Frequncy standard use in audio equipment

[tkamiya]

I'm asking because I'm genuinely curious....

I have been seeing reference oscillators of all kinds, including Rb, OCXO, and GPSDO for audo use.  Some report says they can "clearly hear the difference."  I'm thinking for syncing digital recording, etc, regular crystal or maybe TCXO is sufficient and that human ear can't possibly distinguish the difference in phase and frequency stability. 

Is this yet another case of snake oil?  Some of those goes for $4KUS or so. 

[james_s]

It's no different than fancy expensive speaker cables and other audiophool nonsense. The placebo effect is real, if you believe some gadget will make your audio sound better then it will sound better to you. A reference oscillator would have to be REALLY bad to have any sort of audible effect though, and it's unlikely the oscillators in the gear used to record and master the audio were anything exotic either.

[Gyro]

Master clocks and distribution are used in recording studios for equipment synchronisation I believe. I don't know to what accuracy though.

For consumer audio, the reference oscillators you mentioned are total snake oil. What does matter is noise and jitter. Better quality gear tends to have better internal clock oscillators - starting from the basic on-chip inverter oscillator through to better quality canned ones and those using discrete components. These (all things being equal) do seem to improve sound quality - Different DACs have different jitter rejection, likewise clock recovery in things like S-PDIF receiver ICs.

Absolute frequency accuracy is totally inaudible though, even with the cheapest and most basic crystal oscillator.

[ataradov]

Clearly hearing the difference is not enough. Can you measure it with measurement equipment? If not, then you are full on audiophool. 

And for most audio application $1 crystal is sufficient.

[SiliconWizard]

Master clocks and distribution are used in recording studios for equipment synchronisation I believe. I don't know to what accuracy though.

Yes, but synchronizing several digital sources and the frequency stability of a single source are two completely different things.

The typical use case for an end-user is just a single source. So there's usually nothing to synchronize.

Of course, as said above, for a single source, jitter matters much more than absolute frequency. We'd need to know if those shit expensive oscillators have at least much better jitter figures than modest ones. We can suppose so, but that's really the main point to consider.

Now whether the typical jitter of even a modest crystal-based oscillator (as long as it's not pure crap) has any measurable impact on distortion (harmonic/phase), that remains to be seen. That may be measurable if you directly measure the ouput of a DAC with a very clean output stage, but once it goes through some kind of amplifier and speakers (or headphones), all bets are almost certainly off. The latter will have a much more severe effect on phase and harmonic distortion than the digital clock.

And, now even when it's actually measurable, which as said above is likely not the case with a typical audio system, whether it can actually be "heard" by our normal ears, it's yet another question. The most sensible answer to this is probably, not at all.

But as we say on a regular basis, as long as said equipment has measurable technical benefits, even if it makes no difference to the ears, then it's just a matter of luxury IMO. If you want luxury equipment, that's your call. Now OTOH, if there is absolutely NO measurable technical difference, then it's snake oil, and a different category, although certainly some people won't be able to tell the difference. But I like the "luxury" argument. If you're buying a $20k watch made with very good engineering and great materials, that may not give you much more accurate time than a $20 watch (or so), but it IS luxury. Now if you're buying a $20k that is actually really a rebranded $20, that is just rip-off.

[tkamiya]

Even the jitter though, regular plain crystal oscillator has decent figure.  Can one actually hear the difference if say 32K fork quartz (the watch kind) or whatever they use in cheap gear was replaced with good quality one?  (or vise versa...  replace good one with bad?)

If jitter was the problem, certain Rb units won't be usable.  I can even see the jitter using simple oscilloscope.

[Gyro]

For audio gear, you tend to be talking at least 11.2896MHz, 16.9344MHz, 27MHz (AV), 48MHz or faster depending on sample rate, not down at 32kHz. At those sort of frequencies, it's more down to the oscillator design and decent RF layout you'd never hear the difference between individual crystals (unless faulty).

I mentioned above "(all things being equal)". Of course they never are. Better gear tends to have more attention to PCB layout and routing, component quality, grounding, PSU noise etc, in addition to 'better clock circuits.

Yes, some of the snake oil alternatives are clearly inferior in the parameters that matter.

[tkamiya]

So....  having bad crystals doesn't turn Waltz into Polka? 

[ataradov]

Jitter of the regular crystal oscillators should be perfectly acceptable. Again, as soon as a person does not accept double blind test, they can be dismissed as audiophool. Let them pay $4K for a clock source and $8K for a mains cable. Suckers need to be milked.

[tkamiya]

Master clocks and distribution are used in recording studios for equipment synchronisation I believe. I don't know to what accuracy though.

But as we say on a regular basis, as long as said equipment has measurable technical benefits, even if it makes no difference to the ears, then it's just a matter of luxury IMO. If you want luxury equipment, that's your call. Now OTOH, if there is absolutely NO measurable technical difference, then it's snake oil, and a different category, although certainly some people won't be able to tell the difference. But I like the "luxury" argument. If you're buying a $20k watch made with very good engineering and great materials, that may not give you much more accurate time than a $20 watch (or so), but it IS luxury. Now if you're buying a $20k that is actually really a rebranded $20, that is just rip-off.

I used to collect watches.  One of the things about luxury watches is that they'll always have some value.  Most will depreciate but some appreciate.  Like you say, cheap OCXO housed in a fancy box is probably worth nothing.  Unless of course it gets sold to another like minded people.

[penfold]

Was it not the introduction of dither on the sampling clock which made CD audio viable? Or am I wrong on that?
I seem to recall that the measurable sampling noise in audio-type signal digitisation is worse without dither... and a little drift never hurt anyone.

As an poignant point on audio-pragmatism, it is currently lashing down with rain outside with constant rumbles of thunder, I can barely tell whether my CD player is switched on, let alone what type of oscillator it contains.

