cocamycola89
New member
Knowledge is power my man... besides isn't this a recording, mixing and mastering forum?
Understanding Dynamic Processors (Compression)
- Thread starter WeissSound
- Start date
MoonDragon
MoonDragon
THis post is very good.... and still being read
WeissSound
Engineer
Dynamic processors are not necessarily digital. The majority of them still exist as analog hardware. Additionally, compressors should not change the pitch (which is the perception of fundamental tone), nor will it make significant alterations to any wavelength or composite wavelength of the sound. I would even argue that "intensity" is not equivalent to "amplitude", and while the intensity of the sound will change shape through a dynamic processor, it's inherent intensity is [subjectively] printed.
Sorry for being confrontational, but your phrasing/information is misleading.
Sorry for being confrontational, but your phrasing/information is misleading.
MoonDragon
MoonDragon
I also have been a sound engineer for 20 years and totally concur with WeissSound's comment
HighGradeMuzik
New member
WeissSound
Thanks 4 this post man, it helps a lot!
Just started producing my own tracks and mastering is a big issue LOL
Big UP! 1 BASS
Thanks 4 this post man, it helps a lot!
Just started producing my own tracks and mastering is a big issue LOL
Big UP! 1 BASS
Last edited:
dumkidz(sam)
New member
very great post made me realize alot of beat makers dont even use compression but it made it alot easier 2 understand
dynamic_producer
New member
That was a very informative post...
WeissSound
Engineer
this is not the right thread. start a new one on live sound and one on tuning (and what you are trying to tune for that matter).
---------- Post added at 10:07 AM ---------- Previous post was at 10:05 AM ----------
One of the things that makes compression so "mysterious" or "difficult" is that all the settings influence each other. It's not like an EQ where the you turn a frequency band up or down. It's more like if turning up one band on an eq makes another band's Q get wider - that would make EQing a little more complicated.
In addition, analog circuitry and emulation will give you different tones based on the input, output, and "work" that the compressor is doing. In other words, if the signal comes in 5db hotter, turning the threshold up 5db will not necessarily produce the same results.
However, taking compression one step at a time is a really good way to approach things.
For example, you can start with your tone. This to me is 50% of compression. Tone tone tone tone tone. Run your signal into the front end of the compressor and circle it back in to your DAW/Tape Machine. Put the ratio at 1:1, set the attack to super long (or bypass if available), set the release medium, set the threshold all the way up to 0db. Basically, get the compressor doing the least amount of work possible, so you're just getting a return sound from the actual gain staging of the compressor. Set the level of the signal coming in, and the level of the signal coming out in a way where the TONE sounds best to your ear. This will help you learn your compressor, and will allow you to select different compressors more effectively.
Once you hear the tone you want, then you can select the compression and timing parameters that fit the bill.
For the purpose of this thread, I'm not going to explain what attack, release, threshold, etc. all mean. Plenty of resources that explain these just fine. What I'd like to do is explain the relationship between them.
Threshold is the ultimate control tool. It's going to effect the magnitude of the ratio, the speed of the attack, and the speed of the release. Compression is all relative to the signal that is coming in, and threshold defines that relativity.
The compressor reacts as soon as the voltage surpasses threshold. If the attack is constant, lowering the threshold will effectively make the compressor react sooner, because the threshold is breached quicker. Similarly, the compressor lets go once the signal is ducking the threshold, so with a constant release, the compressor will hold on longer with a lower threshold. Lastly, the ratio will have an overall greater amount of reduction BUT since more of the signal is being effected it actually makes the tonal change more homogenous. You end up hearing less "compression", though the tonal changes coming from the reduction actual touch more of the signal. Kind of confusing, I recommend a little experimentation on that one.
Essentially, threshold is your number one control as it determines just what you are effecting - the attack and transients of your signal? The upper part of the sustain? The deeper part of the sustain? The release? In a sense it's KIND OF like an EQ, you're exaggerating whatever's lurking in certain parts of the signal, except instead of parts of the frequency spectrum you are targeting parts of the temporal band.
