Pushing the limit..er

[JohnRoberts]

Which is easy to do ITB, albeit not in (musically meaningful) real time.

That was a popular strategy used in cleaning up noisy vinyl recordings to remove surface noise ticks and pops.
===
BTW wrt synchronizing changes to signal zero crossings. One of my early electronics jobs back in the 70s was at a company performing pitch change for speech compression (speeding up playback of talking book recordings for blind listeners, by normalizing pitch). The early analog process created the pitch shift using a ramped clock driving a BBD analog shift register. The hard part involved splicing together different pitch corrected samples and discarding the unwanted erroneous output. The audibility of artifacts caused by combining the mismatched samples was dramatically reduced by stopping and starting pitch corrected samples at zero crossings. There is both undesirable HF and LF energy involved with splicing away from zero crossings.

A similar benefit occurs when coordinating gain changes with signal zero crossings. In that case it's multiplying the gain change times 0.

JR

 

[scott2000]

Interesting stuff... Never really thought much if at all about zero crossings in general.
Pretty significant information...
Another thing.......
What's the deal with checking and or correcting phase rotation in a waveform? I guess as it relates to more headroom
I have seen it mentioned in a few places.

 

[JohnRoberts]

Interesting stuff... Never really thought much if at all about zero crossings in general.
Pretty significant information...
Another thing.......
What's the deal with checking and or correcting phase rotation in a waveform? I guess as it relates to more headroom
I have seen it mentioned in a few places.

Superposition .... When you assemble a complex waveform by summing multiple pure tones their peak excursion will be larger than the individual components. When these tones are unrelated wavelengths there is nothing you can do about how they combine, but when combining harmonics of a single root pitch the phase relationship between these overtones can impact the peak signal. In theory phase shifting one or more of these components could minimize the peaks. I have no practical experience trying to do that.
===
Back years ago when I was still messing with tuning drums I developed a prototype DSP platform that had its own onboard 16b dac. I tried to quickly parse out the drum head tuning resonances by shotgunning a stimulus sound sample to excite the drum head. This stimulus was made from 10 or 12 spaced frequency sine waves. Each one of these component sine waves had to be 1/10th or 1/12th max output so the combined result would not saturate the output. I blasted the drum head with this franken sound source then read the response coming back with a FFT. It worked more or less, much faster than running a sweep, but I abandoned the project because drummers mainly wanted a cheaper, smaller tuner.

JR

 

[john12ax7]

I've done my fair share of abusing L2 over the years. While there are cleaner limiters these days they all still turn to mush when pushed too hard.

If looking for perceived level you need to get it from additional places other than just pushing the 2bus. Get the mix and individual tracks right so the 2bus doesn't need to work so hard. Some will even go so far as putting limiters on every individual track, then use automation to bring back some excitement.

You can also go in and edit things like drum transients which can eat up the limiter. Clip gain down the big offenders and the limiter will work much better.

 

[scott2000]

If looking for perceived level you need to get it from additional places other than just pushing the 2bus. Get the mix and individual tracks right so the 2bus doesn't need to work so hard.

Yes for sure.
I was kinda looking for the things that let me know when it is working too hard. If the mix or an individual track is decent enough, I'm guessing a limiter can actually go pretty far, just having difficulty finding that. I mean, even trying to use them in a mix environment, I'd probably have the same questions.
Once I get back in town I'll take some of this info and put it to practice...God willing of course..

 

[Script]

You can also go in and edit things

One or two massive peaks of less than 2ms easily pulled down manually from zero to zero crossing.

If a whole series of high peaks, a soft limiter on offensive track.

 

[Jelly]

That was a popular strategy used in cleaning up noisy vinyl recordings to remove surface noise ticks and pops.
===
BTW wrt synchronizing changes to signal zero crossings. One of my early electronics jobs back in the 70s was at a company performing pitch change for speech compression (speeding up playback of talking book recordings for blind listeners, by normalizing pitch). The early analog process created the pitch shift using a ramped clock driving a BBD analog shift register. The hard part involved splicing together different pitch corrected samples and discarding the unwanted erroneous output. The audibility of artifacts caused by combining the mismatched samples was dramatically reduced by stopping and starting pitch corrected samples at zero crossings. There is both undesirable HF and LF energy involved with splicing away from zero crossings.

