The 31 band broadcast limiter

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

emrr

Well-known member
Joined
Apr 12, 2006
Messages
8,525
Location
NC, USA
Courtesy of Radio World

http://www.nxtbook.com/nxtbooks/newbay/rw_20111116/#/36
 
Interesting, their pitch is to allow stations to drive more loudness however loudness is traditionally achieved with heavy AGC, some compression and a minimal amount of limiting (due to the limited no of limit bands). This Limiter alone is not the full picture, one would think with such granularity in a limiter you could back off the compression and let the limiting provide most of the peak control, but there is the general gain riding required that an AGC typically does, to push the average levels up and it is the high average level that has most impact on loudness. You really need to look at the whole chain here because one without the other is just not going to cut it.

Still, this limiter coupled with a multi band agc might just work, the real test is the listening test, yes our ears can be fooled I agree, but I do know that the ear cannot be fooled by science and theory. I went down this track, bought some Omnia's over the usual Optimods, on spec alone, the Omnias coming with one more band and every conceivable control to mankind, so in theory my brain said I can make this sucker sound better than the Optimod which has limited user control ..... wrong!, that Omnia really sucked, spent months tweaking in small steps and listening, still sucks.

We like progress and I would like to hear this thing, but coupled with what other products?
 
The thing is that the multiband limiter does not provide the necessary protection of the transmitter.
In fact, the more bands, the less protection.
Let's consider this 31 band thing. If one wanted to rely on it for transmitter protection, the energy in each band should be 1/31th of the total energy allowed. If you put pink noise in there, each band would heve equal energy and the sum would be the maximum permitted. Now since most audio program is not pink noise, the total energy would be lower - too low for the perceived loudness stations want to achieve. Typically, with pop/rock, most of the energy will be in 3-4 bands, giving a total energy of about 8dB below maximum. In comparison, a 2-band limiter would be only 2-3 dB below max.*

So the system will always rely on a peak limiter at the end of the processing chain. The multi-band system has been designed to provide more ear-friendly spectrum-shaping before peak-limiting. The most important energy increase is the gain applied between the multi-band limiter and the final peak-limiter.

There is nothing new in the concept of this 31-band limiter, just an evolution on the 6, then 11 band Optimod.
The problem with multi-band limiters is the band-filtering. the interaction between bands has an undesirable effect that one band may tend to disappear completely when it's even slightly compressed and the adjacent bands are not.
Two generations of broadcast designers have worked on this problem, using various tricks in the side-chain control and tweaking band-pass filters.
Today, the ability of creating almost perfect brickwall digital filters brings an answer to that problem.

The claims of "subtle audio details being revealed" is probably true in comparison with a limiter with a lesser number of bands, but I'm not sure when the signal has passed the crushing gate of the final peak-limiter there will be much audible difference. As usual, the law of diminishing returns prevails.

*These figures are to be taken with a large handful of salt; I haven't included the effects of pre-emphasis and initial spectral balance.
 
So funny to see this brought up here now - this seems to be a technology related to a new product we're currently developing over at Gyraf, the G21 CutterClipper. I had no idea that something like this existed already - although it always seemed too strange that no one had the idea before :) So much for my literature-search skills...

Jakob E.
 
abbey road d enfer said:
but I'm not sure when the signal has passed the crushing gate of the final peak-limiter there will be much audible difference.
Very true, I made the point re-AGC and similarly a final peak limiter or clipper is necessary as the end stop before the transmitter, so much of what is likely to be the end product will be the result of interaction between these devices. Bring it on, lets hear it, but in a complete system, not in isolation
 
radiance said:
Isn't this what Paul Frindle does with his DSM pluggin?
Somewhat; the DSM is FFT-based so it has probably 1024 or more bands of equal Hz bandwidth (about 20Hz wide at 20Hz center, one-octave-wide, and also 20Hz wide at 20kHz center, about 1/100th-octave wide).
Also unique is the capture mode, that is based on capturing in real-time a complete set of 1024(?) thresholds.
But the overall effect is probably similar.
 
Update:

G21_AudioShaver.jpg


..The Gyraf G21 AudioShaver prototype 1...

Stereo and ENTIRELY passive, i.e. no power connector on the back  ;D

Yes, there's still some work to do - like making the low-cut frequencies useable.

Jakob E.
 
It's kinda' like emphasis prior to processing - clipping high frequencies at lower "thresholds" than low frequencies.

subjectively nice  :)

At 3dB/oct for "pink" (like pink-noise filter or "thrust" in compressor sidechain) and at 5dB/oct for "red"
 

Latest posts

Back
Top