How does the stereo image in digital audio shrink?

[alcohol]

Really, how can that be? If I pan two distinct sounds in Pro Tools, and then monitor them separately, the other signal is not heard.

 

So how does the stereo image or field shrink?

 

Tape and analog consoles can have channel crossover problems that might shrink the stereo image.

 

Is it because digital isn't producing a phase distortion, which might gives the illusion of a wider field, that the analog recorder might be producing?

[MikeRivers]

This is one of the effects that's often attributed to clock jitter. The phase relationship of the two channels wanders a bit, and at a fast enough rate so that you don't actually hear the stereo image move, but it does blur slightly.

 

Or it could be just because it's digital.

[Rudolf von Hagenwil]

Every recording technique follows different rules for panning, some can not be panned at all.

 

Shrinks in what parameter? What is the panning? What is the stereo microphone technique?

[blue2blue]

 

This is one of the effects that's often attributed to clock jitter. The phase relationship of the two channels wanders a bit, and at a fast enough rate so that you don't actually hear the stereo image move, but it does blur slightly. Or it could be just because it's digital.

 

 

Is that the going thinking why some folks prefer external clocking (which tends to increase jitter, as the slaved clock circuit tries to keep in sync with the master clock source)?

 

It's certainly a tempting explanation... but it almost sounds too pat.

 

Yet, an "enhanced stereo field" is often what the kool-aid drinkers... er, adherents to the external-clocking-improves-sound argument promoted by a certain company's advertising as well as a lot of word-of-mouth suggest.

 

Still, I'm having some trouble figuring out how.

 

As jitter increases, it tends to decrease dynamic accuracy -- but don't all the sample measurements fire at the same time in a given converter box?

 

I can easily see how timing inaccuracy from sample to sample (jitter) would degrade the transcription of a given waveform... but wouldn't the frequency of a given wave have to be quite high for such degradation to manifest as a phase shift?

 

Also, isn't jitter almost a kind of extremely high frequency flutter... ie, it's an interstitial irregularity of sample firing, right? So any phase shift would be irregular from sample to sample, right?

 

 

[Have I mentioned lately, that, had I ever declared a major, I would likely have been an English major? My apologies for mangling any science in advance.]

[Ethan Winer]

 

Really, how can that be?

 

 

Digital audio does not "shrink" stereo - that's a myth. Good stereo imaging is all about hearing more direct sound than reflected sound. When reflections are not absorbed the resultant comb filtering skews the response, differently at each ear, and that's what affects imaging. This article explains how bad listening environments can falsely lead to all sorts of audio-related beliefs:

 

A common-sense explanation of audiophile beliefs

 

--Ethan

[Han]

 

Digital audio does not "shrink" stereo - that's a myth. Good stereo imaging is all about hearing more direct sound than reflected sound. When reflections are not absorbed the resultant comb filtering skews the response, differently at each ear, and that's what affects imaging. This article explains how bad listening environments can falsely lead to all sorts of audio-related beliefs: A common-sense explanation of audiophile beliefs --Ethan

 

 

 

Sorry Ethan, but how can it be that a vinyl record sounds better, with a much more precise placement in the stereo field, than a CDR from that very same vinyl, in the same room, with the same amp and speakers.

 

And yes, it was recorded with the Fostex CR300, the same machine Beck has.

[franknputer]

Umm...my first thought would be that a CD of an LP and the LP itself are not the same thing?

 

I know this sounds simplistic - however, there's been loads of discussion as to why CDs might not be as accurate in practice as in theory (even recently).

 

Consider your proposal: basically, you have a turntable going straight to an amp, OR you have a turntable going to an AD converter & a DA converter, THEN to an amp. So even if the sound is not mangled in some way in the digitizing & burning process (which is quite likely anyway) you've added 2 stages of electronics into the signal paths.

 

Simply put, there are many stages of variables added into the process. The signal is going to be affected.

 

Now - better? Worse? That's pretty subjective. But definitely different.

 

I'm sure this will make Ethan's teeth grind. And I would tend to agree that the environment is a huge factor. My point is that, alongside all of the perfections of science there are variables, and these variables add up & affect the outcome of any real-life application.

[Anderton]

 

Consider your proposal: basically, you have a turntable going straight to an amp, OR you have a turntable going to an AD converter & a DA converter, THEN to an amp. So even if the sound is not mangled in some way in the digitizing & burning process (which is quite likely anyway) you've added 2 stages of electronics into the signal paths.

