess I honed in on the word problem and wondered if it was more than just not in keeping with good engineering practice.
Yeah something more is happening. You have to get HDMI data to a DAC at a rate that matches the DAC's clock on average. There are two different ways you could do that, use your own clock which you speed up and slow down slightly so that its average sampling rate matches the HDMI's packet rate or run the DAC async from the rest of the receiver off of a clock you derive from the HDMI. If you do the former, the pitch wobbles some tiny, tiny amount at a small fraction of a Hz but you don't care about the HDMI clock. I doubt this is done much, and the wobble is probably much too low frequency to be detected. If you do the latter, the obvious way to do this would be to take the HDMI clock and feed it into a PLL which would generate the clock you wanted while stabilizing it, making the HDMI clock irrelevant as well.
If this problem really only happened over HDMI (as opposed to merely having a broken DAC), I'd guess they did something crazy like try to divide down the ~160 MHz HDMI clock to 48k without a PLL. That would certainly give you some jitter. It would probably also give you a pitch error since I doubt the HDMI clock is always an integer multiple of 48,000 . . .
The HDMI spec contains a chapter that describes the clock regeneration in a receiver. While an implementation may choose any method it sees fit, the overall process is pretty clear and straightforward. However, there are two modes, depending on whether the video and audio clocks are coherent or not.
Ha, looking at the spec, the two recommended ways of dealing with this problem are exactly the two solutions I described above. Not a bad guess. But it also suggests that the HDMI data clock and audio clock may be set to be integer multiples, so I suppose those jittery devices did not use a PLL and instead simply divided down the data clock. Not great design.
Yeah something more is happening. You have to get HDMI data to a DAC at a rate that matches the DAC's clock on average. There are two different ways you could do that, use your own clock which you speed up and slow down slightly so that its average sampling rate matches the HDMI's packet rate or run the DAC async from the rest of the receiver off of a clock you derive from the HDMI. If you do the former, the pitch wobbles some tiny, tiny amount at a small fraction of a Hz but you don't care about the HDMI clock. I doubt this is done much, and the wobble is probably much too low frequency to be detected. If you do the latter, the obvious way to do this would be to take the HDMI clock and feed it into a PLL which would generate the clock you wanted while stabilizing it, making the HDMI clock irrelevant as well.
If this problem really only happened over HDMI (as opposed to merely having a broken DAC), I'd guess they did something crazy like try to divide down the ~160 MHz HDMI clock to 48k without a PLL. That would certainly give you some jitter. It would probably also give you a pitch error since I doubt the HDMI clock is always an integer multiple of 48,000 . . .
The HDMI spec contains a chapter that describes the clock regeneration in a receiver. While an implementation may choose any method it sees fit, the overall process is pretty clear and straightforward. However, there are two modes, depending on whether the video and audio clocks are coherent or not. You can't meaningfully discuss the situation when you don't know which mode is active. Furthermore, the signal source has a lot of influence on the process because it determines the mode and supplies the division/multiplication factors to the receiving device. It makes it harder to determine the guilty party when the result turns out to be wanting.
There is no need for the resulting clock to be jittery. If both sender and receiver are working well, the resulting clock can be just fine. People seem to extrapolate the problems some devices have to the HDMI interface in general.
Agreed that there is no need for the resulting clock to be jittery. One of the hallmarks of digital data transmission that as long as the error rates are not too high, the digital data can be recovered as accurately as desired at the receiving end. It is just a matter of properly processing the data properly before transmission, and again after reception. If you have to add parity or CRCs or whatever, so what? If you have to reconstruct, buffer, reclock, whatever, so what? Digital processing and data transmission are cheap and getting cheaper. If it is not practical this week, wait a while.
There are no generally accepted standards (that I know of) for what constitutes audible jitter. Of course there are no standards (that I know of) for what constitutes audible THD, IM, or FR and phase differences either. Therefore whenever someone says there is too much jitter (or THD, IM, FR or phase changes), it would appear to lack general meaning. That leaves the door open for people to say whatever they will, for whatever reason or no reason.
I hope I put this in the right forum. You've probably been asked this a million times, but I trawled the web (mostly searching "dan lavry" + "jitter", as I figured he was the "master") as well as his forum, and I couldn't really find my answer, so I asked on his forum (which gave this response: http://www.lavryengineering.com/lavry_foru...f=1&t=7449) (http://www.lavryengineering.com/lavry_forum/viewtopic.php?f=1&t=7449)), and I'll also ask here to hear your opinion:
If we're talking home audio listening (not recording), is jitter then audible, and how does it sound? I'm only talking about playing CDs and digital audio files from either a CD player, or your computer or a streamer through a dac into your amplifier.
