mixIO 1.3

[IXix]

nothing inside of buzz (before master) clips.

Are you sure about that?

[UNZ]

nothing inside of buzz (before master) clips.

Are you sure about that?

yes ? its all floats, why would it clip ?

[fetus ink]

nothing inside of buzz (before master) clips.

Are you sure about that?

I know I'm not sure about that. Here's a pretty massive clip generated with a patch in Qsamo. Nothing more. And definitely clipping at Mr. Jedshiva, which is obviously before master.

[UNZ]

nothing inside of buzz (before master) clips.

Are you sure about that?

I know I'm not sure about that. Here's a pretty massive clip generated with a patch in Qsamo. Nothing more. And definitely clipping at Mr. Jedshiva, which is obviously before master.

of course it clips BECAUSE YOU ARE RUNNING AT +6 DB INTO MASTER! IT DOES NOT CLIP --BEFORE-- MASTER TOUGH!
to proove this set jedshiva -> master connection to -10db or so: it will not sound clipped.
that jedshiva shows a lit clip led is kind of wrong. this is exactly the confusion / missconception i want to avoid. you just prooved my point that displaying a clip led is harmfull.
only the master VU in the toolbar should have a clip led.

[Joachip]

I guess fetus ink just proved your point Antonio. I guess the "clip" LED does indeed mislead people into not understanding how floating point numbers work.

[ecslist]

unz is crazy UNZ is mad

[fetus ink]

Well, I admit. I don't understand then. I'm not an audio guru. I just made a patch, and ... as far as I can tell am not 'increasing by 6dB anywhere' , and well, it clips. So don't cap me, just inform me.

[UNZ]

Well, I admit. I don't understand then. I'm not an audio guru. I just made a patch, and ... as far as I can tell am not 'increasing by 6dB anywhere' , and well, it clips. So don't cap me, just inform me.

there's only three ways this is possible:
1) your qsamo patch outputs at +6db (this is the most likely case, you probably have a lot of drive or whatever)
2) jedshiva gains the signal up +6fb (very unlikely)
3) the jedshiva connection to master is +6db (also unlikely i guess)

send me the bmx and i'll find out.

the thing about floating point numbers is this: as long as the signal arriving at the master doesn't go above 1.0, you can add or remove gain as much as you want in your signal chains. +2.0 +3.0 +100.0, it doesn't matter much as long as you gain it down again before master

(disregarding machines that clip internally (say distortion) or that can't handle float correctly or have a range they work at best such as -1.0 to 1.0)

[IXix]

Sorry, for some reason I had it in my head that the samples were int type. I was sure you used to have to do an extra calculation to turn them into normalised floats. I must mistaken.

Anyway, if the samples are normalised floats then yeah, you have to really abuse them to cause clipping.

[UNZ]

Sorry, for some reason I had it in my head that the samples were int type. I was sure you used to have to do an extra calculation to turn them into normalised floats. I must mistaken.

Anyway, if the samples are normalised floats then yeah, you have to really abuse them to cause clipping.

they are not ints, but you still have to divide (and later multiply) the "buzz floats" by int16.max thats probably where your memory comes into play.

[fetus ink]

send me the bmx and i'll find out.

Won't have to, I already understand it now. It was a resonance issue. You do have a point that its misleading, but as far as I always knew, a "clip" was when the meter hit the max and distorted the sound. Whether thats happening in the machine or the master, theres still a threshold being passed thats effecting what I hear. Something showing that you had "gone over in volume" if even for one second is still a good way to show you that "hey, you went over the peak at some point, you may want to turn that the fuck down" ... or not. Depending on what kind of mess you're going for. This float stuff is obviously something I know nothing about. What I do know, is that I like knowing when something 'exceeded its peak' ... is that the better phrase to use here? I know the machine itself may not actually be clipping the sound, but either way, there is a max threshold in the master, and if its reached, it sounds clipped/distorted, and I think thats something worth knowing about. Having it on a mixer is a good way to show you exactly what chain is "exceeding its peak".

I also know that Qsamo is pretty good at exceeding said peak when playing around, and a 'master out' parameter would be great for saving with presets so I don't blow out my ears when flipping through them. I can never seem to adjust the sound under that peak level and keep the sound I originally wanted in the few cases its happened.

-added- I guess what I'm trying to say is that if you add the 'clip' feature, optional or not, I'll still use it even if its not 'what its used for'. Its like using a butter knife instead of a screw driver. Sure, it might not the right tool, but it works for me. And it helps me to get the job done.

