Best name for incremental limiter

[kingozrecords]

Hi all, de-clipping can cause a lot of distortion, but I'd realized; if you merely de-clip many times in a slowly decreasing method repeatedly there is much less.

I hope the idea is handy to anyone, but with that said; could anyone suggest the best name for the idea?

Robert

and to further illustrate, how it would work is like as follows (six steps):

1. 25% above ceiling
2. 12.5% above ceiling
3. 6.25% above ceiling
4. 3.125% above ceiling
5. 1.5625% above ceiling
6. 100% ceiling.

[JCJR]

Hi kingozrecords. From your explanation context, am guessing you mean running a signal thru multiple series limiters to avoid clipping when the float input signal exceeds 0 dB full scale peak? (edit: you clarified your question after I posted the reply, making some of my reply look dumber than usual )

If so then that has been done either by a user manually stringing multiple series limiter plugins, and some limiter plugins have featured multiple stages. Dunno if there is a name for the practice. Maybe "Multi-Stage Limiter"?

I doubt that multiple layers of limiting would "just naturally" work better than a single layer of a well-designed limiter. But am ignorant and possibly wrong.

A possible advantage: If a first limiter stage works so "rude" that it introduces new over-level inter-sample peak artifacts as a side-effect of treating the original over-level inter-sample peaks. Requiring a second limiter to clean up artifacts introduced by the first limiter. Then perhaps a third limiter needs to clean up the second limiter's artifacts! Ad infinitum. Introduce cats to control the rat population, then introduce dogs to control the cat population, then introduce wolves to control the dog population!

I was interested in "sane conservative" limiting. Perhaps radical balls-to-the-wall limiting to create Louder-Than-Dammit masters would need more desperate action. Dunno.

So far as I know, the term "De-Clipping" differs from "Limiting". So far as I know, De-Clipping is a process which tries to repair an otherwise-ruined audio file--
_1_ Recorded or bounced too hot into an integer format. Permanent file clipping which can't be removed merely by reducing the gain.
_2_ Or any recording, integer or float, which got clipped in the Mic, preamp, or other parts of the recording input chain. In that case you could have a well-behaved float file maybe peaking no higher than -24 dB or whatever, but permanently horribly distorted because of clipping which occurred before the audio got digitized and recorded.

A De-Clipper attempts to un-do the damage. It MAY be easier in a non-realtime offline process rather than a realtime plugin. So that the process can make multiple random-access read/analyze/write passes as it tries to guess what the original undamaged waveforms looked like.

Maybe there are good real-time plugin de-clippers. Am out of touch. I wrote some non-realtime de-clipping long ago. IIRC never tried to make a realtime de-clipper.

[kingozrecords]

Yeah, well I've been surprised, it would seem that each step adds aliasing, but then the next step removes the aliasing it created and because it's proportionate (that's the name - Proportional Limiter :nods) so when it's all said and done; it's smooth like an envelope.

It's achieved just by setting min and max for the clip; but by doing it in this manner before a conventional limiter (with high release time). Give it a try, it's simple and boss.

[BertKoor]

I know what clipping is.

Is what you call "de-clipping" the reverse operation?
How does that work? Patent-pending technology I presume?

[kingozrecords]

Just a brickwall; it does not even need an envelope. But yes, simple as this may be; it works better than most envelopes I've tried. lol@patent.

If more people would chime in with simple ideas that work we wouldn't be listening to egos telling us how smart they are, sharing ideas so complicated that it's hard to tell whether they're wrong or not lol.

Good luck with this working idea.

[BertKoor]

If you're brick-wall limiting, then de-clipping == clipping

[JCJR]

Just a brickwall; it does not even need an envelope. But yes, simple as this may be; it works better than most envelopes I've tried. lol@patent.

If more people would chime in with simple ideas that work we wouldn't be listening to egos telling us how smart they are, sharing ideas so complicated that it's hard to tell whether they're wrong or not lol.

There have been various "mastering anti-aliased clipper" plugins. I do not recall names but you can search around on the gearslutz mastering forum to find numerous discussions. Some folks think it is a wonderful idea but I'm not among them. Which does not have relevance as to whether it is a worthy concept-- I just wouldn't want to butcher my music in that fashion.

An older but cogent paper: https://static1.squarespace.com/static/ ... +Radio.pdf

However, with the rise of the clipped CD, things have changed. The phase rotator radically changes the shape of its input waveform without changing its frequency balance: If you measured the frequency response of the phase rotator, it would measure “flat” unless you also measured phase response, in which case you would say that the “magnitude response” was flat and the phase response was highly non-linear with frequency.

The practical effect of this non-linear phase response is that flat tops in the original signal can end up anywhere in the waveform after processing. It’s common to see them go right through a zero crossing. They end up looking like little smooth sections of the waveform where all the detail is missing—a bit like a scar from a severe burn.

This is an apt metaphor for their audible effect, because they no longer help reduce the peak-to-average ratio of the waveform. Instead, their only effect is to add unnecessary grungy distortion.

That older paper is about radio transmitter processing, but anything past the clipping stage can add phase shifts and produce above-described "scarred audio" results. Such as further processing while creating a "master mix", processing related to creating lossy compressed distribution files, EQ and other software processing done by a user's audio player device or program, such as silly microsoft, dolby or apple "audio enhancements", EQ and other effects of the playback amp, and phase shifts added by the speakers and speaker crossovers.

IMO the popularity of mastering with clipping tools is one reason modern pop songs sound unbearably distorted, but some folks like it and therefore I suppose clipping is not all bad since it makes some people happy.

[mystran]

If I understand your concept correctly, you're basically thinking of some sort of maximizer.

I would suggest a slight different approach though: use the same threshold for all stages, but start with a slowly-acting "loose" limiter (ie. one with some attack time, no lookahead). This won't give you brick-wall, but it will bring down segments that stay too high for longer time. Then use a slightly faster-acting limiter to do a second round, but still with some finite attack. This still won't give you brick-wall, but it will bring short-term transients closer to your limit. Finally do a third round with a very fast-acting lookahead brickwall limiter (eg. I'd suggest finite attack+release that's about 1ms at most combined); at this point we're basically clipping, but we can still smooth it out a bit.

With suitably-tuned attack/release times for the long- and medium-scale stages, this can work quite well. The trick is to find time-constants that manage to fool the ear to thinking the dynamics haven't really changed (at least assuming you don't push it too hard).

ps. To answer the inevitable question of why long-scale first: this makes the medium-scale limiting action less obvious, by allowing it to act less agressively if the signal stays consistenly high-level for longer periods of time, making "gain-riding" not nearly as obvious.

[kingozrecords]

Yes, I had made the same realizations. But I'd wanted to avoid any envelopes and wanted this merely as a tool so as to employ the math min and max function, so as to later be able to limit what was left with an envelope.

It makes work easier on the envelope, I thought it a simple strategy and maybe it's useful to someone.