Compression

[Jeff Carter]

Re: Compression

The digital compressors are more forgiving because of how they handle the waveform by "looking ahead" and lowering the entire intact signal.

Isn't "look-ahead" more of a recording/post-processing thing? I don't see how it could be applied to live sound without introducing latency.

 

[John Chiara]

Re: Compression

Isn't "look-ahead" more of a recording/post-processing thing? I don't see how it could be applied to live sound without introducing latency.

You only need microseconds of look ahead to work. I am sure they build in a set latency for all the available DSP processors.

 

[Tim McCulloch]

Re: Compression

Isn't "look-ahead" more of a recording/post-processing thing? I don't see how it could be applied to live sound without introducing latency.

There is AD/DA and processing latency to start with, and every thing you add creates more. Giving up a few samples for look-ahead is fairly trivial.

 

[Eric Cagle]

Re: Compression

Isn't "look-ahead" more of a recording/post-processing thing? I don't see how it could be applied to live sound without introducing latency.

That is a good topic for disussion on its own. I am not an expert but I do remember reading that my first digital console did all of the processing during a 12 sample window or something like that. The thing I have noticed in the field is the sound of digital board compression is really smooth without many of the artifacts of the analog compressors I usually use. It sounds more like lowering a fader. The first time I noticed this was on an A&H iLive a while back when they first came out.

 

[John Roberts]

Re: Compression

You only need microseconds of look ahead to work. I am sure they build in a set latency for all the available DSP processors.

Cough... microseconds?

The only practical stand-alone live audio rack product using look-ahead was a RANE noise gate, that was able to apply a short delay to the path so a signal that was gated off, could be turned on and the circuit would not smear the leading edge of the sound and click while turning on (circuits can't turn on and off instantly).

With general dynamic processing (like gain reduction) for the look-ahead to be really useful, it needs to be long enough to ramp a gain change slowly so that gain change itself doesn't become an audible distortion. If gain changes are fast in the context of a single waveform period, it just becomes distortion to that cycle of the waveform. Look ahead lets you reduce gain slowly before the transient gets there. Even if you used rapid gain reduction with look ahead, the look-ahead anticipation time would need to be at least a quarter wavelength of the lower frequencies being processed. To not distort the first cycle of a burst. Again more than microseconds.

The RANE noise gate exhibits 1.6 milliseconds of total latency (so look-ahead is less than that). Some were apprehensive about even this much delay in live sound but it's only equivalent to spacing the mic a foot and change further away.

For general dynamics processing useful look-ahead could be full seconds for very slow gain changes, but minimally a fraction of a second to be effective ramping gain changes. It takes time to be slow.

JR

 

[Eric Cagle]

Re: Compression

I believe the DBX supergate employed the same technology as far as a analog processer using a predelay on the processed signal with the key tapped from the origonal signal . My understanding of some of the digital compressors is that there is some time (a few samples worth) alotted to define and reshape the waveform much more accurately than in the analog compressor processing world. More accurately may not be the correct word but in the case of a gate it doesn't clip the leading edge off as John mentioned. Also the artifacts that are added by the analog components themselves such as a VCA, are eliminated with digital.

 

[John Roberts]

Re: Compression

I believe the DBX supergate employed the same technology as far as a analog processer using a predelay on the processed signal with the key tapped from the origonal signal . My understanding of some of the digital compressors is that there is some time (a few samples worth) alotted to define and reshape the waveform much more accurately than in the analog compressor processing world. More accurately may not be the correct word but in the case of a gate it doesn't clip the leading edge off as John mentioned. Also the artifacts that are added by the analog components themselves such as a VCA, are eliminated with digital.

Digital signal processing can use alternate techniques to analog to reduce artifacts of gain changes. Using simple continuous attack and release side chain algorithms creates waveform distortion (especially fast release at lower frequencies). Digital could co-ordinate step gain changes to occur at zero crossings only, so the gain step multiplied times 0v generates no artifact.

Not to repeat myself too much, to synchronize with the zero crossing at the start of a signal level change, you need to read enough about that level change that hasn't happened yet to be useful. So again that takes time (a fraction of a period at lowest dominant frequency of interest).

I am a huge fan of digital but it takes time to use time.

JR

 

[John Chiara]

Re: Compression

The DBX Supergate did the same thing and I believe they listed the Look ahead delay as 20microseconds. That's where I got that idea from.

 

[John Roberts]

Re: Compression

The DBX Supergate did the same thing and I believe they listed the Look ahead delay as 20microseconds. That's where I got that idea from.