The ratio is pretty straight forward. It's basically how hard the compressor is going to work. Really pushing the ratio tends to drive the gain reduction circuit which can have other artifacts outside of the compression. Particularly when the attack and release are very fast. This is why fast, high ratio compression is generally dedicated to a device that is designed specifically for that purpose: A Limiter. But sometimes you want the distortion or pump or breathing that can come from high ratio settings - sometimes it can sound really cool (aka Fairchild). Usually, the way I approach ratio is "as little as necessary to accomplish the task." Most people start at a 4:1 ratio for vocal work, I usually start at 2:1 and move to higher settings if I feel I'm not tucking things in enough - or if I'm getting too much pumping with a lower threshold, I might raise the threshold a bit and raise the ratio as well.
Attack is one of the first places people start getting a little confused, but it's actually pretty simple. Gain reduction has to occur over time. It's unrealistic to have an instantaneous change in voltage - although some compressors can get pretty damn fast. There will be a speed to the gain reduction, and the attack controls how fast or slow the reduction is accomplished. What makes attack confusing is how that idea translates to an actual signal. Basically, the total gain reduction is an idea, -Xdb, but that amount of gain reduction will probably not occur. The sound is probably moving too quickly for total gain reduction to every be met. The attack almost functions like a ratio of the ratio - how much of the gain reduction can be met relative to the shape of the signal that surpasses the threshold. Yikes! A simple affect of attack is how much the initial transient goes untouched. Longer attack settings will ease the amount of gain reduction on the transient of the signal. Part of the transient will always poke through, so you if you are trying to eliminate or soften a transient you really have to set the threshold low and the attack carefully - otherwise you get a "spike" at the attack of the signal.
Knee is the curvature of the attack. All compressors have a knee shape. Some have a "hard" knee where the gain reduction is applied in a linear fashion over time. Others have a "soft" knee, where the gain reduction is milder at first, and stronger as it kicks in a bit more (optical compressors generally have "soft" knees). Knee is just a finer control over the attack of the compressor, one more way to more precisely target the effect you want. Some compressors have a variable knee.
Release is the one that leaves most people in the dark. It's really not much different from the attack. It's how long the compressor takes to ease off of the signal once it's back below the threshold. The effect of the release can be most easily heard on low thresholds - so it's not a bad idea to exaggerate your threshold setting while determining the release time. Release is usually a play between "thickness" and "naturalness." The faster you release, the more of the quiet parts of the signal will be preserved, but too fast of a jump causes notable compression artifacts (many actually distort in a less than pleasing way). Or, you end up getting too much of the lower parts of the signal - so you might get something thick, but perhaps too thick or just too much of something you wouldn't normally hear from the sound. You may lose cut or impact. Too long however, and you end up exaggerating the front end of the signal - the quiet parts are being reduced as well. On low thresholds you can actually hear the compressor letting go and the signal ramps up. This is called "pumping". Then again, this is sometimes desirable. Sounds annoying on things like vocals or a 2-track though.
Here's an example of these ideas working in tandem. Let's say you have an upright bass. It sounds a little dull, and a little soft. You want to bring out the snap of the string, and you want to "thicken" the body of the tone. The string snap is living in the attack of the signal. The tone is the resonance of the basses body which is living at the quieter part of the sustain. You may have bridge sound and fundamental tone living in the decay and upper part of the sustain. You set your threshold pretty low, just above where the release of the bass starts. You set your attack slow, to allow the attack and decay to poke through. You set your release fast, so that the compressor isn't clamping down the resonance of the body. Effectively, what you are doing is reducing the sustain part of the signal. You use the make up gain to return the sustain up to it's original volume, so now you have more attack and decay, and more release - OR more string snap, and more body resonance. You choose a compressor that imparts a nice tone to accent the overtones in the bass signal which also helps give it more presence. You also might choose a compressor that has a rounder knee, to really target just the sustain - perhaps an optical compressor.