 

 

What many people forget is that there is a lot of ANALOG electronics involved in the conversion process. It may be that any perceived problems with digital relate more to the analog circuitry that helps encode and decode.

[franknputer]

'zackly!

[Rudolf von Hagenwil]

I thought alcohol was talking about a stereo track he recorded and narrowed with panning...

[Han]

 

What many people forget is that there is a lot of ANALOG electronics involved in the conversion process. It may be that any perceived problems with digital relate more to the analog circuitry that helps encode and decode.

 

 

But I'm sure that when I record that vinyl to my Studer two track, the stereo image remains the same. The sound wil not be exactly the same, but the stereo image will. And there's a whole lot of analog electronics involved in that process.

[Rudolf von Hagenwil]

What model Studer two track you use? We have twelve A-80 here for sale.

[Han]

 

What model Studer two track you use? We have twelve A-80 here for sale.

 

 

I have a Studer B67, a Telefunken M10 and a full tube professional Philips Pro 50 which is as big as the fridge and heavy as a small car.

[rlindsey0]

Somebody mentioned phase.... I had an experience recently that may or may not have anything to do with any of this imaging, but here it is anyway.

 

I was recently given a dual mono mixdown of an analog-recorded drum track, which I placed in two separate mono files within my DAW (Tracktion). The kick was driving me crazy because it sounded like it was over to the left, even though the levels (there was a section of the tune with nothing but kick drum) seemed exactly matched between left and right, and even though when I listened to either left or right individually, the levels sounded exactly the same.

 

It turned out that when I looked at the two tracks very closely, the left track had somehow gotten to be ever so slightly ahead of the right. The difference was VERY slight and didn't appear at all until I really jacked up the magnification. I wouldn't have thought it was enough to matter. But even so, seemingly there was enough of a difference that my ear heard the left kick drum attack first and consequently registered the kick drum as coming from somewhere on the left.

 

When I slipped the track ever so slightly so that the peaks aligned more exactly, the kick moved back to the center, where it was supposed to be. After I did this, there weren't any differences in the levels, at least none that my admittedly fairly crude meters could detect. But the placement of the kick had very definitely shifted. It was unmistakable.

 

So, though hardly conclusive, this does make me wonder if phase issues might occasionally come into play with perceived shifts in imaging.

[MikeRivers]

 

Is that the going thinking why some folks prefer external clocking (which tends to increase jitter, as the slaved clock circuit tries to keep in sync with the master clock source)? It's certainly a tempting explanation... but it almost sounds too pat.

 

Consider it one of the many possible explanations. An external clock may or may not increase jitter. It will if the PLL on the receiver is poor, it won't if it's good. But then some people think that when they use an external clock it helps to "glue the mix together." This can be a result of narrowing the stereo field, which you'd get from crosstalk on an analog tape deck. And round and round it goes.

 

As jitter increases, it tends to decrease dynamic accuracy -- but don't all the sample measurements fire at the same time in a given converter box?

 

Jitter causes sidebands to appear around each frequency. That's all it can do. I'm not sure what it is that you're really asking here, but whatever, it's a really complex issue not easily discussed informally. There are a couple of reasonably readable technical papers about jitter on the TC Electronic web site if you feel like digging for them.

 

I can easily see how timing inaccuracy from sample to sample (jitter) would degrade the transcription of a given waveform... but wouldn't the frequency of a given wave have to be quite high for such degradation to manifest as a phase shift?

 

The amplitude of the jitter would have to be quite high for that to happen, and with most external clock generators today, it isn't. So that's not really what's happening. What's happening is that you have extra frequencies thrown into both channels as a result of the jitter. The jitter frequency determines how far removed from the actual sampled frequency they are.

 

Also, isn't jitter almost a kind of extremely high frequency flutter ... ie, it's an interstitial irregularity of sample firing, right? So any phase shift would be irregular from sample to sample, right?

 

It turns out that jitter is both random and frequency-coherent. A small amount is random noise, but the more significant portion of the jitter spectrum is at specific frequencies. It could be the power line frequency, the frequency of some internal clock (or a beat between that and another clock), or some other design issue. Tricky clocks like the Apogee Big Ben use noise shaping to get the jitter frequency up to a point where the PLL will do a good job of rejecting it. By knowing the characteristics of the PLL (which, in the case of other Apogee equipment) they can tweak the external clock so that it actually makes the system sound better than with the internal clock. But this isn't necessarily the case when it's used to clock a different converter.

