Is this really how Pro Tools does panning? Take the test

[u b k]

Originally Posted by Lynn Fuston

Mono should always be a consideration.

mono should sometimes be the *goal*.


gregory scott - ubk
.

[Nishmaster]

Originally Posted by oky****

...yeah, nishmaster's posts are erroneous / deceptive. i'll get to it....

I'm kicking myself for continuing to bother, but I feel this is an important issue that needs to be resolved properly. I am not trying to deceive anybody, and as you can see, I've provided as clear and concise an example as I possibly can. The math is correct here.

also, i am smart enough to know when they are gunning for me. it is a group that generally works together. and at a certain point it makes no difference what you say, they will try to protect their turf.

I am not gunning for anybody. I'm not working with anybody either. I don't know Mr. Frindle personally aside from the one or two very brief exchanges we've had on this forum. I have no turf to protect; I'm merely a lowly engineer at a lowly studio who happens to have a great deal of interest in electrical engineering and design, and I'm trying to present clear facts with the appropriate math and evidence to back them up.

and, beyond that, i am alternately laughing my ass off at how inconsistent and off-track some of my "detractors" are, and, on the other hand, being annoyed at how they are willing to be so misleading just for the sake of their economic interests and common oneupmanship. i don't like to see that.

I am willing to accept that I may be inconsistent, incorrect, or generally wrong at any point. I am the first to admit when I've made a mistake. But the math here is very simple, and you have yet to show me any alternate avenues that I may have forgotten, or in fact comment directly on anything I've presented, even after following the rules you have laid out. Again, I have no intention, nor do I believe that I am misleading anyone. I have no economic interest in this thread, nor do I wish anyone to be "oneupped." I simply wish for the correct information to prevail.

At this point, unless you wish to comment directly to the aforementioned post (which I am fairly certain you won't), I'm not going to take this thread any further. I have presented the mathematics as they apply to real world situations and facts both in detail in some responses and concisely in others, and I hope that the people reading can pick the signal from the noise in this thread, as it were.

Total noise does not equal noise source one plus noise source two.

Total noise equals the square root of the sum of noise source one squared plus noise source two squared. My final example.

Noise source 1, -100dBu of noise. -100dBu = 0.000007746 volts RMS.
Noise source 2, -100dBu of noise. -100dBu = 0.000007746 volts RMS.
Total noise = SQRT(0.000007746^2 + 0.000007746^2).
Total noise = SQRT(6.00005160000e-11 + 6.00005160000e-11)
Total noise = SQRT(1.20001032000e-10)
Total noise = 1.09544982541e-05 volts RMS
Total noise = -96.989662692dBu

[Paul Frindle]

Originally Posted by oky****

hi,

there is an old pro tools plug-in that did that kind of thing. the one with the picture of the guy's head that let you point to where you wanted the sound to be coming from. i think you could point to up or down in addition to left right. some sort of reverb thing too, i think.


right.

I vaguely recall such a thing - but not sure it was a PT plug-in?

Most panners are for surround (multiple speaker systems of various kinds) and try to intelligently do the level control between them. It's even less obvious what the pan laws should be when more speakers are added - in fact the more you have the worse it gets in many ways.

We had much discussion about this for the surround panning in the OXF-R3 - and now I can't remember which particular line was actually taken. The work was done by another engineer. The only thing I do remember about it was that it tried to maintain positional and level similarity between 5.1 and 7.1 for film production.

But of course none of this would produce up/down or delayed signal of any kind - it was level only.

[Lynn Fuston]

Originally Posted by Paul Frindle

I vaguely recall such a thing - but not sure it was a PT plug-in?

Most panners are for surround (multiple speaker systems of various kinds) and try to intelligently do the level control between them. It's even less obvious what the pan laws should be when more speakers are added - in fact the more you have the worse it gets in many ways.

We had much discussion about this for the surround panning in the OXF-R3 - and now I can't remember which particular line was actually taken. The work was done by another engineer. The only thing I do remember about it was that it tried to maintain positional and level similarity between 5.1 and 7.1 for film production.

But of course none of this would produce up/down or delayed signal of any kind - it was level only.

Took some looking, but is this what he is referring to? Never seen if before.

ARL Sound Stage



According to their description:

Step onto a virtual soundstage! The ARL Sound Stage™ plug-in, which includes ARL Pan™ and ARL X-Over™ provides a unique listener point of view mixing environment for surround sound. Add incredible depth and motion effects to your surround mixes. Head-related delays, volume, and frequency response dynamically change as the sound sources move about the virtual sound stage.