That should give some foundational ideas for compression, aside from just basic definitions. Ultimately, you have to hear everything working together to get a feel for compression and how it works. That was really long, so I'll talk about some parallel compression stuff in a later post. I think that was enough for now
---------- Post added 05-02-2011 at 12:14 PM ---------- Previous post was 04-29-2011 at 10:07 AM ----------
EDIT: It was brought to my attention that a VCA is not in fact a catch-all phrase for dynamic processors. This was incorrect and I apologize, so please take note. A VCA refers to one specific type of compression. A VGA - variable gain amplifier - is used in several different types of compressor designs, making it a bit more universal, but still not inherent to dynamic processing.
Different types of compression circuits include VCA, Optical, Diode, FET, Variable-Mu, and DSP.
---------- Post added at 10:07 AM ---------- Previous post was at 10:05 AM ----------
One of the things that makes compression so "mysterious" or "difficult" is that all the settings influence each other. It's not like an EQ where the you turn a frequency band up or down. It's more like if turning up one band on an eq makes another band's Q get wider - that would make EQing a little more complicated.
In addition, analog circuitry and emulation will give you different tones based on the input, output, and "work" that the compressor is doing. In other words, if the signal comes in 5db hotter, turning the threshold up 5db will not necessarily produce the same results.
However, taking compression one step at a time is a really good way to approach things.
For example, you can start with your tone. This to me is 50% of compression. Tone tone tone tone tone. Run your signal into the front end of the compressor and circle it back in to your DAW/Tape Machine. Put the ratio at 1:1, set the attack to super long (or bypass if available), set the release medium, set the threshold all the way up to 0db. Basically, get the compressor doing the least amount of work possible, so you're just getting a return sound from the actual gain staging of the compressor. Set the level of the signal coming in, and the level of the signal coming out in a way where the TONE sounds best to your ear. This will help you learn your compressor, and will allow you to select different compressors more effectively.
Once you hear the tone you want, then you can select the compression and timing parameters that fit the bill.
For the purpose of this thread, I'm not going to explain what attack, release, threshold, etc. all mean. Plenty of resources that explain these just fine. What I'd like to do is explain the relationship between them.
Threshold is the ultimate control tool. It's going to effect the magnitude of the ratio, the speed of the attack, and the speed of the release. Compression is all relative to the signal that is coming in, and threshold defines that relativity.
The compressor reacts as soon as the voltage surpasses threshold. If the attack is constant, lowering the threshold will effectively make the compressor react sooner, because the threshold is breached quicker. Similarly, the compressor lets go once the signal is ducking the threshold, so with a constant release, the compressor will hold on longer with a lower threshold. Lastly, the ratio will have an overall greater amount of reduction BUT since more of the signal is being effected it actually makes the tonal change more homogenous. You end up hearing less "compression", though the tonal changes coming from the reduction actual touch more of the signal. Kind of confusing, I recommend a little experimentation on that one.
Essentially, threshold is your number one control as it determines just what you are effecting - the attack and transients of your signal? The upper part of the sustain? The deeper part of the sustain? The release? In a sense it's KIND OF like an EQ, you're exaggerating whatever's lurking in certain parts of the signal, except instead of parts of the frequency spectrum you are targeting parts of the temporal band.
The ratio is pretty straight forward. It's basically how hard the compressor is going to work. Really pushing the ratio tends to drive the gain reduction circuit which can have other artifacts outside of the compression. Particularly when the attack and release are very fast. This is why fast, high ratio compression is generally dedicated to a device that is designed specifically for that purpose: A Limiter. But sometimes you want the distortion or pump or breathing that can come from high ratio settings - sometimes it can sound really cool (aka Fairchild). Usually, the way I approach ratio is "as little as necessary to accomplish the task." Most people start at a 4:1 ratio for vocal work, I usually start at 2:1 and move to higher settings if I feel I'm not tucking things in enough - or if I'm getting too much pumping with a lower threshold, I might raise the threshold a bit and raise the ratio as well.