[MikeRivers]

 

I was recently given a dual mono mixdown of an analog-recorded drum track, which I placed in two separate mono files within my DAW (Tracktion). The kick was driving me crazy because it sounded like it was over to the left, It turned out that when I looked at the two tracks very closely, the left track had somehow gotten to be ever so slightly ahead of the right. When I slipped the track ever so slightly so that the peaks aligned more exactly, the kick moved back to the center, where it was supposed to be.

 

So, did you learn a new way of panning from your experience here? There are two mechanisms at work here, in two different, but still rather small, time difference ranges.

 

Between about 0.1 and 0.7 milliseconds, our brain is sensitive to time differences between the two ears. By adjusting the time in this range (roughly 10 samples at 44.1 kHz) you can pretty much linearly shift an image from full left to full right if you're wearing headphones so you're hearing only the time difference resulting from the delay and not from reflections in the room.

 

In the range of 0.7 ms up to 50 ms, we have the precedence effect, when we hear two versions of the same sound, we hear it as coming from the side from which it arrives first.

 

This is a clever way of panning a source without changing the amplitude. Try it some time when you need to use it.

[blue2blue]

 

Sorry Ethan, but how can it be that a vinyl record sounds better, with a much more precise placement in the stereo field, than a CDR from that very same vinyl, in the same room, with the same amp and speakers. And yes, it was recorded with the Fostex CR300, the same machine Beck has.

 

Just because one likes something more, does not make it more accurate.

 

As someone who built his first stereo when he was 12, when stereo was young and the Beatles had not been on Ed Sullivan yet, I'm pretty aware of the technical limitations of grooved disk stereo reproduction. You may like the sound of stereo as it comes off a grooved disk -- but what you are getting is far from ideal stereo in terms of channel separation, not to mention dynamic range, and a host of related issues.

 

That you like the sound of a grooved record is nice. Quaint.

 

That you confuse that with accuracy is kind of sad.

[rlindsey0]

 

So, did you learn a new way of panning from your experience here? There are two mechanisms at work here, in two different, but still rather small, time difference ranges. Between about 0.1 and 0.7 milliseconds, our brain is sensitive to time differences between the two ears. By adjusting the time in this range (roughly 10 samples at 44.1 kHz) you can pretty much linearly shift an image from full left to full right if you're wearing headphones so you're hearing only the time difference resulting from the delay and not from reflections in the room. In the range of 0.7 ms up to 50 ms, we have the precedence effect, when we hear two versions of the same sound, we hear it as coming from the side from which it arrives first. This is a clever way of panning a source without changing the amplitude. Try it some time when you need to use it.

 

 

Never thought of it that way, but you're right. I'll have to remember that.

[blue2blue]

 

Somebody mentioned phase.... I had an experience recently that may or may not have anything to do with any of this imaging, but here it is anyway. I was recently given a dual mono mixdown of an analog-recorded drum track, which I placed in two separate mono files within my DAW (Tracktion). The kick was driving me crazy because it sounded like it was over to the left, even though the levels (there was a section of the tune with nothing but kick drum) seemed exactly matched between left and right, and even though when I listened to either left or right individually, the levels sounded exactly the same. It turned out that when I looked at the two tracks very closely, the left track had somehow gotten to be ever so slightly ahead of the right. The difference was VERY slight and didn't appear at all until I really jacked up the magnification. I wouldn't have thought it was enough to matter. But even so, seemingly there was enough of a difference that my ear heard the left kick drum attack first and consequently registered the kick drum as coming from somewhere on the left. When I slipped the track ever so slightly so that the peaks aligned more exactly, the kick moved back to the center, where it was supposed to be. After I did this, there weren't any differences in the levels, at least none that my admittedly fairly crude meters could detect. But the placement of the kick had very definitely shifted. It was unmistakable. So, though hardly conclusive, this does make me wonder if phase issues might occasionally come into play with perceived shifts in imaging.

 

 

This is called the Haas Effect.

 

 

[uPDATE: I see this has already been Covered. Mike R, I'll check out the TC white papers. Thanks! One thing, though -- I've read Dan Lavry's rather extensive discourse on and dialog about external clocking and he is quite persuasive to me in arguing against any increase in signal accuacy related to the introduction of an external clock source, even one as expensive or otherwise capable as a Big Ben.]

[Beck]

I'm not sure why the soundstage appears to narrow when going from analog tape to CD, but I've always noticed it even compared to cassette. I'm speaking specifically of making a one-to-one copy direct from 2-track analog to 2-track digital in this case, so it doesn't involve room acoustics.

