Multiband Compressor Idea

[Gold]

I've had a concept for a multi band compressor for a bunch of years now. I don't have the skills to do the design and it wouldn't make sense as a one off. I've been unsuccessful at enticing anyone to take this on.

If there are any manufacturers lurking out there who think a multi band compressor makes sense for them please get in touch. I'm not looking to make any money. I just want to see this thing made.

The only current competing products are the Maselec and the Tube Tech. I don't think the Tube Tech is very popular. I think my concept brings something new to the table.

Hope springs eternal.

 

[JohnRoberts]

maybe take a dbx multi band expander from the 80s and connect it backwards...  :

it seem there are several broadcast processors that do some variant on that..  I've never done more than two bands (de-ss and wide-band comp/limit) inside one box, but early broadcasters ran signals through speaker crossovers then separate comp/limiters on each band-pass before recombining... as you can imagine it didn't sound very good,  but that didn't stop them. It was louder and sounded "different",.

JR

 

[Gold]

It’s fashionable to hate multiband compressors which is probably why there is a lack of interest.

I love multiband compressors. Lots of them sound terrible. It’s not easy to get right which is probably another reason for a lack of interest.

 

[weiss]

It’s fashionable to hate multiband compressors which is probably why there is a lack of interest.

I love multiband compressors. Lots of them sound terrible. It’s not easy to get right which is probably another reason for a lack of interest.

i think people use it wrong. of course it changes the character of a whole song if you smash it on a mix. but putting it on a bass guitar or a piano makes much more sense.

I wouldn't say no to a linear phase multiband comp though..  ;D

 

[Gold]

This is kind of like a Maselec and GML had a baby.

 

[pucho812]

I had a known mastering engineer tell me he hates multiban compressors because they take away from mix.

He is known but known does not equal good.

His argument had to do with splitting the music with a crossover and then summing.

While I get his point I am not about to say anything pro or con.

Paul what did you have in mind? I know some guys who would love to hear this.

 

[Gold]

The knobs don’t turn themselves.  It’s perfectly possible to live life without one and make goood masters. Coming up with some pseudo technical mumbo jumbo as to why they suck is another thing.

It’s a five band. Fixed crossover points for the high and low band. Two crossover choices between bands two and three and bands three and four.

Dual detector VCA with separate controls for the peak detector and the RMS detector.

I can post a drawing of the controls of one section tomorrow.  A picture being 1000 words.

 

[pucho812]

Hmmmm.  Sounds interesting.  Yeah let's see it

 

[JohnRoberts]

It seems I've participated in discussions about a hypothetical multi-band dynamics processor before, either here or on Wayne's forum, or both. If you plan to divide the dry audio path into multiple bands, process, and recombine them again, this is not trivial and the optimal crossover topologies for loudspeakers are not the same as ideal for this.. I would used "derived" crossovers that have a less steep slope (in one passband), but allow you to put humpty back together again neatly (ignoring the gain processing)... (1-x)+x=1. Changing the gain of the bandpasses will hinder perfect reconstruction  (1-x)+Nx  does not equal 1, but whatever.  ??? 

===

Back last century I though about designing entry level sound system mixers (for beginners) that used dsp and were smart enough to recognize the different musical instruments or vocals and EQ the individual channels to deliver a target sonic template and final mix. My conundrum was that the technology to do this was too expensive to sell to entry level customers, and the professionals who could afford it would never tolerate a robot mix person inside the (top)box.  :

JR

PS: I would be one of those curmudgeons unenthusiastic about an after the fact "fix the final mix box", but I am not your target demographic. 

 

[iampoor1]

It seems I've participated in discussions about a hypothetical multi-band dynamics processor before, either here or on Wayne's forum, or both. If you plan to divide the dry audio path into multiple bands, process, and recombine them again, this is not trivial and the optimal crossover topologies for loudspeakers are not the same as ideal for this.. I would used "derived" crossovers that have a less steep slope (in one passband), but allow you to put humpty back together again neatly (ignoring the gain processing)... (1-x)+x=1. Changing the gain of the bandpasses will hinder perfect reconstruction  (1-x)+Nx  does not equal 1, but whatever.  ??? 

===

Back last century I though about designing entry level sound system mixers (for beginners) that used dsp and were smart enough to recognize the different musical instruments or vocals and EQ the individual channels to deliver a target sonic template and final mix. My conundrum was that the technology to do this was too expensive to sell to entry level customers, and the professionals who could afford it would never tolerate a robot mix person inside the (top)box.  :

JR

PS: I would be one of those curmudgeons unenthusiastic about an after the fact "fix the final mix box", but I am not your target demographic. 

Is, let's say,  is a 12 or 24db per octave linkwitzreily crossover not ideal?

 

[JohnRoberts]

Is, let's say,  is a 12 or 24db per octave linkwitzreily crossover not ideal?

good for speakers because it is dealing with a 3 dimensional problem... the 4 pole L/R is so far out of phase at the crossover point that it is back in phase (+180' - (-180')= 0' at least for a sine wave. ... and the two passes being down -6dB at crossover is 1/4th power so good for speaker driver protection.

Look up derived crossovers... you subtract one several pole HPF or LPF from unity to make the other pass band. Slope is only -6dB /octave for the derived output, but the two passes sum back together perfectly (unless you change the level with a dynamic processor).  :

JR

 

[gyraf]

Interesting!

 

[hazel]

Maselec uses 6dB/octave slope. Quite gentle.

5 bands would be great. Maybe 3 would be more feasible.

How about splitting bands with multiple secondaries transformer? Or ICs make more sense here?

 

[gyraf]

The splitting is not the problem - the re-combining is.

I've tried attacking this from so many different angles, with results that were like "meh..." at best of times.

Ended up in seriously doubt if it can be done (in a good way) at all - which is why I'm very interested in your idea..

Jakob E.

 

[weiss]

What's the "maselec-approach"? The device seems relatively popular.

 

[Gold]

What's the "maselec-approach"? The device seems relatively popular.

6dB/ oct crossover slope. Like John said it’s the only way to get a mathematically perfect sum. I think tha’s the right approach. I don’t hear the crossover in the Maselec.

 

[Gold]

 

[Gold]

No one has anything to say about it?

 

[scott2000]

Tubes?

 

[Gold]

Tubes?

THAT fest. VCA and probably line receiver and driver. This thing would be big enough without tubes and transformers. This is supposed to be ultra clean. This is not the place for mojo. I don't use tubes anyway. The only piece if tube gear I currently have in use is a Waveforms VTVM.