Let's also for the time being assume that the analogue to digital conversion during recording/mastering was done well to begin with, and let's assume that the converter you use is of fairly good quality.
I discussed this with an audiophile friend. My own impression is that with 16 bit audio, jitter starts to become audible at -96 dB (without dithering), so with music recorded at sensible levels, it will not be audible, as all jitter is in the noise floor, and the music would mask the jitter. An exception could of course be orchestral music with a huge dynamic range recorded at a very low volume (so the noise floor would be raised). As for how jitter sounds, the video "Digital show & tell" on Xiph.org says quantization noise sounds like tape hiss from analogue tapes, but I might be mixing up quantization noise with jitter (is it the same?). On that website, there's also a file available for download with a 1 kHz tone at -105 dB. When playing that file there's background noise, which I assumed was jitter. In other words, jitter is present in all digital audio, but the amount is so low that you can never hear it except for in those extreme cases mentioned above. I asked Ethan Winer who said:
"Jitter manifests as noise 100+ dB below the music, and is never audible. Nor does it create 'a lack of depth, solidity and a smearing of the stereo image.' You’re thinking of wow and flutter. :->)"
In Ethan Winer's Youtube video "AES Damn lies workshop", he shows the following picture:
(http://s3.postimg.org/rqukd9xqb/AES_Damn_Lies_Jitter.png)
My friend's attitude is that jitter is omnipresent and always audible and smears the stereo image etc. (like the comment above), so although he exclusively listens to digital audio he is starting to think vinyl might be the way to go to get rid of the issue of jitter. Surface noise, pops, clicks, etc. from vinyl can be filtered out by our brains, whereas jitter is an omnipresent 'grating' and unpleasant sound.
I of course understand that when creating converters like Dan Lavry does, it's important to minimise jitter as much as possible in order to come closer to creating the perfect product, as well as creating an A/D converter that will have minimal jitter so the recording artists can raise and lower levels on different tracks as much as they like. But as mentioned, I'm only interested in audibility and listening at home – not in the technical aspect (measuring, etc.).
Marantz claims to have Jitter reducing technology. Now I don't think a company as large and as well regarded as Marantz would include a technology in their equipment that didn't have real value ... I trust that their engineers know exactly what they're doing and that reducing jitter matters. I don't think Marantz would spend money on a "non issue".
Marantz claims to have Jitter reducing technology. Now I don't think a company as large and as well regarded as Marantz would include a technology in their equipment that didn't have real value ... I trust that their engineers know exactly what they're doing and that reducing jitter matters. I don't think Marantz would spend money on a "non issue".
Whether jitter is audible or not, it is perceived as negative to have it in the audio signal. If it can be measurably reduced for small cost then I consider a commercial manufacturer would do it if they could. Of course there may be many variables and factors for a manufacturer to consider in researching and implementing jitter reduction. I just mentioned the most obvious 2 factors above.
Marantz claims to have Jitter reducing technology. Now I don't think a company as large and as well regarded as Marantz would include a technology in their equipment that didn't have real value ... I trust that their engineers know exactly what they're doing and that reducing jitter matters. I don't think Marantz would spend money on a "non issue".
"Audiophiles" claim jitter is a problem. Manufacturer wants to sell products to "audiophiles". Manufacturer introduces technology that doesn't really do very much at all and makes no difference to what can be heard. But "audiophiles" see "jitter reduction" and spend lots of money. Manufacturer is happy.
Or....you don't really understand how companies sell products to consumers do you?
Ironically a $6000 McIntosh is more jittery than a Sony PlayStation 1
https://www.stereophile.com/content/case-jitters-less-cd-quality
So high jitter is only for rich people, not poor guys like the rest of us :D
Nice argument from authority (https://en.m.wikipedia.org/wiki/Argument_from_authority), though.
I trust that their engineers know exactly what they're doing and that reducing jitter matters. I don't think Marantz would spend money on a "non issue".
Sorry, can't engineer away a mental health issue. Swing and a miss. Wrong authority to appeal.
Btw, it isn't a non-issue, the effect on audiophiles is real. Just not of DUT origin.
Or....you don't really understand how companies sell products to consumers do you?