[okydoky]

Originally Posted by Nishmaster

The math is correct here.

Total noise does not equal noise source one plus noise source two.

Total noise equals the square root of the sum of noise source one squared plus noise source two squared. My final example.

Noise source 1, -100dBu of noise. -100dBu = 0.000007746 volts RMS.
Noise source 2, -100dBu of noise. -100dBu = 0.000007746 volts RMS.
Total noise = SQRT(0.000007746^2 + 0.000007746^2).
Total noise = SQRT(6.00005160000e-11 + 6.00005160000e-11)
Total noise = SQRT(1.20001032000e-10)
Total noise = 1.09544982541e-05 volts RMS
Total noise = -96.989662692dBu

hi,

watch, listen [feast your glazzies, and viddy this, me droogies]

http://www.youtube.com/watch?v=5LeLAELIxKY

"karma police, arrest this man, he talks in math.
he buzzes like a fridge, he's like a detuned radio." ]radiohead[


right.

[Nishmaster]

Originally Posted by oky****

hi,

watch, listen [feast your glazzies, and viddy this, me droogies]

http://www.youtube.com/watch?v=5LeLAELIxKY

"karma police, arrest this man, he talks in math.
he buzzes like a fridge, he's like a detuned radio." ]radiohead[


right.

Well, at least I know for sure now that you were trolling and not just missing the concept.

But thanks for the link! I hadn't seen the video to Karma Police before. Very cool stuff.

[okydoky]

Originally Posted by Nishmaster

Well, at least I know for sure now that you were trolling and not just missing the concept.

But thanks for the link! I hadn't seen the video to Karma Police before. Very cool stuff.

hi,

no, you're wrong again.

why would you thank me for the link?


right.

[Nishmaster]

Originally Posted by oky****

hi,

no, you're wrong again.

why would you thank me for the link?


right.

Why wouldn't I? It was a cool video. I'm not insulted; I have no bone to pick.

Good day to you sir.

[okydoky]

Originally Posted by u b k

t
if i .. miss out on ...brilliance, i'm okay with that.



gregory scott - ubk
.

hi,

link:

TR-i


right.

[okydoky]

Originally Posted by Nishmaster

Why wouldn't I? It was a cool video. I'm not insulted; I have no bone to pick.

Good day to you sir.

hi,

yeah.


right.

[okydoky]

hi,

greetings, salutations, and belutations.


for your perusement, amusement, and un-confusement we offer the following: [hopefully not too many "mistakes"].


"Note that dither can only increase the resolution of a sampler, it cannot improve the linearity, and thus accuracy does not necessarily improve." - wikipedia [more about dither to come]



anyhow, sorry if nishmaster had to rewrite. i think he was posting a new one while i was writing the one that he answered, or something.

anyhow, the analysis he presented is defective, which is no big deal. however, what i do take exception to is the fact that it appears to be deliberately misleading.

i don't really mind the fact that a lot of people may be naive, and trying to learn, but deliberate deception to try to make a case for "in the box mixing", or anything else, really, is not cool.

there was failure to include the entire signal path in the "post 266" example, by leaving out the mix bus input on the "pan itb" example. including the mix bus, using your figures and method of calculation, gives a noise level of -96.46dB.

again using your figures and method of calculation, the "pan otb" example actually yeilds a noise level of -99.99dB [not -99dB, as he posted].

thus, there is a 3.54dB increase in noise floor in the "itb" example, properly calculated up to the mix bus input.

that is more than twice as big a noise floor.

generally, unless you deliberately try to massage the numbers or pick and choose numbers to "your" advantage [as was also done to some extent], you will have a 3dB increase, or so, with both noise sources under the best of conditions.

3dB is twice as much noise.
even a 1dB increase is 33% more.

1dB = 33% more noise power
2dB = 66% more noise power
3dB = twice as much noise power

there has also been an attempt to improperly trivialize certain things [most particularly the significance of a 1, 2, or 3dB increase in noise] to support your position, and a naive reader may be persuaded. 3dB does not look like a big number on paper.

in any event, it is always a game of inches, and even 1 dB of unnecessary noise should be avoided where it can be, because it will definitely be amplified and exacerbated later in the chain, and, perhaps more importantly, in mastering. and if you have 48 or 60 channels, and out of those you have 10 or 15 or 30 that have unnecessary noise [including dither noise], it adds up.

what you have attempted is what manufacturers often do, and that is to present the numbers in a specific way, and to frame the issues in such a way that your argument is artificially supported.

this gain scaling thing is just common sense, and the fact that any of you are trying to prove me wrong on such a no-brainer is very telling.

as i believe mr. gwinn concurred, the scenario of introducing a noisy signal at a low level to a console, is "bad, very bad", particularly where you have the option to do otherwise.

it seems to me that mr. nischan is an audio engineer only. i don't mean that as an insult.

however, mr. gwinn is an electrical engineer at benchmark media systems, and also an audio engineer. he stated:

Originally Posted by elias gwinn

"Oops....