Attack is one of the first places people start getting a little confused, but it's actually pretty simple. Gain reduction has to occur over time. It's unrealistic to have an instantaneous change in voltage - although some compressors can get pretty damn fast. There will be a speed to the gain reduction, and the attack controls how fast or slow the reduction is accomplished. What makes attack confusing is how that idea translates to an actual signal. Basically, the total gain reduction is an idea, -Xdb, but that amount of gain reduction will probably not occur. The sound is probably moving too quickly for total gain reduction to every be met. The attack almost functions like a ratio of the ratio - how much of the gain reduction can be met relative to the shape of the signal that surpasses the threshold. Yikes! A simple affect of attack is how much the initial transient goes untouched. Longer attack settings will ease the amount of gain reduction on the transient of the signal. Part of the transient will always poke through, so you if you are trying to eliminate or soften a transient you really have to set the threshold low and the attack carefully - otherwise you get a "spike" at the attack of the signal.
Knee is the curvature of the attack. All compressors have a knee shape. Some have a "hard" knee where the gain reduction is applied in a linear fashion over time. Others have a "soft" knee, where the gain reduction is milder at first, and stronger as it kicks in a bit more (optical compressors generally have "soft" knees). Knee is just a finer control over the attack of the compressor, one more way to more precisely target the effect you want. Some compressors have a variable knee.
Release is the one that leaves most people in the dark. It's really not much different from the attack. It's how long the compressor takes to ease off of the signal once it's back below the threshold. The effect of the release can be most easily heard on low thresholds - so it's not a bad idea to exaggerate your threshold setting while determining the release time. Release is usually a play between "thickness" and "naturalness." The faster you release, the more of the quiet parts of the signal will be preserved, but too fast of a jump causes notable compression artifacts (many actually distort in a less than pleasing way). Or, you end up getting too much of the lower parts of the signal - so you might get something thick, but perhaps too thick or just too much of something you wouldn't normally hear from the sound. You may lose cut or impact. Too long however, and you end up exaggerating the front end of the signal - the quiet parts are being reduced as well. On low thresholds you can actually hear the compressor letting go and the signal ramps up. This is called "pumping". Then again, this is sometimes desirable. Sounds annoying on things like vocals or a 2-track though.
Here's an example of these ideas working in tandem. Let's say you have an upright bass. It sounds a little dull, and a little soft. You want to bring out the snap of the string, and you want to "thicken" the body of the tone. The string snap is living in the attack of the signal. The tone is the resonance of the basses body which is living at the quieter part of the sustain. You may have bridge sound and fundamental tone living in the decay and upper part of the sustain. You set your threshold pretty low, just above where the release of the bass starts. You set your attack slow, to allow the attack and decay to poke through. You set your release fast, so that the compressor isn't clamping down the resonance of the body. Effectively, what you are doing is reducing the sustain part of the signal. You use the make up gain to return the sustain up to it's original volume, so now you have more attack and decay, and more release - OR more string snap, and more body resonance. You choose a compressor that imparts a nice tone to accent the overtones in the bass signal which also helps give it more presence. You also might choose a compressor that has a rounder knee, to really target just the sustain - perhaps an optical compressor.
That should give some foundational ideas for compression, aside from just basic definitions. Ultimately, you have to hear everything working together to get a feel for compression and how it works. That was really long, so I'll talk about some parallel compression stuff in a later post. I think that was enough for now
---------- Post added 05-02-2011 at 12:14 PM ---------- Previous post was 04-29-2011 at 10:07 AM ----------
EDIT: It was brought to my attention that a VCA is not in fact a catch-all phrase for dynamic processors. This was incorrect and I apologize, so please take note. A VCA refers to one specific type of compression. A VGA - variable gain amplifier - is used in several different types of compressor designs, making it a bit more universal, but still not inherent to dynamic processing.
Different types of compression circuits include VCA, Optical, Diode, FET, Variable-Mu, and DSP.
NobleSounds
Noble
Very useful information. Thank you for sharing so much knowledge and experience.
expljoesion
Reason 5 / Sonar 8 PE
This is incredible. Thank you for sharing this information!
WeissSound
Engineer
It depends on the task at hand. There's really two determining factors I think about when selecting a compressor - it's inherent tone, and it's action. If I want to emphasize the low mid harmonics, get some gruffness in there, an 1176 is a good choice - whereas if I want an upper midrange glow, I like the Alan Smart C2. Rvox has a really nice way of rounding out the low range and low mids in a smooth way. So tone has a lot to do with it. If I need something to effect fast transients, I'm looking at RComp, or an 1176. If I want something with a rounder approach on the attack and release, I might go for an LA2A or CL1B. There's a lot of choices out there both in the hardware and software world. If you shoot me a specific task I can make recommendations - otherwise - well they're all good.
