Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[JustinChase]

I can't remember exactly where I ran across this video, but I just watched it after downloading it several weeks ago.

I wonder if this helps explain to others some of the common misconceptions/theories/arguments about bitdepth, conversion, dither and other "audiophile" discussions.

https://www.youtube.com/watch?v=cIQ9IXSUzuM

[~OHM~]

sweet.....thanks

[blgentry]

It's an interesting video that I've seen before.  It doesn't really address much that the audio community seems to argue over.  Still, worth watching, even if it seems to raise more questions.

Brian.

[JustinChase]

I found the part about shaped dithering especially interesting.  But moving most of the dithering to the upper range it seemed to eliminate all audible noise.  I wonder if MC uses or implements shaped dithering in this way.

It also made me realize that higher bitdepth is probably just a waste of disk space for me in pretty much all situations.  However, even with that said, for the same reason I prefer to keep lossless audio, I'll still keep higher bitdepth files, just to have some headroom, and a bit of futureproofing.

[dtc]

He talks about the noise introduced when the signal is bandwidth limited, but does not discuss using higher sampling rates to move that noise to higher frequencies. In the paper he referenced at the beginning of the video, he dismissed using higher sampler rates based on standard arguments, such as the fact that humans cannot hear continuous tones above 20KHz and he also referenced the Boston Audio Society paper There are problems with both those arguments which we do not need to go into here.

The video is a good overview of basic digital audio concepts, but I think he is on shakier grounds in his paper. Despite his arguments that higher sampling rates are not necessary, my wife continues to hear differences between 44K and 96K files. I understand his arguments but my wife's hearing wins. I am not about to tell her she actually cannot hear the differences she routines identifies.

[kstuart]

There are various simplifications of digital audio that are used to explain them to people who are not technicians.

Monty (and several other people who have made a name out of it), then take these oversimplifications and use them as if they actually describe what is going on  ("You can't hear blah blah because less than blah db.... ").

The Monty stuff has been rebutted many times years ago.  But,the idea that cheap equipment and expensive equipment sound exactly the same is oh so very sexy and irresistible - it will be around for hundreds of years.

For example, Monty acts as if the one and only difference between 16-bit and 24-bit is signal-to-noise ratio.  Since, in theory, 16-bit has "enough" then you "cannot hear" any greater SNR.

But Barry Diament, professional recording and mastering engineers, says:

In more practical terms, in a 16-bit recording, one is making the most of all 16 bits when the signal is loudest -- at the top 6 dB of the possible levels (6.02 dB for those who want more precision). Now of course, with real music the signal is not always at the top of the level scale. In other words, not every part of the song is at the loudest point. Further, there are things in the background that are lower in level than things in the foreground. Instrumental harmonics (the things that differentiate a Steinway from a Baldwin or a Gibson from a Fender) are considerably lower in level and spatial cues (whether real or studio generated) are lower still.

To put it into numbers, the harmonics might be 20 dB lower in level than the loudest sounds. Spatial cues might be 40 dB lower in level. (I'm just picking the numbers arbitrarily to illustrate the point.) With a 16-bit recording, those harmonics that are 20 dB lower in level will be encoded using about 13 bits, not 16. The spatial cues that are 40 dB lower in level will be encoded using about 10-bits, not 16. This accounts for the coarsening of the sound and thinning of instrumental harmonics many have notices with 16-bit audio since the beginning.

Now let's look at the resolution of the same items with a 24-bit recording. The harmonics, at -20 would be encoded using about 21 bits. The spatial cues, at -40 would use about 18 bits. Both still exhibiting more resolution than a full level 16-bit recording. So if you make the same recording at 16/44 and at 24/44, you'll find the complexity of instrumental harmonics much better preserved on the latter. Same with the spatial cues: where the 16-bit version defocuses the space, the 24-bit version makes the room boundaries clear and easy to hear by comparison.

[Hendrik]

Well, you just quote another engineer, which you say is more credible than others. But is he? Maybe he just needs to convince everyone that his job is important? People have been known to do that.

This is not about cheap or expensive equipment though. No-one disputes that a good DAC will produce a better analog signal.
The entire argument made in that video is that digital "quality" is often overrated and mis-presented.