Yes, scenario 2, which is an example of proper gain staging, will have a better signal to noise ratio.

Scenario 1 is bad. Very bad.

atb,
e
__________________
Elias Gwinn

Electrical Engineer
Benchmark Media Systems, Inc"

note that i am pretty much giving your position every benefit of the doubt, and even assuming absolutely incoherent noise signals, although you know there is some coherence, unless you have a gasoline [or natural gas] powered studio [most of us are using electricity out of the wall]. coherent noise is going to add much more sharply.

you can play number games till you drop [and engineers and manufacturers frequently do], but anyone who actually has any experience with professional audio knows better than that.

please see the francis buckley video clip i posted.

please also read pretty much any available text on gainstaging that exists on this planet. this issue is not even controversial.

everyone please note that these guys are suggesting bringing a signal into the console at a level potentially at or below the noise floor of the console itself [-90dB is not unusual for an analog console, and -90 may have been a proper figure for input and a proper figure for the mix bus].

as discussed below, they are proposing a scenario where it would be impossible to apply e.q. or compression or anything else on that channel, because you cannot e.q. or compress a channel with that poor a signal to noise characteristic. if you [try to] jack the channel up at the input trim in the hopes of getting enough level so that some theoretically perfect compressor may have a fighting chance of grabbing the signal, then you’re going to have to do the same on the other half of the stereo pair, where you will be screwed because you're already 90dB hotter on that side.

your other option would be to open the compressor's input wide open and hope to suck the noisy signal in so that the compressor can amplify all that noise.

so you have one channel that is too low to process and / or one that is too hot to handle. take your pick.

anyhow, in addition to the points already made, the example that was posted, and the entire idea of panning in the box, fails to take into account the effect of introducing any type of signal processing on a console insert.

in point of fact, by bringing in a signal at such a ridiculously low level you would effectively be prevented from get a compressor or other processor to work in that channel at all. and even if you did, it would skyrocket your noise floor. moreover, you would have the "challenge" [putting it nicely] of two panned channels, at radically different levels, to compress, instead of one. no way to even use a stereo linked compressor.

the idea of panning a mono signal in the box, and then introducing into the console is literally stupid. it requires the unnecessary use of twice as many channels on the console, and a host of other absurdities. it defeats any reasonable scheme for processing at the console.


note that you did not say, "in my example it makes no difference" [which might have actually given you an out, because you could have said, "well i did it like oky**** said", but instead you stated , essentially, "it could never make a difference", which is completely untrue under any analysis.

as a rule [and I know that term generally elicits a gag reflex on this forum ], you should always try to keep the signal of interest at its highest level above noise, for as far along into the signal path as possible. you have a long way to go until, God willing someone puts your product in his or her cd player, including all the way through mastering, where more compression will almost inevitably be utilitzed, and it will finally be dithered and truncated to 16 bits.


i am not directly addressing at length the fact that you would never be able to get your "in the box" panned signal at a mere -90, because it would be impossible for you to keep that channel fader at zero if you are mixing in the box. there is a whole host of other stuff that i am not addressing at all that would tend to further support my position, because this is not a book.


by the way, according to the data sheet, the dac's linearity on a digidesign 192 i/o is shown as good down to about -110, where it gets a little strange [but still pretty good] and stranger by the time it gets to -130. your -90dB panned in the box signal could actually be down between the -110 and -130 area, or lower.

thd+noise is actually between -98dB and -100dB generally, @ 44.1

lower amplitude signals may possibly be more compromised by pre-ringing and post-ringing. so there's that to consider.

the examples given construe signal to noise ratio relative to the loudest component [or perhaps average] of the sound, which is deceptive.

a complex signal, such that the harmonics essential to its natural tonality would lie below the dynamic range of the dac when the signal is attenuated is not going to be helped by dither's ability to remove quantization error. rather, those harmonics will simply be missing altogether from the reproduced signal. this would of course be alleviated by keeping the entire signal within the dynamic range of the dac [my plan].

and, as the quote at the beginning of this post points out, dither does not necessarily help accuracy.