A similar benefit occurs when coordinating gain changes with signal zero crossings. In that case it's multiplying the gain change times 0.

JR

"The hard part involved splicing together different pitch corrected samples and discarding the unwanted erroneous output. The audibility of artifacts caused by combining the mismatched samples was dramatically reduced by stopping and starting pitch corrected samples at zero crossings."

Correct me if I'm wrong, but isn't this what the ALG3 board in Eventide H949 does? I realize the Eventides use D to A and A to D to store samples in RAM(not BBD), but as far as splicing the pitch shifted samples I mean...?

 

[AlgoRhythms]

In mastering you should really have a reason to do what you’re doing. First what, then why, then how. Like ‘the vocal is muddy’. Why change it? Because it’s clouding the instrumental arrangement. How to fix it? EQ, compression? Then you start turning knobs.

This is about as good advice as I've ever seen. Tape this under your master buss meters so it's always a reminder.

 

[JohnRoberts]

"The hard part involved splicing together different pitch corrected samples and discarding the unwanted erroneous output. The audibility of artifacts caused by combining the mismatched samples was dramatically reduced by stopping and starting pitch corrected samples at zero crossings."

Using zero crossings to stop and start pitch corrected samples reduced the clicks and thumps.

Correct me if I'm wrong, but isn't this what the ALG3 board in Eventide H949 does? I realize the Eventides use D to A and A to D to store samples in RAM(not BBD), but as far as splicing the pitch shifted samples I mean...?

I have no idea**** what Eventide (Richard Factor?) did inside his H949. The difficult problem is splicing together non-time related samples.

The SOTA for good sounding pitch correction was a "rotating head" magnetic tape machine. Not unlike inside VCRs a rotating head machine allows manipulation of the relative speed of the tape WRT the head (this is how slow moving VCR tape can record HF video signals). The four, 90' spaced audio heads inside a rotating head audio machine, pick up LF audio stronger and before the HF audio signals providing an inherent smoothing of the merge between samples. The rotating head pitch shift machine was an exotic and experimental Bell Labs technology. Back in the 70s I saw efforts to mimic the frequency shaping for smoother sounding splices electronically using OTAs to make voltage/current controlled filters.

This is about as good advice as I've ever seen. Tape this under your master buss meters so it's always a reminder.

bus.

JR

[edit: Actually typical digital pitch shift was performed using a bi-frequency method. Data gets clocked into digital memory at one data rate, and clocked out at different data rate. The ratio between input and output clocks determines the pitch shift. Pitch shifting down results in more output time than input so sampling must be reset and excess data discarded, to start over. Pitch shifting up results in less output time than input so the presentation has gaps between samples. Back in the day (early 70s) Dr Lee at Lexicon made a digital pitch shifter for talking book cassette playback using 8 bit A/D/A conversion. /edit]

 

[scott2000]

so just to be clear, are there any possible issues running an older, slower computer that would manifest itself in goofy behavior from limiting with say the Pro L2? I mean, with a couple of eqs and a couple of other things, I could see things getting taxed. Maybe I should look at computer performance/usage to be sure....
It's pretty obvious with kick and sustained bass together but, I can only get a couple of dBs of limiting before I can hear it start to fall apart. But like I mentioned earlier, it's not as obvious to me unless I monitor at a real low level....
It almost has a computer struggling/slight breaking up kinda sound so it made me wonder...
That would actually suck...lol

I know it's most likely the sounds not where they need to be, I just wanted to make sure. It's not like the limiter will work but not be as effective and have weird issues with less processing power? Because it seems to work, just not as aggressively as I anticipated it would..But it's new to me...
Guess I can run through a hardware piece to check behavior too...