[ekral]

I am really looking forward for the new mixIO - especially that level metering is nice. Is there any chance to implement clip detection?

i thought about adding clip leds, but i probably remove them again because it doesnt really make sense: nothing inside of buzz (before master) clips. dont want to give the impression that stuff clips when it doesn't. maybe i'll make it optional to enable clip leds..

Thanks for answer

Btw. I am just curious, what happens if it goes above 20db and what is the relation between floating number and decibels?

[UNZ]

I am really looking forward for the new mixIO - especially that level metering is nice. Is there any chance to implement clip detection?

i thought about adding clip leds, but i probably remove them again because it doesnt really make sense: nothing inside of buzz (before master) clips. dont want to give the impression that stuff clips when it doesn't. maybe i'll make it optional to enable clip leds..

Thanks for answer

Btw. I am just curious, what happens if it goes above 20db and what is the relation between floating number and decibels?

to understand the K-20 meters in mixio (there will be others too) you HAVE to read this:
http://www.digido.com/level-practices-p ... ystem.html

It's just where the meter ends. there's nothing special about +20db, but in this case its defined as "Full Scale Digital" (FSD), you can also define +12db or +14db or 0 (or really anything) as that. It's just a convention. Its a way to calibrate your monitoring system and help you mixing when you keep the RMS (NOT PEAK) around 0 (and 0 corresponds to a fixed SPL) and the peaks go up to + whatever (how much dynamics you want). Its a way to judge the RMS to PEAK ratio (as in how compressed is the signal)


NOTE THAT THE TOP IS ALWAYS FULL SCALE DIGITAL!
All of this obviously only makes sense when mixio is the last machine before master or you don't adjust the amp after mixio at all. (in which case a clip led on the master output of mixio would make sense i guess)

the relation with floating point and decibels is such that you usually define 1.0 to be FSD. I'm not sure what you mean with this question, but you can use an amplitude to decibel converter. 2.0 would be +6db, 0.5 would be -6db etc

[IXix]

You know, I've never bothered with mixer machines but this one looks like it might change my mind.

[ekral]

So in the case of K-20:

maximum is 83 + 20 = 103 dB (amplitude 141,253.75446)
minimum is 83 - 24 = 59 dB (amplitude 891.25094)

And as the amplitude in Buzz is represented in single precision floating point number, so there is enough room for amplitude and there is no clipping before master.

Do I understand that right?

[UNZ]

no you got it wrong, in a K-20 meter:

+20 is FULL SCALE DIGITAL = 0dB = 1.0
0 is FSD-20dB = 0.1

the 83 is SPL (Sound Pressure Level) and is measured from the sound comming out of your speakers with a physical loudness meter. The point of this whole exercise is repeatable monitor gain. They want you to calibrate your monitor loudness (with its gain knob) so that 0dB RMS (when playing pink noise) on the K-Meter represents 83dBC SPL comming out of your monitors.

and yes, single precision floats are PLENTY enough. Passing sound around between machines in Double (as some other hosts do) is just silly and unnecessary. Double is only needed for SOME internal calulations in some machines.

[ekral]

no you got it wrong, in a K-20 meter:

+20 is FULL SCALE DIGITAL = 0dB = 1.0
0 is FSD-20dB = 0.1

the 83 is SPL (Sound Pressure Level) and is measured from the sound comming out of your speakers with a physical loudness meter. The point of this whole exercise is repeatable monitor gain. They want you to calibrate your monitor loudness (with its gain knob) so that 0dB RMS (when playing pink noise) on the K-Meter represents 83dBC SPL comming out of your monitors.

and yes, single precision floats are PLENTY enough. Passing sound around between machines in Double (as some other hosts do) is just silly and unnecessary. Double is only needed for SOME internal calulations in some machines.

thanks, now I understand that little bit more . I still have some questions, but I will try to study it on their web. Anyway I am really looking forward for the new mixIO

[UNZ]

OK, IT'S READY FOR TESTING!

http://outofobscure.dyndns.org:88/mixIO_1_2_Beta_1.zip

[zephiris]

error on connecting a machine to it, or just doubleclick-opening gui

[UNZ]

i'll make this big and bold, since a number of people found it difficult to copy two files apparently....

make sure you replace BOTH .dll
mixio.dll
mixio.gui.dll