For a gate, it could be relatively short. You don't need to see how loud it is going to get, only know that it is above threshold, and open up the gate soon enough to not add extra clicks.


JR

 

[Greg Cameron]

Re: Compression

XTA's "D-Max" peak limiters use a look-ahead scheme which they claim use "10 of microseconds" of delay to apply. I have to say, they work rather well with minimal audible distortion.

 

[John Chiara]

Re: Compression

For a gate, it could be relatively short. You don't need to see how loud it is going to get, only know that it is above threshold, and open up the gate soon enough to not add extra clicks.


JR

Trying to visualize. As long as all of the signal is running through the delay line, it should be able to adjust for whatever comes along, although I guess attack time might have an impact.

 

[Misja Hoen]

Re: Compression

The biggest impact to be real. Forget all this above. Athough fair points none the less. I think for dynamic FX the least understood but dramatic feature is the attack and release times.

 

[kristianjohnsen]

Re: Compression

Cough... microseconds?

Just this saturday I had a gig with some room to experiment. I was working with a lead vocalist with a "tricky voice". Rummaging around the various knobs on the multiband compressor in the Yamaha desk I was using, I discovered it had a variable "look ahead" time. I strapped the compressor across the vocal bus, and set the look-ahead to 3ms. I also delayed every other bus "running parallell" by 3ms. Could not perceive any other changes than the vocals sounding better In the grand scheme of things, "moving" the PA by 1 meter is nothing, anyway!

 

[Eric Cagle]

Re: Compression

The biggest impact to be real. Forget all this above. Athough fair points none the less. I think for dynamic FX the least understood but dramatic feature is the attack and release times.

Yes. That is almost the difference between a cardboard box and an actual instrument. Also all the cool effects are done playing with those, especially the release time.

An easy one to learn is using a compressor at a 3:1 or 4:1 ratio to tighten up a bass guitar (which will make it dull) and then backing off the attack time until it livens back up.

Another good example is John Chiara's snare drum setting. Hopefully some more of you will share some favorite settings for people to experiment with.

 

[Per Søvik]

Re: Compression

Just this saturday I had a gig with some room to experiment. I was working with a lead vocalist with a "tricky voice". Rummaging around the various knobs on the multiband compressor in the Yamaha desk I was using, I discovered it had a variable "look ahead" time. I strapped the compressor across the vocal bus, and set the look-ahead to 3ms. I also delayed every other bus "running parallell" by 3ms. Could not perceive any other changes than the vocals sounding better In the grand scheme of things, "moving" the PA by 1 meter is nothing, anyway!

Since it is good practice anyway to delay the signals in the pa to align with sound coming off the stage, the distance from your source to the pa kind of defines the available processing time. This is particularly true for smaller venues where the ratio between stage sound and pa sound is small enough to make a difference to the quality of what you hear out in the audience. With floor stacks in front of the stage, the available processing time could easily be 10-15 mS at front of stage and as much as 50 mS for the backline and drums. With some smaller mobile stages and small format arrays hanging more or less in line with the stage lip, the scope can be far less.

 

[Per Søvik]

Re: Compression

I also think that if you use compression and aren't good at it, yoSu do much more harm than good

Damn, I hear this so often. All you need to do is change the threshold control!!
This goes back to my gripe, that if sound mixers don't have a way to practice, how in the hell are you going to get good at something? Get some free DAW software, grab some prerecorded tracks from any number of Internet sites and learn how to use the damn tool. Really... Anything else in 2013 is just being lazy. If you are passionate about mixing, learn the craft. A magical performance still needs technical know how.

If I have got very little time for sound check or no time at all, which is often the case, I find that things go much smoother if I take the lazy approach and start out with guesstimated compressor/limiter settings for everything. That way I can start out with a resemblance of balance in the mix even when the source loudness/gains are all over the place. Yes, the opening number might be a bit dull, but it means one can quickly start focusing on the money-channel(s) and open up a bit as each channel gets under control.

 

[John Roberts]

Re: Compression

Trying to visualize. As long as all of the signal is running through the delay line, it should be able to adjust for whatever comes along, although I guess attack time might have an impact.

The concept of look ahead involves using prior knowledge of the future so you can ramp gain changes slowly, and still get there in time to prevent overload or some other situation. The amount of look-ahead time matters. You need to see enough of the signal to impute useful gain change information.

In recording this look ahead is often performed in non real-time, so you could look ahead the whole performance, say for mastering. Non real-time processing involves tricks like half speed mastering, to double the apparent HF bandwidth of a cutting lathe, or even running recordings backwards to trade sudden attacks, for gentle decay envelopes.