Personally I doubt any average person could detect the difference between 24 and 16-bit recordings done in the same quality. Any differences are so low in volume that its rather doubtful you can hear them, as other sounds at much higher volume are going to play at the same time.

[RD James]

24-bit matters for recording. You need the extra dynamic range to avoid harsh digital clipping.
24-bit matters for playback. You need the extra dynamic range to avoid hissing when you lower the volume control.
16-bit is enough for delivering a mastered file. Music does not have anything close to 96dB dynamic range.

Bit-depth reduction does not lower resolution in a properly dithered signal chain. The only thing that changes is the level of the noise floor.
Many professionals are good "field engineers" that know exactly how to properly treat a room and set up mics to create fantastic sounding recordings, but it does not mean that they are technically proficient.

[pschelbert]

yes I have absolutely the same opinion.
There was much discussion on a paper in the AES from "Meyer Moran, around 2007" about 16-bit and 24-bit and sample-rates. Everbody (or at least a lot) cried that they did this and that wrong. Okay may be...
But now I want to see the proof form the ones who blamed  thjat Meyer Moran was wrong... and surprise I cannot find any proof, I mena a scientific test which proofes that Meyer Moran in 2007 was wrong.

Okay until it is proven, we can comfortably believe they were right, otherwise it would have been easy in the past 8 years to make a counter-study....

Lets see who comes up with a proof....

A correct mastered 16-bit CD gets to the max quality the ear is able to distinguish.
Its easy to measure more however.

Well, you just quote another engineer, which you say is more credible than others. But is he? Maybe he just needs to convince everyone that his job is important? People have been known to do that.

This is not about cheap or expensive equipment though. No-one disputes that a good DAC will produce a better analog signal.
The entire argument made in that video is that digital "quality" is often overrated and mis-presented.

Personally I doubt any average person could detect the difference between 24 and 16-bit recordings done in the same quality. Any differences are so low in volume that its rather doubtful you can hear them, as other sounds at much higher volume are going to play at the same time.

[dtc]

yes I have absolutely the same opinion.
There was much discussion on a paper in the AES from "Meyer Moran, around 2007" about 16-bit and 24-bit and sample-rates. Everbody (or at least a lot) cried that they did this and that wrong. Okay may be...
But now I want to see the proof form the ones who blamed  thjat Meyer Moran was wrong... and surprise I cannot find any proof, I mena a scientific test which proofes that Meyer Moran in 2007 was wrong.

Okay until it is proven, we can comfortably believe they were right, otherwise it would have been easy in the past 8 years to make a counter-study....

Lets see who comes up with a proof....

A correct mastered 16-bit CD gets to the max quality the ear is able to distinguish.
Its easy to measure more however.

First, this discussion will go nowhere. It has been hashed out time and time again, to no conclusion. And, shortly, Jim will shut this discussion down, because nothing ever gets settled. Anyone who has been around knows both sides of the argument, knows there is no real conclusion. Some "prove" that 16/44 is fine and others say they can hear the difference. We should just leave it at that way. The question is certainly not going to be settled here It is settled in my house, however, since my wife says she can hear the difference

That said, just because nobody has published a paper in the ASE journal refuting Meyer and Moran's findings, does not mean they are right. One could just as easily say that since they have not repeated the experiment taking the criticisms into account, that the results are highly in doubt. Simply declaring them right - or wrong -  does not settle the disagreement.

But really guys, this will go nowhere. Let it rest.

[JustinChase]

To be clear, it was not my intention to fan the flames of this issue. I just thought the video was interesting and did a good job of explaining the basics of some of these issues.

[dtc]

To be clear, it was not my intention to fan the flames of this issue. I just thought the video was interesting and did a good job of explaining the basics of some of these issues.

Thanks for pointing out the video. Actually, I am the one who stared the discussion about the hi rez issue. I just wanted to point out that his paper was somewhat controversial and hoped it would stop at that. I should have known better.

[JimH]

I just wanted to point out that his paper was somewhat controversial and hoped it would stop at that. I should have known better.

Yes.

[kstuart]

To be clear, it was not my intention to fan the flames of this issue. I just thought the video was interesting and did a good job of explaining the basics of some of these issues.

You are missing a subtle point.  It's actually not the basics of the issues, it's his oversimplification of these issues.

