What Richard meant when he said that computers cant distort audio?

[Ragnar]

irony ends up being that poor analog to digital conversion causes the most warmth. Or would that still be analog's fault. So confused

[Ragnar]

uh so I do try something from time to time, in a digital environment/laptop u can invert the spectrum (high pitches become low pitches sort of thing, the spectrogram would be a mirror image) simply by inverting every other sample I guess? So I'm messing around with adding subtle distortion in this mode and then flipping the spectrum back with another pass of the effect. Problem is I dunno wtf it's doing if it's generating some sort of harmonics or not. I guess I'll try with a sine wave or somethin

 

tried it, only thing is I can't really eyeball a spectrogram and go THOSE ARE EVEN HARMONICS so uh either way the spectrogram is different than without inversion?

 

Edit: SPECTRAL inversion, not just inversion like changing 1 values to -1

[Ragnar]

agh I just tried my fractal waveshaper again on 100% wet and it's hella warm somehow without sounding crackly/digital/just a dither

[ZoeB]

You have to stop thinking in terms of digital waveforms looking like staircases. They don't. They're run through a lowpass filter which doesn't just smooth them out arbitrarily, it smoothes them out into the original waveform. The only information you lose when going to digital and back is frequencies that are higher than the lowpass filter. There's no practical difference between running a signal through a 20kHz lowpass filter, digitising it, analogueising it again, running it through another 20kHz lowpass filter, and hearing it, than there is to running it through a 20kHz lowpass filter and hearing it. (Assuming a decent bit depth, as we're just talking about the sample rate here.) As we can't hear anything over 20kHz, us humans haven't lost anything in digitising the signal.

[spunktronics]

Summing of signals on a analogue desk can not be achieved the same way on a computer.

Harmonic distortion is a beautiful creature!

 

wanna try it just buy a cheap soundcraft and have a play.

[rst]

You have to stop thinking in terms of digital waveforms looking like staircases. They don't. They're run through a lowpass filter which doesn't just smooth them out arbitrarily, it smoothes them out into the original waveform. The only information you lose when going to digital and back is frequencies that are higher than the lowpass filter. There's no practical difference between running a signal through a 20kHz lowpass filter, digitising it, analogueising it again, running it through another 20kHz lowpass filter, and hearing it, than there is to running it through a 20kHz lowpass filter and hearing it. (Assuming a decent bit depth, as we're just talking about the sample rate here.) As we can't hear anything over 20kHz, us humans haven't lost anything in digitising the signal.

 

i don't usually participate in any discussion that involves the words digital and analogue as most of the times those will make me angry. however i wanna thank you at least once for repeatedly dispelling dsp myths. kudos!

[rst]

that said i completely agree that analogue gear has definite merits, analogue clipping / saturation being one of them. many times though ppl seem to feel the need to justify their preferences with unfounded half-truths. instead of simply saying "it sounds nice to me."

 

bit OT, i guess, soz. also posting in a row: couldn't edit.

[ZoeB]

Thanks. :)

 

Yeah, I'm really not up on what happens when you overdrive an analogue signal. It clips in a nice way, but what's happening mathematically, and how close careful digital emulation of it is, isn't clear to me. All I know is that you need to go out of your way to emulate it if you want that sound in your digital setup, as it doesn't do it inherently, and that the reason for that has nothing to do with the sample rate or bit depth, it's to do with how digital systems work.

[khov]

digital distortion generally just isn't nice.. analogue distortion can be pleasing to the ear/nostalgic

 

i beg your pardon ?

 

[Guest Wander]

^That's a very nice track, cheers.

[rst]

wow. thx for posting my fav pita track, khov!!!

i love his early stuff. tracks like this are much more punk than most punk rock can ever hope to be.. he really embraced imperfection there.

 

edit: in fact i think this is what black metal should be!

[rst]

btw. pita nowadays uses guitar/bass amps with his electronics when playing as ktl. apparently soma suggested it to him and he liked it ;-)

[enki]

I took the "computers can't do distortion" comment in a broader context, to include all of the various non-linearities (i.e. 'distortion') that are present in analog signal chains that typically aren't modelled in DSP systems, or which aren't modelled well. You'd probably have to do some pretty intenstive physical modelling to capture the effects of crosstalk between circuits, sagging supply voltages, etc. Throw mechanical feedback into the mix (as in a guitar-amp situation), and things start getting intractable.

 

So, I wouldn't take it as a hard "can't" (in most cases), but more of a "don't" due to the engineering effort and computational resources needed to reproduce what essentially comes "for free" in analog systems.

[khov]

btw. pita nowadays uses guitar/bass amps with his electronics when playing as ktl. apparently soma suggested it to him and he liked it ;-)

 

yeah, its true, 2015 should see some Pita release !! i saw him live last month & the new tracks are truly awesome

[rst]

 

You'd probably have to do some pretty intenstive physical modelling to capture the effects of crosstalk between circuits, sagging supply voltages, etc. Throw mechanical feedback into the mix (as in a guitar-amp situation), and things start getting intractable.

 

yeah, i guess these artifacts is where it gets really interesting & unique! (instead of more transparent processes like tape saturation e.g.)

