Dither - My Ears can't hear it

[SouthSIDE Glen]

Thus there is a form of "noise-shaping" occurring with the piano signal . . . it just so happens that its noise is audible, in the lower frequencies


Yep, a quick retest of the test signal with added simulated 60Hz hum . . . no more quant. distortion.

It sounds like you're saying that the hum is acting as a form of dither itself, in a way. It of course has the downside of being audible and rather a form of overkill as compared to real dither; but it does demonstrate the real-live plausability of using one signal to remove the quantitaztion distortion from another.

G.

 

[mshilarious]

It sounds like you're saying that the hum is acting as a form of dither itself, in a way. It of course has the downside of being audible and rather a form of overkill as compared to real dither; but it does demonstrate the real-live plausability of using one signal to remove the quantitaztion distortion from another.

G.

Correct. The trouble for me is it will be very difficult to remove that last bit of hum. I still have a bunch of crap dimmer switches to replace, and the good ones are $50 each We're talking -90dBFS peaks, or less, so it's not really a sound quality issue (unless you are playing an electric guitar!) It's not terribly audible; in comparison with the white noise from the electronics, anyway.

I also had a 40Hz rumble from something . . . maybe the heat pump outside. Downstairs was off, upstairs could have been on. I could have killed some of that if I'd shockmounted the mic . . .

 

[Ethan Winer]

Even so, I couldn't find the quantization distortion! It should be there, but I can't generate it! And I tried pretty hard!

Indeed, and this is 180 degrees counter to your earlier assumption, "As for dither--I am confident based upon my earlier analysis that I could prove audibility."

I don't say this to pick on you! But to show how rampant it is that people assume this or that based on what they read in a book or magazine. I also agree with TerraMortim that taken in isolation this is not a big deal. But it's part of a much larger problem - people readily accept as true things that are not true, and sometimes they even get very testy about it.

--Ethan

 

[mshilarious]

Indeed, and this is 180 degrees counter to your earlier assumption, "As for dither--I am confident based upon my earlier analysis that I could prove audibility."

I don't say this to pick on you! But to show how rampant it is that people assume this or that based on what they read in a book or magazine. I also agree with TerraMortim that taken in isolation this is not a big deal. But it's part of a much larger problem - people readily accept as true things that are not true, and sometimes they even get very testy about it.

--Ethan

Well, in my defense, I didn't assume anything was true without testing it first. And so I did demonstrate quant. distortion with a test signal. And I believe I came very close to matching that test signal with a reasonable, real world signal. If I work the hum issue a bit more, I might even find real-world quant. distortion.

Even if I don't, to suggest that I accept assumptions untested is clearly not the case.

PS Right before I went to bed, just for fun I truncated to 8 bit. Yowza! That's some distortion!

 

[SouthSIDE Glen]

Indeed, and this is 180 degrees counter to your earlier assumption

You also cherry-picked an old quote, there Ethan. If you keep reading, it turned out that the Q.D. wasn't there because it was being dithered out by other unwanted signals. All this last test really did was prove the concept that dithering does indeed reduce Q.D. It didn''t address the question of the audibility of that Q.D. whatsoever - one way or the other.

It's that kind of misleading cherry-picking of quotes that just keeps the entire debate wallowed in mud and serves only to prolong the debate; it serves you just as badly as it serves everyone.

And I'd like to take that opportunity to ask everyone, just what does "audible" mean here? Is a dog whistle "audible" or not? Is subliminal advertising in retail Muzak "audible" or not? Are the low levels of white noise pumped through some casinos "audible" or not (FTM can one discern without test gear the higher oxygen levels in thse same casinos or not?)

G.

 

[mshilarious]

I did measure Q.D. (that seems like a Latin abbrev., n'est-ce pas?) on a kick drum tonight. This particular drum is missing a head at the moment, so it sounds more like a tympani, with a nice long decay.

However, the Q.D. was at -105dBFS when the signal was fading at -80dBFS (with the attack at -6dBFS), so I am not yet my own personal hero . . .

 

[Ethan Winer]

it turned out that the Q.D. wasn't there because it was being dithered out by other unwanted signals.

This is an excellent point because it bolsters my argument that dither is meaningless for almost all practical content since it's 20+ dB below the ambient noises already in the recording.

Is a dog whistle "audible" or not?

Not.

Is subliminal advertising in retail Muzak "audible" or not?

Not.



--Ethan

 

[Ethan Winer]

to suggest that I accept assumptions untested is clearly not the case.

I didn't mean that as an attack, or to single you out! There are at least four "does dither matter?" threads going on right now in various forums, and in every one you'll find people saying of course it matters even though not one of them has risen to my challenge to show a file where the difference between dithered and truncated can be clearly heard. Hopefully Glenn will be able to do that. And yes, I do mean hopefully. I'd love to be proven wrong. I doubt I will, but I want to know The Truth as much as anyone else.

--Ethan

 

[mshilarious]

This is an excellent point because it bolsters my argument that dither is meaningless for almost all practical content since it's 20+ dB below the ambient noises already in the recording.