It's like reading an article on a political issue and thinking "He makes good points" and then later reading an article from the opposite side and thinking "He makes good points".

That does not actually tell you anything.

Media works by giving people impressions that they have been "informed", when you do not know what has been left out.  For example, if I read an article about Syria, I do not know whether the seemingly "good points" are leaving out anything important, and actually I will never know.

Personally I doubt any average person could detect the difference between 24 and 16-bit recordings done in the same quality.

I don't think anyone is saying that.  Detecting sound quality differences requires practice.  Most people cannot run a 4-minute mile, and it requires practice to do so.

[pschelbert]

Okay sound differences can be proven with sample persons. Several did extensive trials on that.
Also subjectively perceived sound-differences for example if you see and hear can be proven. The eyes change the perceived sound, if you believe it or not, its a fact (clearly the physical sound is not changed though).
Its a known phenomen in psychology, nothing new.
Nice work on this issue on audio (loudspeakers mainly) was done by Floyd Toole, Harman (formerly JBL).

And a joke about opinions:
If there are two lawyers, the have at least three differing opinions. We are better off we have only two...

[RD James]

I don't think anyone is saying that.  Detecting sound quality differences requires practice.  Most people cannot run a 4-minute mile, and it requires practice to do so.

But there is no difference between 24-bit and 16-bit sound quality.
You just get some "tape hiss" at -84dB with 16-bit audio instead of -132dB with 24-bit.
Hearing that hiss at -84dB while music is playing is not a matter of "training" it's beyond what humans can perceive.

[BillT]

Point of information.

Theoretical noise level of 16 bit audio is -98dB. Properly dithered, as it should be, about -92db.

Theoretical noise level of 24 bit audio is -146db. This is, of course, impossible in the real world. In reality you'll be lucky to get -110db.

In the real world the potential dynamic range of 16 bit recordings exceeds the dynamic range that is practically achievable by any sort of domestic reproduction equipment. Can't say I'd want to listen to music with peaks at 120db SPL either.

[pschelbert]

Point of information.
Theoretical noise level of 16 bit audio is -98dB. Properly dithered, as it should be, about -92db.

probably just a typo: 16-bit audio dithered around -112dB (not -92dB)

And yes recordings I have do not reach even 16-bit, noise is clearly audible and you hear fade-in-fade-out...
For mastered music in my perception its enough.
And no, I do not hear explosions or gunshots (reach 160-170dB SPl shorttime, no micro could record the full range 0-170dB nor any AD converter available).
My 24-bit RME has about 110dB N+THD (119dB SN weighted) and there is may be subtle better only, regardless the price!

[mojave]

How loud a signal that can be recorded or played back is restricted by microphones, mic preamps, or AD converters, but that limit is quite high. In other words it can be done at very high SPL with the right equipment. Shure has condenser mics that record up to about 190 dB and piezoelectric mics go up much higher. Bit-depth has nothing to do with maximum volume.

Can't say I'd want to listen to music with peaks at 120db SPL either.

At the JRiver room at Rocky Mountain Audio Fest we were listening with peaks over 130 dB measured 4 meters from the speakers. Those in the room, including me, were loving it!

[pschelbert]

with SPL I mean a max SPL with a noisefloor which is low as of best mics available around, 20dB

If you go with 190 dB SPL, a noise floor of 20dB you would need a 170 dynamic range,
-would need a 28bit AD
-mic with 20dB floor, 190dB max (<1% THD?)
I have not seen equipment for that. Not really needed as well is my opinion.

What you can buy on the shelf is 24-bit AD with around 20-bit precison (THD+N)
Mics like Schoeps or Neumann are around 20dB floor, 130-140dB max SPL (<1%)
Fits well: 110-120dB dynamicrange on mic with AD 24-bit (20bit effective) gives the 110-20dB range.

[BillT]

probably just a typo: 16-bit audio dithered around -112dB (not -92dB)

No typo. Dither increases noise level (and reduces distortion), it doesn't reduce noise by 14dB.