[Bassman59]

Was it not the introduction of dither on the sampling clock which made CD audio viable? Or am I wrong on that?

No. It was better mastering for the medium and better converters.

Dither is used when taking, say, 24-bit words down to 16 bits, instead of simply truncating.

[SiliconWizard]

Was it not the introduction of dither on the sampling clock which made CD audio viable? Or am I wrong on that?

No. It was better mastering for the medium and better converters.

Dither is used when taking, say, 24-bit words down to 16 bits, instead of simply truncating.

Yep. And this is what needs to be done when mastering CDs these days, as most recording is now 24-bit. Of course, a lot more than this needs to be done, such as pretty "drastic" compression, so that the end result can still be listened to with small speakers/crap headphones.

[vk6zgo]

Human hearing is remarkably forgiving.
For years, movie soundtracks recorded at 24 f.p.s. film rates were replayed on TV at 25 f.p.s., shifting all audio frequencies by around 4.2%.
Did anyone notice?

[SiliconWizard]

Human hearing is remarkably forgiving.
For years, movie soundtracks recorded at 24 f.p.s. film rates were replayed on TV at 25 f.p.s., shifting all audio frequencies by around 4.2%.
Did anyone notice?

For that much shift, you would definitely notice when there is music and you have "absolute pitch". Which is a pretty rare thing. Otherwise, nope.

Speaking of music, even without absolute pitch, if you're playing along some music with some instrument, then you would definitely get very annoyed by such a large shift. But even a basic crystal oscillator gets you around 100 ppm or so... so, nah. You'd have to be very good to hear a difference with a 100ppm frequency shift.

[james_s]

Human hearing is remarkably forgiving.
For years, movie soundtracks recorded at 24 f.p.s. film rates were replayed on TV at 25 f.p.s., shifting all audio frequencies by around 4.2%.
Did anyone notice?

Today a lot of reruns of older TV shows are sped up slightly, they do that in addition to editing out bits and pieces in order to squeeze in more commercials. It's annoying and noticeable if you've seen the show before.

[artag]

Today's methods of changing speed do it without a pitch shift, so may be less obvious.

[Haenk]

Jitter of the regular crystal oscillators should be perfectly acceptable. Again, as soon as a person does not accept double blind test, they can be dismissed as audiophool. Let them pay $4K for a clock source and $8K for a mains cable. Suckers need to be milked.

There might arise issues with separated source/DAC, when transfered data is mixed together (via SPDIF eg.). Never experienced this myself, but could happen (according to "others"). Not sure, if that really leads to serious sound issues.
This should never happen when the data/clock is directly connected via a short distance (like in a CD player).

However one should still remember the (very early) method of channel switching in the very early days, when one single channel DAC like the 1540 was constantly switching left/right channels. That's probably where all the badmouthing about awful CD sound quality came from. Because it was an awful method

[coppice]

Jitter of the regular crystal oscillators should be perfectly acceptable.

There are certainly some crystal oscillators whose jitter is sufficient to degrade audio, but it doesn't require anything particular fancy to be a good clock source.

At 20kHz a 1 bit error in 16 bits is equivalent to about 500ps of jitter, so it takes a pretty nasty crystal oscillator to affect things there. However, at 20kHz a 1 bit error in 20 bits (more relevant to recording than to replay) is equivalent to about 30ps of jitter. A lot of crystal oscillators give that. Maybe you aren't too concerned about the errors (or the noise they manifest as) at such high frequencies, but to be really clean you are looking for jitter levels in the ballpark of what a good crystal oscillator gives.

[PlainName]

Can you measure it with measurement equipment? If not, then you are full on audiophool. 

You need to know what to measure before you can say that not reading anything is meaningful.

Sometimes it's not obvious. The brain can use the minute difference in arrival time of sound at each ear to determine direction. If you just looked at that time interval in isolation you might well consider it pretty much irrelevant compared to the frequency range of the human ear.

And don't forget eyes: movies at 25fps are fine, but gaming screens talk about 120Hz. But there are frequencies where something can flicker and we can't tell it is flickering, but if it moves it's very obvious it is (roadwork beacons and LED rear lights on older cars are good examples). What would you measure there? In fact, it took a long time before the problem was accepted to be real, so would you have even bothered to think about measuring anything?

We could go on and note that making an object flicker at a specific frequency can change the perceived colour of it. If someone said film displaying at that frequency was 'more colourful' would you write them off a videophools?

I am not suggesting that any of the stuff laughed at here is pukka in any way, but I am suggesting that a knee-jerk thoughtless dismissal is unwarranted.

[ataradov]

For stuff having to do with human perception double blind test is a way to go. But that is usually denied even more vigorously by those types of people. They know they are full of it.

[coppice]

For stuff having to do with human perception double blind test is a way to go. But that is usually denied even more vigorously by those types of people. They know they are full of it.

Double blind testing is very important. However, it takes a lot of testing to reach a meaningful conclusion. It takes a lot of different material to flush out every audible quirk a system may have. If you ever worked on lossy compression you will know this VERY well.

[Dubbie]

When I was into audio as a teenager, I ran a blind test on myself and was absolutely shocked at how high the THD could get before I could reliably detect any distortion. After that exercise, I realised that most audio "tweaks" were a waste of time.

If you have enough clean amp power and decent speakers and room, tweaks up the chain such as DACs etc, were generally a waste of time for me.

[ataradov]

Double blind testing is very important. However, it takes a lot of testing to reach a meaningful conclusion.

"clearly hear the difference"  should not take that long. If the differences are so minor that it takes forever to detect them reliably, then is the upgrade really worth the money?

[Circlotron]

One way to avoid all this clock jitter stuff is to only listen to obviously "superior" vinyl...








LOL.