 

[ruairioflaherty]

so just to be clear, are there any possible issues running an older, slower computer that would manifest itself in goofy behavior from limiting with say the Pro L2? I mean, with a couple of eqs and a couple of other things, I could see things getting taxed. Maybe I should look at computer performance/usage to be sure....
It's pretty obvious with kick and sustained bass together but, I can only get a couple of dBs of limiting before I can hear it start to fall apart. But like I mentioned earlier, it's not as obvious to me unless I monitor at a real low level....
It almost has a computer struggling/slight breaking up kinda sound so it made me wonder...
That would actually suck...lol

I know it's most likely the sounds not where they need to be, I just wanted to make sure. It's not like the limiter will work but not be as effective and have weird issues with less processing power? Because it seems to work, just not as aggressively as I anticipated it would..But it's new to me...
Guess I can run through a hardware piece to check behavior too...

Yes, if you are only getting a few dB of reduction without things sounding very obviously wrong then your mix is likely not balanced properly. EQ is your most valuable tool when it comes to making things feel loud (not meter loud). Again this all goes back to the quality of your monitoring.

Processing power errors sound very different, digital doesn't fall apart in these kinds of ways, it just stops making sense and glitches.

Consider sending a track to a Mastering Engineer to get an idea what someone good would do, it would likely help you a lot as a kind of reference point. But all roads leads to a better playback situation (does not need to be expensive).

 

[Pinokio]

Nice thread, learning a lot.
How could the waveulator fit in this story. its a zero attack/release limiter.
https://www.waveulator.com/forum/php/viewtopic.php?p=17268

 

[thomasdf]

Nice thread, learning a lot.
How could the waveulator fit in this story. its a zero attack/release limiter.
https://www.waveulator.com/forum/php/viewtopic.php?p=17268

I have been very curious about this design but I cannot find demo or even user review... Ever heard it ?

 

[JohnRoberts]

I have been very curious about this design but I cannot find demo or even user review... Ever heard it ?

maybe look on Wayne's forum

https://www.proaudiodesignforum.com/forum/php/viewtopic.php?t=1331

JR

 

[scott2000]

How could the waveulator fit in this story. its a zero attack/release limiter.

I'm sure it would fit great. Only thing is, if the mix isn't right for what you're after , nothing is going to get it to where a mix that is will be capable of...
I mean, have you seen one of Izotope's patents?..lol

 

[Newmarket]

Superposition .... When you assemble a complex waveform by summing multiple pure tones their peak excursion will be larger than the individual components. When these tones are unrelated wavelengths there is nothing you can do about how they combine, but when combining harmonics of a single root pitch the phase relationship between these overtones can impact the peak signal. In theory phase shifting one or more of these components could minimize the peaks. I have no practical experience trying to do that.
===

fwiw phase shifting of harmonics has been used in synthesis to enable "quasi-square waves" eg IIRC Korg DW8000.

 

[JohnRoberts]

fwiw phase shifting of harmonics has been used in synthesis to enable "quasi-square waves" eg IIRC Korg DW8000.

Are they trying to make waveforms that "look" more like square waves?

AFAIK they all sound the same if not clipping.

JR

 

[Newmarket]

Are they trying to make waveforms that "look" more like square waves?

AFAIK they all sound the same if not clipping.

JR

This was back in the 90s (see DW "Hybrid" synth). But IIRC the opposite. Taking a mathematically perfect square wave and phase shifting the (odd) harmonics to make the result more realisable. Given that rise/fall times are always limited in reality.

 

[JohnRoberts]

This was back in the 90s (see DW "Hybrid" synth). But IIRC the opposite. Taking a mathematically perfect square wave and phase shifting the (odd) harmonics to make the result more realisable. Given that rise/fall times are always limited in reality.

Rise time limited is effectively a one pole LPF. So that would just scrape off some very high harmonic overtones.

Fiddling with phase shift of some overtones to reduce max slew rate sounds good on paper, I don't know about IRL.

JR