I like the distortion I get from the Art Tubepac. That thing is a cheap piece o'crap - I use it all the time!
slanderson89
New member
This is really great.
TrackPizza
New member
thanks for taking the time to post that - lots of info to digest
WeissSound
Engineer
You're welcome guys/gals. It's good to feel appreciated.
Hey Weiss,
many thanks for this very inspiring post. You really made my day! I really appreciate it.
The reason why I post in this thread today is fairly simple. I´m one of those guys who´s only working inside Reason Record. No rewiring. I don´t own pro equipment and stuff like that so I have to work with what I´ve got and do the best out of it. But at least it´s licensed and it´s the only thing I could afford at the moment (don´t like using cracked stuff).
Anyway, I´m working with this setup for more than 2 years now in my spare time and I´ll try to understand each tool only by listening carefully to it. I don´t have those fancy VSTs GUIs where I could see whats going on ( and believe me - that was one thing I hated so much in the beginning but I don´t miss it anymore). Most of the time my only chance to measure something out was (and already is) to take a look at my meters and trusting my ears. But I have to admit that I had some serious problems reading my meters correctly most of the time until today.
I asked goolge for more precise information about headroom, level, VU, PPM, peak, analog and digital, then and now and all that stuff and after understanding most the basic things I decided to take a closer look at the manual (which is really great btw) and stumbled across something called PPM mode: "...In PPM (Peak Program Meter)Mode, the meter response is 0 ms rise and 2.8s/24 db fall. The PPM mode is perfect for detecting transients in the sound. There is no peak segment in the meter. ...". I remembered your post about compression and transients and thought a little bit about it before I started testing.
I was interested in doing some tests compressing a selfbuild analog modelled KONG kickdrum (very pure sound - 808 stlye with lots of dynamics).
I built up my chain with the kick sound in the beginning and a compressor in the end (no inserts and/or sends). My goal was to have some kind of A/B comparison with a ppm meter. The kick had a peak somewhere between -12db and -10db so I adjusted the output gain of the comp everytime I touched the attack, release, threshold, ratio etc for a better comparison. What a blasting experience, really. The more I play with it the more I hear the differences. While having some kind of extreme settings in the beginning to get a feeling for it I´m now hearing the impact of it with only minor corrections. I´m playing around with the punch of it and it´s a blast. And watching the PPM is so much fun. I´m not that kind of guy who is using analyzers and stuff like that (I trust my ears) but the PPM gave me the hint in the right direction.
I´m looking forward to work with the compressor while hearing the entire mix (either on one bus or the master, I don´t care at the moment. I just wanna hear sound. )
The last time I had so much fun (and made such a progression in sound) was discovering M/S processing (building up my own chain). I think I´ve learned so many important things about audio only using Reason Record.
So today I discovered my compressors on a completely new level and also learned a lot about metering in the analog/digital domain. Thanks again for this very inspiring post.
Sebastian
many thanks for this very inspiring post. You really made my day! I really appreciate it.
The reason why I post in this thread today is fairly simple. I´m one of those guys who´s only working inside Reason Record. No rewiring. I don´t own pro equipment and stuff like that so I have to work with what I´ve got and do the best out of it. But at least it´s licensed and it´s the only thing I could afford at the moment (don´t like using cracked stuff).
Anyway, I´m working with this setup for more than 2 years now in my spare time and I´ll try to understand each tool only by listening carefully to it. I don´t have those fancy VSTs GUIs where I could see whats going on ( and believe me - that was one thing I hated so much in the beginning but I don´t miss it anymore). Most of the time my only chance to measure something out was (and already is) to take a look at my meters and trusting my ears. But I have to admit that I had some serious problems reading my meters correctly most of the time until today.