+1
How do we know the
engineers thought reducing jitter matters, as opposed to taking marching orders from sales/marketing?
Are we also supposed to trust the "engineers" at Sony?
https://www.sony.com/electronics/hi-res-audio
Jitter can be a problem and reducing it always is good but there is a point of dismissing returns and I always found that most of the time the people is clueless about it and think in it a magical way, it effects are not the same in all situations but people think the contrary. Even depending of the technology the effects can vary, as in CD player with a clock signal to the DAC with 1us jitter can have a disastrous effects if it uses a 1bit DAC but at the same time be unnoticeable if it uses a R2R DAC.
The only time that I was able to hear jitter noise in a CD player was in an +20 years old one from a friend (actually his father), I reworked it and the noise and jitter disappeared (caps where bad, some diodes and transistors where also bad an changed the crystal oscillator module as it drifted to much).
Samples and ABX logs indicating a difference was heard or your post will be binned per rule #8 in our terms of service.
Anecdotes won't cut it.
No expert here, but if someone wants to describe how jitter sounds, wouldn't it be very similar to a Doppler Effect. But It would be a Doppler Effect that changes many times per second in very small acceleration changes.
I don't know much about jitter in audio, but I do know about jitter in networking. So I'm extrapolating the concept. In other words, it's my guess.
For Analog sources, jitter would be due to the change of speed of the media during playback. Not so much the overall speed, but tiny acceleration or declaration of the playback speed. Or instability of the playback speed. I would expect this to be inevitable but at the same time mostly negligible. It's part of the nature of electric motors and belt drives, etc.
For digital sources then you have the instability of the clock. Which will depend on the quality of the oscillator. But then again when you have 44,100 samples per second, can you tell which of those samples came a little faster or slower?
An analogy would be the second hand on your analog watch, or second display on a digital watch. Could you tell which seconds in a minute were actually faster or slower than average. Because of the nature of the oscillator and general quality of the watch, all those ticks will not be exactly the same, but over the span of a minute they are expected to keep within a margin of error.
The only way you could probably hear jitter on a real world scenario would be an Analog system which is pretty much failing dew to mechanical flaws. It could be experiencing excessive mechanical resistance. It could be worn out, Or it could be affected by a power problem like low voltage. So you could probably listen it it slow down for longer periods of time and then go back to normal.
On a computer you could probably listen to some sort of jitter if the computer hangs momentarily. Then you might listen to the audio slow down or pause, and then try to catch up. If there is a slow down or catch up it would be jitter. If it just skips during the problem, then you wouldn't actually be able to hear the jitter it would just be a skip.
EDIT:
This is an interesting concept:
"Headbanging jitter.... Someone had jokingly mentioned that jitter that comes from the doppler effect, if you move your head, is probably higher than jitter introduced by the devices themselves."
http://www.sereneaudio.com/blog/headbanging-jitter
Even depending of the technology the effects can vary, as in CD player with a clock signal to the DAC with 1us jitter can have a disastrous effects if it uses a 1bit DAC but at the same time be unnoticeable if it uses a R2R DAC.
1us of jitter is such a lot that it takes gross incompetence, or even sabotage, to arrive at such figures.
So yes, if you are dealing with this much jitter, I am happy to believe stories that it is audible. But I would discard or repair any piece of equipment that shows this much jitter. I haven't seen anything like that in a device that wasn't defective. And I have seen audiophile devices that were so incompetently designed they produced systematic jitter by design, while at the same time claiming to reduce jitter. Even there, there wasn't this much jitter.
The only time that I was able to hear jitter noise in a CD player was in an +20 years old one from a friend (actually his father), I reworked it and the noise and jitter disappeared (caps where bad, some diodes and transistors where also bad an changed the crystal oscillator module as it drifted to much).
Well, that looks like a defective device to me. Yet, I'm not sure how you determined that jitter was the problem. Have you got sufficient measurement equipment to find out what exactly was wrong?
No expert here, but if someone wants to describe how jitter sounds, wouldn't it be very similar to a Doppler Effect. But It would be a Doppler Effect that changes many times per second in very small acceleration changes.
You are right in principle, but it is more useful to speak about it in terms of modulation. Jitter effectively causes phase modulation of the audio signal. This means that the actual effect depends a lot on the nature of the modulating signal, i.e. the jitter signal. Your doppler effect analogy only holds for very low jitter frequencies, i.e. frequencies a few cycles per second at the most. That isn't typical, and you would need enormous amounts of such jitter to produce an effect that is even faintly like doppler.