Back to live processing, with stand-alone boxes we do not have the flexibility to perform delay alignment, so this is, in my experience, limited to noise gates that get a lot of benefit from a short delay. For general dynamics processing, more is better and several mSec can buy enough time to smooth gain moves so this is great feature for use inside a digital console. Besides the look-ahead, digital dynamics side-chains can use digital decision making and processing power to reduce audible artifacts.

Are we far enough off the reservation yet with this veer?

JR

 

[kristianjohnsen]

Re: Compression

Since it is good practice anyway to delay the signals in the pa to align with sound coming off the stage, the distance from your source to the pa kind of defines the available processing time. This is particularly true for smaller venues where the ratio between stage sound and pa sound is small enough to make a difference to the quality of what you hear out in the audience. With floor stacks in front of the stage, the available processing time could easily be 10-15 mS at front of stage and as much as 50 mS for the backline and drums. With some smaller mobile stages and small format arrays hanging more or less in line with the stage lip, the scope can be far less.

It's important to remember that the sound coming off the stage doesn't travel in one narrow, straight line from the snare drum, past one speaker, and toward one listening position.

While delaying the sound system to match the backline does make sense at first glance (and often does, I agree), you can get weird effects. Examples:

In a shallow, wide room with PA either side of the stage, a lot of people may be further away from the speakers than from the backline. Delaying the mains will just make the offset worse.

Another one would be where people in the audience are up close to a stage and hearing monitors and PA at the same time. Since you can't delay the monitors anyway, you might as well leave the mains alone.

Yet another one is where there are center clustered subs. People in the middle would benefit from delayed subs, while people one the sides would benefit from delayed mains...all the while the people on stage probably would not like delayed subs much at all.

 

[Max Warasila]

Re: Compression

Ok I will weigh in on this one. Let me first say that I came from the recording studio before live sound and I had many many hours behind the monitors learning what my various tools did. I have mixed many shows with NO compression whatsoever and sometimes still do if the musicians and vocalists are good enough and I have racks of DBX 160 and 1066 always available and all of the various digital offerings as well. Here are my thoughts for what it is worth:

The only "have to have" compressor settings I use are:

1)Bass guitar if it is run direct. This instrument can be all over the map with dynamics, even more than a really inexperienced vocalist. A 3:1 ratio (up to a 4:1 max) is a good place to start with the threshold set around 0db if you are gained in correctly. You are trying to tighten up that flappy sound and get a more even output (and save massive headroom) without being dull and muddy. The magic happens when you back off the attack a little to allow the slaps and picking dynamics to come through while still having the tightness of the compressor grabbing the peaks. Use your ears and you will get a feel for what your particular compressor will do.

2)A Vocalist who doesn't know how to work a mic (which unfortunately is most of them). Again I like to start at about the same settings as for the bass except leave the attack at max or just back off a touch if you are wanting to let a little personality through but be careful because you need the compressor to grab the peaks pretty quickly. The trick is to smooth out the peaks and leave the rest of it alone. I usually see no compression happening until the singer starts belting it out and then a 3:1 ratio really makes things smooth without sounding unnatural.

That is it for the "have to have" things. There are hundreds of other things a compressor can be useful for but you start getting into "effects", personal taste, and compensation for poor muscianship. The only other thing I occasionally like to have is a compressor on accoustic guitar but you can go to "lifeless and boxy" really quick if you use too much ratio and you don't back off the attack and release a bit. 2:1 or 3:1 MAX ratio on this one and just tame it a little tiny bit or there will be no "feel" at all.

If you only have one compressor available you can use it on the main output at the 3:1 ratio with it just coming in on the peaks to smooth everything out and get more of a "radio" sound and also catch those vocal moments when you really need a limiter. The DBX 1066 has a contour button which filters the bass frequencys and will let more of the kik drum through without triggering the comp. It also has adjustable attack and release times which can be backed off a little to make everything sound a little more natural.

Just my thoughts. Good luck with it!
-Eric

Can not spam the like button enough. :thumbup:

My two cents: Compressors are not the devil. They are like bleach- incredibly useful but if you add in too much color you get bad results.

 

[John Chiara]

Re: Compression

Yeah, sorry for the tangent. I just get frustrated. I did an opening gig for a National this week and the house guy didn't patch in any comps because he doesn't use them. I am assuming ignorance but Man I am baffled by so many seemingly uninterested mix engineers. Again, trying to imagine another occupation where just getting by is acceptable. I think this prevalence makes it hard to convince owners, promoters or even bands that there is a difference out there.