Again, you can't take a signal peak in dBFS and compare that directly to a dBA noise floor. That is what I am trying to show--20dBA noise with a standard (94dBSPL peak) source is going to be something like -105 to -115dBFS integrated noise across the spectrum. And a QD peak can be above that level, and thus audible.

So far I have gotten derailed by a -90dBFS hum peak. But I still think an excellent, hum-free facility that is doing classical soloist or ensemble tracking needs to think about dither. I should be able to have another go tonight, and this time all hum will be eliminated, at the panel box if I have to.

Implicit in that statement is my feeling that a rock band homerec probably doesn't need to be concerned with dither.

 

[SouthSIDE Glen]

This is an excellent point because it bolsters my argument that dither is meaningless for almost all practical content since it's 20+ dB below the ambient noises already in the recording.

Going around in circles, baby. It's not BELOW anything. There is no masking involved. Dither is not even even noise in the sense you keep coming back to. I don't think you'll ever quite get that.

Not.


Not.

Tell that to the people who can feel dog whistles shoot right through their heads, and to the retailers who have done studies demonstrating a signifigant percentage drop of shoplifting and increase in sales in stores that use subliminal messaging for such purposes, and to the casino owners who use similar methods to keep customers burning their money into ashes in their casinos, and who are famous for being tighter with their money than Jack Benny on anything that isn't a definitive money generator for them.

There's audible and there's audible.

Dither is never going to club anybody over their head with obviousness. It's not even going to tickle some others. But subtlety and non-ubiquity are not the same as irrelevancy or non-existance. This is the very reason why nobody has come up with such a test. It's a game of percentages, and one that can only be tested for via a large enough population study with no room for bias and no obvious artifacting in the design of the test itself. Ain't a whole lot of gear sluts or music geeks that know how to do that.

I can't get past the problem of cheating in the Internet environment though. Any such test, regardless of the result, is open to corruption by those with a bias one way or the other. The only way it could be done properly is in a controlled environment.

G.

 

[Ethan Winer]

subtlety and non-ubiquity are not the same as irrelevancy or non-existance. This is the very reason why nobody has come up with such a test.

Forget the test. I'd be happy to see you or someone else post any pair of files showing that the presence of dither makes any difference at all on any type of music when recorded at sensible levels. If dither actually mattered, someone would have risen to this challenge a month ago.

--Ethan

 

[Golden]

I just read the last few posts... so, I'm not sure if anyone has mentioned the Pow-r dither box by weiss... it's what, $3400?

what's the point of this box? if dither can't be heard, what's the point? Is it just weiss out to get some sweet cash? maybe the box is made out of gold?

I'm honestly just curious...

 

[masteringhouse]

Forget the test. I'd be happy to see you or someone else post any pair of files showing that the presence of dither makes any difference at all on any type of music when recorded at sensible levels. If dither actually mattered, someone would have risen to this challenge a month ago.

--Ethan

Ethan -

Can you show us an example of where Realtraps make an audible difference in a recording? Looking for one where a recording was made without them and then with them (no other variables) that definitively shows the difference.

 

[SouthSIDE Glen]

I'm getting off, this thread just gets me dizzy now.

Like a circle in a spiral, like a wheel within a wheel,
Never ending or beginning on an ever spinning reel,
Like a snowball down a mountain or a carnival balloon
Like a carousel that's turning running rings around the moon,
Like a clock whose hands are sweeping past the minutes of its face,
And the world is like an apple whirling silently in space,
Like the circles that you find in the windmills of your mind.

G.

 

[mattkw80]

Ethan -

Can you show us an example of where Realtraps make an audible difference in a recording? Looking for one where a recording was made without them and then with them (no other variables) that definitively shows the difference.

Check these video's out.

http://www.realtraps.com/videos.htm

You can hear the difference when using and not using the portable vocal booth.

 

[masteringhouse]

Check these video's out.

http://www.realtraps.com/videos.htm

You can hear the difference when using and not using the portable vocal booth.

What if I were close micing a bass cabinet, would bass traps in a room make a difference?

My point is that the audibility of an effect depends on the environment in which it is demonstrated, as has been said in many previous posts. Granted the audibility of dither is subtle. It was intended to be that way, who wants to actually hear dither?

The effect that dither has on reducing harmonic distortion can be masked by other textures on a song. In a recording where harmonic distortion is more noticeable, the effect that dither has is more noticeable. I think that the samples that I posted clearly demonstrate the audibility of quantization distortion and how dither helps. Some might say that they are exaggerated or not at sensible levels. This is true to some extent, but I created them only to allow those who do not know what harmonic distortion sounds like in the first place to know what to listen for. If we added small amounts of harmonic distortion from other sources would we say that it's inaudible and doesn't matter? Maybe some would, but it's not what I would want any pro engineer I hired to say (of course unless it was for effect).