Normally, the term "jitter" refers to modulating frequencies of 10 Hz or more. In other words, the jitter itself is an audio signal. The term that is used for frequences below 10 Hz is "wander". A doppler effect is wander with an enormous amplitude, so that you actually hear the pitch change. If something like that happens in an audio system, something is very seriously wrong. Like a mechanical problem in an analog tape machine, or a excentric hole in an LP.
I don't know much about jitter in audio, but I do know about jitter in networking. So I'm extrapolating the concept. In other words, it's my guess.
It is dangerous and misleading to extrapolate this. Jitter that causes modulation is sampling jitter, and it is associated with the sampling clock that drives a converter (ADC or DAC). Jitter in networking is associated with packet transmission, which hasn't got anything to do with the sampling clock.
The only link is the fact that there is variation in timing, in the most general sense. Don't draw any further conclusions from it.
For Analog sources, jitter would be due to the change of speed of the media during playback. Not so much the overall speed, but tiny acceleration or declaration of the playback speed. Or instability of the playback speed. I would expect this to be inevitable but at the same time mostly negligible. It's part of the nature of electric motors and belt drives, etc.
It is far from negligible. Considerable effort was spent in getting this under control. It is very easy, for example, for mechanical oscillations to build up when tape glides over a pin. Technical term: flutter. Very common. It is actually hard to misdesign a digital device to exhibit such amounts of phase modulation as analog devices frequently do.
For digital sources then you have the instability of the clock. Which will depend on the quality of the oscillator. But then again when you have 44,100 samples per second, can you tell which of those samples came a little faster or slower?
No, you can't tell from the individual samples. Heck, you don't listen to samples, anyway. You listen to a reconstructed, continuous signal. If anything, you hear a modulation effect. And that's unlikely, because it is quite straightforward to build a good crystal oscillator that has orders of magnitude less jitter than necessary. Much better than anything mechanically analog. If you design digital audio gear and don't know how to do that, you're incompetent. That's not snobbery, that's reality.
The only way you could probably hear jitter on a real world scenario would be an Analog system which is pretty much failing dew to mechanical flaws. It could be experiencing excessive mechanical resistance. It could be worn out, Or it could be affected by a power problem like low voltage. So you could probably listen it it slow down for longer periods of time and then go back to normal.
It is a bit more common than that, unfortunately. But people have figured out how to deal with it mechanically. Even service people had wow&flutter meters to measure their analog gear, and see if there was a problem. With digital gear those meters are useless. They just don't have a jitter problem that would fall into the range these devices can measure. For digital, you need measurement gear that is more sophisticated than an analog wow&flutter meter.
On a computer you could probably listen to some sort of jitter if the computer hangs momentarily. Then you might listen to the audio slow down or pause, and then try to catch up. If there is a slow down or catch up it would be jitter. If it just skips during the problem, then you wouldn't actually be able to hear the jitter it would just be a skip.
No, because the relevant oscillator carries on. If the software can"t keep up with it, samples are duplicated or omitted. The effect isn't jitter, but short dropouts.
"Headbanging jitter.... Someone had jokingly mentioned that jitter that comes from the doppler effect, if you move your head, is probably higher than jitter introduced by the devices themselves."
It is most definitely higher, by many orders of magnitude, but the modulation frequency is quite low.
[A] CD player with a clock signal to the DAC with 1us jitter can have a disastrous effects
I looked up my CD player, a Naim CD5X, on Stereophile's measurement page: It has 241 picoseconds of jitter. Other CD players, both cheaper and more expensive, show jitter figures around the same amount, give or take a few hundred picoseconds (well, give, since jitter amounts in any converter don't go down to 41 picoseconds as far as I know).
One nanosecond is 1000 picoseconds. One microsecond (μs) is 1000 nanoseconds. That would be a hell of a lot of jitter in such a CD player! I'm with Pelmazo on this one - the unit must be defective.
The McIntosh MS750, which is one of the worst units that Stereophile has ever measured for jitter, had 14 nanoseconds of jitter.
The $43,440 Zanden 5000 Mk.IV, which the Stereophile editor called, and I quote, "the worst-measuring digital product [he had] encountered", since everything about it was poorly designed, "only" measured 1018 picoseconds of jitter. But of course, CD-hater extraordinaire Michael Fremer called it the best CD player he had ever heard (and reviewers for other publications were ecstatic as well)!