 

 

btw. pita nowadays uses guitar/bass amps with his electronics when playing as ktl. apparently soma suggested it to him and he liked it ;-)

 

yeah, its true, 2015 should see some Pita release !! i saw him live last month & the new tracks are truly awesome

 

looking forward to that. pita is a unique artist. in fact, i'd rather afx (had) worked with him instead of hecker!!

[Brisbot]

Lots of good information in this thread. Just wanna know if anyone has any good in depth readups on distortion?

[ZoeB]

 

 

Bob Katz explains it in "Audio Mastering":

"In a floating point system, you can break all the rules: floating point can literally ignore the individual levels in the chain. It's possible to drop the signal level by 100dB, store the signal as a floating point file, then open the file, raise the gain 100dB and get back the original signal, with little or no deterioration. Or vice versa"

 

This sounds like an odd thing to say. Bit depth is important, not whether you're using floating point arithmetic or integers. A bit depth of 4 is generally pretty lousy, regardless of whether you're counting from -7 to +8 in increments of 1, or 0 to 1 in increments of 0.0625. A bit depth of 16 is much nicer, regardless of whether you're counting from -32767 to +32768 in increments of 1, or 0 to 1 in increments of 0.00001525878. But hey, I'm a simple creature and like programming with integers. :)

 

Bit depth is important indeed, and that's why you won't see FP processing happening in anything less than 32bits. For instance, Reaper lets me choose between 32bit or 64bit FP as the bit depth for uncompressed audio files.

 

Thanks. I was rewatching Monty's A Digital Media Primer for Geeks and get it now. When audio engineers talk about floating point variables, they're specifically talking about IEEE 754 ones, set to at least (and probably only) 32 bits. You're right, thank you for clearing that up! Incidentally, while we're talking about Bob Katz, his video on the loudness war is pretty interesting. It's kind of ironic that just as high quality digital encoding of waveforms allows a greater dynamic range, people are also able to listen to music outdoors more, where the sounds of the music have to compete with the sounds of the outside world, and hence people prefer a reduced dynamic range.

 

As pointed out in the Bob Katz video, check out the Sony PCM-1610's meter:

 

 

The 0 dB indicator is somewhat arbitrarily placed 20 dB from the actual top. Unlike analogue tape, a digital recording won't slightly distort until you reach the hard limit and it gives nasty clipping, so people try to get as close to the top as they can, but that's pretty silly. As this device emphasises, as long as you're not so close to the bottom that the signal to noise ratio becomes noticeable, you don't need to get as absolutely close to the top as you can. Just 3/4 or 4/5 of the way is more than enough, leaving plenty of headroom.

[rst]

 

The 0 dB indicator is somewhat arbitrarily placed 20 dB from the actual top. Unlike analogue tape, a digital recording won't slightly distort until you reach the hard limit and it gives nasty clipping, so people try to get as close to the top as they can, but that's pretty silly. As this device emphasises, as long as you're not so close to the bottom that the signal to noise ratio becomes noticeable, you don't need to get as absolutely close to the top as you can. Just 3/4 or 4/5 of the way is more than enough, leaving plenty of headroom.

 

as with a lot of things it wasn't silly in the past when dsp was little evolved and the signal was prone to degradation from all the computations it is put through when mixing etc. but with virtually every DAW computing at 64 bits internally it's not necessary any more. (keep in mind that is what's important, really, not the bit depth of an actual recording.)

 

still, i guess an actual mathematical imprecision will still happen when you put a dozen fx on a track which runs through several gainstages on its way before it's finally summed with three dozen other tracks... but, i mean, we do mix BECAUSE we want the sound to change, right?

 

also i read that 8 bits of the 64 are actually reserved for headroom BEYOND -1. ... 1., or min ... max, for internal computation, mind you, not the i/o stages. can't rem the details though...

 

 

ps. 39 bits in reaper, anyone???

 

 

 

 

edit: oh shit. post no. 303!!! & it's all about digital...

[MisterE]

i feel like he was 'taking the piss'. people have distorted samplers, including him, for ages. not only the converters but also using them for their 'gritty sound' which usually comes down to low sample rate and bit depth. and that may be inharmonic but it's still a form of distortion and it's very digital. how about all the digital dsp stuff he used? i'm sure he's used a lot of digital distortion and loved it while he was using it. the statement 'digital can't do distortion' is so general that it just seems like a troll statement. someone saying that would know that theres a digital vs analog argument always going on, and it seems like that statement was just made to throw gas on that fire for amusement, to me.

 

if he had specified which type of distortion (subtle harmonics from a mixing desk for example), or clarified that he was stating an opinion that he prefers analog distortion in a particular situation, then it would actually mean something. as it is it's entirely meaningless and imo contradicted by his own use of samplers and digital distortion in the past. or is he actually saying he doesn't like go plastic, dnb, eventides, or dsp in general? i think he said it because he wanted to see people trying to decipher what it meant.

 

that or he just said it without thinking too much about what he was saying, and wasn't asked to and didn't think to explain what he actually meant specifically.

[ZoeB]

as with a lot of things it wasn't silly in the past

 

Good point. So it's vestigial, then. The ham butt problem.

[rst]

 

as with a lot of things it wasn't silly in the past

 

Good point. So it's vestigial, then. The ham butt problem.

 

now what about vestigial 28 organ?? is it exactly like a ham butt? it's not distorted, so i guess no.