 

I haven't explored the phenomenon, but my instincts tell me to start with the fact that analog is amplitude based, while digital relies on algorithms to represent everything, including spatial positioning. Stereo analog consists of two physical tracks, while stereo Redbook CD is one file, which has to be interpreted and "redrawn" into the analog world for human consumption. Quality of converters doesn't seem to make much difference and I noticed it going from analog to PCM U-matic back in the day as well.

 

I could be way off on the whys, but it's an interesting subject and worth further investigation.

 

[blue2blue]

I'm not sure why the soundstage appears to narrow when going from analog tape to CD, but I've always noticed it even compared to cassette. I'm speaking specifically of making a one-to-one copy direct from 2-track analog to 2-track digital in this case, so it doesn't involve room acoustics. I haven't explored the phenomenon, but my instincts tell me to start with the fact that analog is amplitude based, while digital relies on algorithms to represent everything, including spatial positioning. Stereo analog consists of two physical tracks, while stereo Redbook CD is one file, which has to be interpreted and "redrawn" into the analog world for human consumption. Quality of converters doesn't seem to make much difference and I noticed it going from analog to PCM U-matic back in the day as well. I could be way off on the whys, but it's an interesting subject and worth further investigation.

Yes. You're way off.

 

Digital does not rely "on algorithms to represent everything, including spatial positioning." Conventional digital audio is psychoacoustics-agnostic, as it were.* It sets out simply to accurately record one or more analog electrical signals -- just as a tape recorder or phonograph recording system attempts to capture and store that signal.

 

Digital relies on periodic voltage measurements. Within a given dynamic range and frequency bandwidth, a properly functioning digital recording system can recreate an electrical audio waveform with measurable precision far greater than tape or vinyl.

 

That does not mean you will like it. People don't tend to like the raw sound of signal coming into the control room during tracking, either, in my experience. Many a musician or worried tyro producer has commented about things sounding too, dry, too thin, not big enough, etc -- when they're getting audio directly from mic-preamp/board-amp-monitor.

 

Don't worry, we used to tell them -- it'll sound better once it hits tape.

Better, in terms of expectations and subjective judgment, mind you -- not more accurate.

 

 

______________

 

* Perhaps you were thinking of perceptual encoding formats like MP3, AAC, WMA, or ATRAC. These do use all sorts of psychoacoustic trickery in order to achieve their seemingly magical data reduction.

[MrKnobs]

I don't even understand the premise of this thread.

 

I mean, if you record two different tracks to digital, then you pan one left and the other right and then solo them one at a time, there's no measurable crosstalk between them - unlike tape where there was always some tiny amount of crosstalk on adjacent tracks.

 

So then you burn them on a CD, and the stuff you panned to the left comes out the left speaker, and vice versa. What more can you ask of a reproduction system?

 

The stereo image YOU put there is what comes out. How is that any different from using tape?

 

Terry D.

[Beck]

Yes. You're way off. Digital does not rely " on algorithms to represent everything, including spatial positioning." Conventional digital audio is psychoacoustics-agnostic , as it were.* It sets out simply to accurately record one or more analog electrical signals -- just as a tape recorder or phonograph recording system attempts to capture and store that signal.

 

Well, I didn't say I could be off in thinking a digital file representing stereo audio is interpreted by an algorithm... because it is. I said I could be off in thinking it has anything to do with why we perceive a narrowing of the stereo field.

 

Yeah, I know how digital works (whispers to self

[Han]

 

Just because one likes something more, does not make it more accurate. As someone who built his first stereo when he was 12, when stereo was young and the Beatles had not been on Ed Sullivan yet, I'm pretty aware of the technical limitations of grooved disk stereo reproduction. You may like the sound of stereo as it comes off a grooved disk -- but what you are getting is far from ideal stereo in terms of channel separation, not to mention dynamic range, and a host of related issues. That you like the sound of a grooved record is nice. Quaint. That you confuse that with accuracy is kind of sad.

 

 

Kind of sad is that you don't read very well.

 

I am a studio owner and moderator of a hi fi forum. I have the opportunity to test

[MikeRivers]

 

I've read Dan Lavry's rather extensive discourse on and dialog about external clocking and he is quite persuasive to me in arguing against any increase in signal accuacy related to the introduction of an external clock source,

 

In general, he's correct. As I said, knowing the characteristics of the PLL can allow you to design an external clock that at least won't make jitter any worse and may make it better. One case where it will improve things is if the internal clock is really poorly designed, but today's converters simply aren't that bad, at least those with an external clock input.