How loud a signal that can be recorded or played back is restricted by microphones, mic preamps, or AD converters, but that limit is quite high. In other words it can be done at very high SPL with the right equipment. Shure has condenser mics that record up to about 190 dB and piezoelectric mics go up much higher. Bit-depth has nothing to do with maximum volume.
At the JRiver room at Rocky Mountain Audio Fest we were listening with peaks over 130 dB measured 4 meters from the speakers. Those in the room, including me, were loving it!

Yes, it's possible to record very high SPLs and even reproduce them with sufficiently robust speakers and powerful amplifiers, although it's probable that any other qualities except loudness will be lost.

I meant that the signal to noise ratio of 16 bit recordings is entirely adequate for domestic reproduction. The quietest domestic room will have a noise floor of more than 30db spl. That means that the loudest sounds will be more than 122dB SPL when using the maximum DR of 16 bit recordings. That is beyond the capability of most domestic equipment and is a pretty undesirable level. 120db spl carries a significant risk of instantaneous hearing loss.

Then there's the issue of musical sources having much dynamic range anyway. The largest orchestral concert will produce peaks of about 120db spl in the audience, but the ambient noise in the auditorium is going to over 45 db spl, so you only need a dynamic range of about 75db. Smaller orchestral forces will give much lower spls, hence need even lower dynamic range.

In my experience, popular music has a dynamic range of 20-30dbs at best. In live performance it's usually accompanied by ear piercing levels of distortion and painful amounts of colouration. It might be loud, but it doesn't have much in the way of quiet bits.

[Hendrik]

No typo. Dither increases noise level (and reduces distortion), it doesn't reduce noise by 14dB.

While this is true, dithering can actually increase the perceived dynamic range of a 16-bit signal up to around 120 dB when using proper noise-shaped dithering, by taking advantage of the characteristics of the human ear (moving noise into areas where the ear is less sensitive)

[pschelbert]

about dithering and how it sounds here a nice webpage:

http://www.audiocheck.net/audiotests_dithering.php

[kstuart]

But there is no difference between 24-bit and 16-bit sound quality.
You just get some "tape hiss" at -84dB with 16-bit audio instead of -132dB with 24-bit.
Hearing that hiss at -84dB while music is playing is not a matter of "training" it's beyond what humans can perceive.

This is classic.

You did not read the quote in my post:

http://yabb.jriver.com/interact/index.php?topic=101638.msg705521#msg705521

about why it is not just about signal-to-noise ratio.

Since it is clear that using the built-in "quote" insures that the quote will not be read by anyone, then here it is again without the "quote-of-death":

Barry Diament, professional recording and mastering engineer (Bob Marley, Led Zeppelin, Yes, ELP, CSN, Enya, etc.) wrote:

" In more practical terms, in a 16-bit recording, one is making the most of all 16 bits when the signal is loudest -- at the top 6 dB of the possible levels (6.02 dB for those who want more precision). Now of course, with real music the signal is not always at the top of the level scale. In other words, not every part of the song is at the loudest point. Further, there are things in the background that are lower in level than things in the foreground. Instrumental harmonics (the things that differentiate a Steinway from a Baldwin or a Gibson from a Fender) are considerably lower in level and spatial cues (whether real or studio generated) are lower still.

To put it into numbers, the harmonics might be 20 dB lower in level than the loudest sounds. Spatial cues might be 40 dB lower in level. (I'm just picking the numbers arbitrarily to illustrate the point.) With a 16-bit recording, those harmonics that are 20 dB lower in level will be encoded using about 13 bits, not 16. The spatial cues that are 40 dB lower in level will be encoded using about 10-bits, not 16. This accounts for the coarsening of the sound and thinning of instrumental harmonics many have notices with 16-bit audio since the beginning.

Now let's look at the resolution of the same items with a 24-bit recording. The harmonics, at -20 would be encoded using about 21 bits. The spatial cues, at -40 would use about 18 bits. Both still exhibiting more resolution than a full level 16-bit recording. So if you make the same recording at 16/44 and at 24/44, you'll find the complexity of instrumental harmonics much better preserved on the latter. Same with the spatial cues: where the 16-bit version defocuses the space, the 24-bit version makes the room boundaries clear and easy to hear by comparison. "

[glynor]

Sigh. Hendrik already responded to this:

Well, you just quote another engineer, which you say is more credible than others. But is he?

Interact is not the place to hash out these disagreements. The end.