there is a tendency to think of signals as being at a single dynamic range, but that is not the case. any complex sound contains harmonic elements existing at various areas of the dynamic range, not just the place where the fundamental is measured. people tend to think that everything is just a single sine wave, but that is not the case. people like to say, "this sound is n dB loud", but in reality, a component of the sound may be n dB loud, and other components may be at, say n-3dB, or n-10dB, or n-50dB, or whatever. if you attenuate a signal enough "in the box", by panning or otherwise, it is possible to create a scenario where some of the lower amplitude components are simply no longer represented at all by the system.

in my way, you have the possibility that the sound will be heard through the transparent analog noise floor. in the way proposed by the others, you have a possibility that the sound [or components of it] will not even make it to the console at all!

get the signal of interest as far along the path as possible, without unecessary attenuation.

i realize that i myself limited the analysis to exclude many of the variables. i had to do that to get some sort of statement that was not so convoluted as to be incomprehensible.

i am only mentioning other issues now because I have already shown is that, even excluding the variables, and creating a scenario that is wildly favorable to the "in the box" option, that option stills fails.

when the actual real life variables are included, the idea of unnecessarily presenting a musical signal to a console at a significantly lowered level becomes ever more absurd.


love always,


right.

[okydoky]

hi,

info / quote about quantization for your consideration and comments [emphasis added] [wikipedia, i think]


"Analysis of the quantization error of low-amplitude signals reveals that teh spectrum is a function of the input signal. The error is not noiselike (as with high-amplitude signals); it is correlated. At the system output, when the quantized sample values reconstruct the analog waveform, the in-band components of the error are contained in the output signal. Because quantization error is a function of the original signal, it cannot be described as noise; rather, it must be classified as distortion.

As noted, when quantization error is random from sample to sample, the rms quantization error E (sub)rms = Q(12) sup.1/2. This equation demonstrates that the magnitude of the error is independent of the amplitude of the input signal, but depends on the size of the quantization interval; the greater the number of intervals, the lower the distortion. However, the relevant number of intervals is not only the number of intervals in the quantizer, but also the number intervals used to quantize a particular level.

A maximum peak-to-peak signal (as used in the preceding analysis) presents the best case scenario because all the quantization intervals are exercised. However, as signal level decreases, fewer and fewer levels are exercised as shown in Fig. 2.8. For example, given a 15-bit quantizer, a half-amplitude signal would be mapped into half of the intervals. Instead of 65,536 levels, it would se 32,768 intervals. In other words, it would be quantized with 15-bit resolution.

The problem increases as the signal level decreases. A very low-level signal, for example, might receive only single-bit quantization or might not be quantized at all. In other words, as the signal level decreases, the percentage of distortion increases. Although the distortion percentage might be extremely small with a high level, ) 0 dBFS, its percentage increases significantly at low-amplitude levels.

The error floor of a digital audio system differs from the noise floor of an analog system, because in a digital system the error is a function of the signal. The nature of quantization error varies with the amplitude and nature of the audio signal. For broadband, high amplitude input signals the quantization error is perceived similarly to white noise.


right.

[gsharp]

Originally Posted by oky****

"karma police, arrest this man, he talks in math.
he buzzes like a fridge, he's like a detuned radio." ]radiohead[


right.

right

[okydoky]

Originally Posted by gsharp

right

hi,

gsharp, i was going back through this thread to tidy up, and i noticed i forgot to tell you that the synclavier's dacs run at full scale may actually yield a superior performance to more recent technology.

also, i think you were looking for some type of summing device, but decided against it because of the lack of level and pan controls. in case you are interested, neve makes a summing mixer that has that stuff [8016, i think]. people seem to like it.


right.

[yeloocproducer]

Just another reminder...

There's an ignore list via your GS user profile. It takes a bit of work to find it but it's worth it. You never have to look at someone's posts again if you don't want to; take a moment to familiarize yourself with it and your day will be brighter...

Right.

*edit* just saw ubk posted the same thing, but it bears repeating.

[okydoky]

Originally Posted by yeloocproducer

Just another reminder...

There's an ignore list via your GS user profile. It takes a bit of work to find it but it's worth it. You never have to look at someone's posts again if you don't want to; take a moment to familiarize yourself with it and your day will be brighter...

Right.

*edit* just saw ubk posted the same thing, but it bears repeating.

hi,

strange that you would feel so compelled.

actually, the ignore function is really easy to find. if you think that takes work, you are extremely lazy [or showing your ignorance].

i kid, i kid.



right.