I asked goolge for more precise information about headroom, level, VU, PPM, peak, analog and digital, then and now and all that stuff and after understanding most the basic things I decided to take a closer look at the manual (which is really great btw) and stumbled across something called PPM mode: "...In PPM (Peak Program Meter)Mode, the meter response is 0 ms rise and 2.8s/24 db fall. The PPM mode is perfect for detecting transients in the sound. There is no peak segment in the meter. ...". I remembered your post about compression and transients and thought a little bit about it before I started testing.
I was interested in doing some tests compressing a selfbuild analog modelled KONG kickdrum (very pure sound - 808 stlye with lots of dynamics).
I built up my chain with the kick sound in the beginning and a compressor in the end (no inserts and/or sends). My goal was to have some kind of A/B comparison with a ppm meter. The kick had a peak somewhere between -12db and -10db so I adjusted the output gain of the comp everytime I touched the attack, release, threshold, ratio etc for a better comparison. What a blasting experience, really. The more I play with it the more I hear the differences. While having some kind of extreme settings in the beginning to get a feeling for it I´m now hearing the impact of it with only minor corrections. I´m playing around with the punch of it and it´s a blast. And watching the PPM is so much fun. I´m not that kind of guy who is using analyzers and stuff like that (I trust my ears) but the PPM gave me the hint in the right direction.
I´m looking forward to work with the compressor while hearing the entire mix (either on one bus or the master, I don´t care at the moment. I just wanna hear sound.
)The last time I had so much fun (and made such a progression in sound) was discovering M/S processing (building up my own chain). I think I´ve learned so many important things about audio only using Reason Record.
So today I discovered my compressors on a completely new level and also learned a lot about metering in the analog/digital domain. Thanks again for this very inspiring post.
Sebastian
Last edited:
WeissSound
Engineer
Thank you for an inspiring post! I hope people who are learning about mixing read and re-read your post because there are some real gems in there.
First: cracked software. Good for you not doing that. Aside from the moral reasons, there is an important psychological reason not to steal software. If you BUY it, you will value it. By taking the time to learn Reason Record you will find all sorts of tricks and capabilities that someone who just steals software would never bother to learn. Software are tools just as much as hardware.
Second: When learning, there is a great temptation to rely on meters and analyzers. You don't, and while that can be difficult at first, you notice it's getting easier to hear everything you need to hear. Eventually you get in touch with your ears which can tell you far more than an analyzer can. Just using the meters as a reference (or maybe not at all) is really all you need them for.
Third: You're experimenting. Take the time to learn everything and try things. Especially now - once it becomes a full time gig the time for experimentation dwindles. So keep keep keep experimenting and finding what you like. Using a PPM detector to keep the peak height equal while changing the shape of the transient is an EXCELLENT experiment and really reveals pretty much the whole point of this long winded post.
Other things to check out - using different compressors to achieve the same results. You will notice tonal variations and practical variations which will help you choose which compressor for which task. Also, compression and M/S processing can be fun to check out. After you play with the master buss, split the mix M/S and compress just the sides, or just the mids and listen to how it effects the dimensions of the music. It can be cool (usually messes up the mix though).
First: cracked software. Good for you not doing that. Aside from the moral reasons, there is an important psychological reason not to steal software. If you BUY it, you will value it. By taking the time to learn Reason Record you will find all sorts of tricks and capabilities that someone who just steals software would never bother to learn. Software are tools just as much as hardware.
Second: When learning, there is a great temptation to rely on meters and analyzers. You don't, and while that can be difficult at first, you notice it's getting easier to hear everything you need to hear. Eventually you get in touch with your ears which can tell you far more than an analyzer can. Just using the meters as a reference (or maybe not at all) is really all you need them for.
Third: You're experimenting. Take the time to learn everything and try things. Especially now - once it becomes a full time gig the time for experimentation dwindles. So keep keep keep experimenting and finding what you like. Using a PPM detector to keep the peak height equal while changing the shape of the transient is an EXCELLENT experiment and really reveals pretty much the whole point of this long winded post.
Other things to check out - using different compressors to achieve the same results. You will notice tonal variations and practical variations which will help you choose which compressor for which task. Also, compression and M/S processing can be fun to check out. After you play with the master buss, split the mix M/S and compress just the sides, or just the mids and listen to how it effects the dimensions of the music. It can be cool (usually messes up the mix though).