I guess that it all comes down to your attitude on audio. For some "good enough" isn't "good enough" and they are always looking for ways to improve their chain, gear, and ears. The mere fact that there's an element of compromise, though inaudible to some, bothers them. Also subtleties tend to add up making the overall quality better. Bass traps by themselves don't make great audio anymore than dither does, but the combination of both gets you closer. Putting a dish towel over a glass table (see Ethan's video on room setup) is an example of something that isn't going to be audible in and of itself, but in combination with other treatments helps. The sum is greater than its parts.

I like to think that this forum is for those that want to be the best engineers that they can be within a home recording budget. Dither is cheap and has been shown to improve the quality of your audio when used correctly. Why wouldn't you use it?

 

[Ethan Winer]

it's what, $3400? ... I'm honestly just curious.

You nailed it. To me this stuff is as much about consumerism and value as audio quality.

The same applies to expensive outboard word clocks that reduce jitter which is also inaudible. Same for expensive outboard summing devices that solve the non-problem of DAW summing. Same for outboard anything else that adds small amounts of very expensive distortion - except instead of calling it distortion they use words like warmth, color, glue, character, etc. The key word here in all cases is "expensive."

--Ethan

 

[Ethan Winer]

Can you show us an example of where Realtraps make an audible difference in a recording?

Yes! This is very easy to prove both on paper and audibly. Besides the PVB videos Matt linked, the MiniTraps Demo video also lets you hear how the sound quality improves as traps are added one by one over the course of the video.

Now, in some cases close miking can reduce the affect of room treatment. But many instruments cannot (or should not) be recorded close up. More to the point, bass traps and other treatment are crucial in a mix room. No matter what happens when recording, your only chance to fix problems and make things right is when mixing. The improvement in a listening room after adding bass traps and other treatment is clearly audible to everyone, and the improved response and ringing are also very easy to prove with graphs.

Since you're a mastering engineer, I'm certain you're contriving this example just to make a point, and you don't really believe that acoustic treatment is a placebo, right?

Also, please understand that my comments here have nothing to do with my business of manufacturing acoustic treatment products. This is me talking, not the head of RealTraps.

Dither is cheap and has been shown to improve the quality of your audio when used correctly. Why wouldn't you use it?

I've already acknowledged that dither is usually cheap (discounting the Weiss box mentioned above). This is not the point. As best I can tell dither has never been shown to improve the quality of anything. This is why I keep asking for someone to post an example of dithered versus truncated that shows the improvement clearly and without question. If you have such an example, I'll be glad to be proven wrong.

--Ethan

 

[masteringhouse]

As best I can tell dither has never been shown to improve the quality of anything. This is why I keep asking for someone to post an example of dithered versus truncated that shows the improvement clearly and without question. If you have such an example, I'll be glad to be proven wrong.
--Ethan

I have shown an example, you are just unwilling to admit that it's audible, or want to prove your point by saying that at "normal" listening levels (whatever that is) it is irrelevant. For every example that you show where dither is "inaudible" I can create a scenario where the use of bass traps are "inaudible". So what ...

I do agree that good bass traps help with room acoustics, but they are not always that audible or measurable. It depends on the room. Likewise with dither. It doesn't mean that either do not have their proper role.

Ethan, the pseudo-science you are propagating is truly disturbing from someone who is supposed to be an audio professional. I know of no other pro audio engineer worthy of this title that says dither and jitter are inaudible or are ploys for generating income. Granted some companies do distort facts in order to sell gear, but to say that truncation distortion and jitter are just ruses for selling gear is like saying a cold or flu are sicknesses used to sell cough medicine.

 

[Ethan Winer]

I have shown an example, you are just unwilling to admit that it's audible

I never bothered to download those files originally because I saw the comments from others that the levels are very low and you used sine waves versus music etc. But I just downloaded all three files now, and it's clear you made a grievous error in your processing.

You say you started with the 24 bit file and reduced it to 16 bits truncated and dithered. I have no idea what you did, but the damage in both cases is way beyond what I get when I bit-reduce the original to 16 bits. Neither of your files resembles the original sine wave, even though the recorded level is at a very reasonable -14 dB. Versus my truncated version that looks identical to the source. Here's a screen cap showing all four files in Sound Forge, with the same vertical and horizontal scaling:

So whatever you think is being proven by your files, you're mistaken. Perhaps you accidentally reduced them to all the way down to 8 bits instead of 16?

I do agree that good bass traps help with room acoustics, but they are not always that audible or measurable. It depends on the room. Likewise with dither.

The improvement in response and ringing after adding bass traps is always audible and measurable, assuming good bass traps and an appropriate number of them. Like at least two for a bedroom size listening room.

I know of no other pro audio engineer worthy of this title that says dither and jitter are inaudible

You need to get out more.

Seriously Tom, there are a lot of professional engineers who agree with me on this, and they are just as upset as I am to see so much time and energy wasted on non-issues like dither and jitter. In fact, you can expect to see an article about this by a respected columnist in a major magazine this spring or early summer.

--Ethan