Above is the uncompressed sample. I notice that on the compressed sample, at the start of the waveform, there is a large peak at the start, which is ruining it. Both are normalised to peak at 0 dB, but of course the compressed sample does not look as I intended it to. The compressor seems to spend some time taking effect.
The compressor is set as follows:
Is it the attack value which is making it not have an effect on the start of the sample? If this is the case, is it possible to compress everything? Why can I not set the attack to zero?
If the lower one is the resulting waveform (post-compression), then the compressor is doing exactly what you set it to do. It’s compressing the heck out of the initial transient. What you interpret as not compressing at all, is in fact a large amount of compression that has had makeup gain applied to it. The compressor is evening out the dynamics of the transient. The actual compression would show up better in the waveform if there was no makeup gain applied.
If you’re looking to “elongate” the kick, I’m not sure you’ll succeed with this sample. This particular kick seems to have very little reverberance to elongate. It seems to die out to almost zero very quickly, if I’m reading your screenshot correctly. And if it’s not there in the first place, you won’t be able to create it out of thin air. Try similar compressor settings on a sample that has a much longer tail. You’ll then hear that tail jump out after the initial transient. For a longer tail still, you could imprint it with some reverb as well (assuming it’s got some decent lower-mid frequencies in it to reverberate).
I believe the compressor is meant to reduce the volume of anything beyond a certain threshold. This, combined with makeup gain, creates a “squashed” (“compressed”) wave. This is reducing the dynamic range of the sound, right?
So I am expecting to see basically almost the same as the uncompressed waveform, except the low frequency after the initial transient persists and stays louder for longer. So it looks as though there is less “fade out”. Also, the initial transient would be slightly louder, as this is not the highest peak.
Is this not what the compressor does? When I looked up the function of a compressor, this is what I understood.
Your understanding is generally correct, but the thing is there’s hardly ANY tail on this sample. So there’s no time for the compressor to release (even at it’s shortest setting) before the tail has already finished playing.
The sample is very short, I agree. I will experiment with longer samples. But this doesn’t answer the question about why the initial transient peaks so extremely, then it goes so quiet.
You’ll notice at the very end of your waveform, the compressed signal IS louder than the original signal. That’s probably the point at which the compressor has finally started to release, but by that time, it’s too late… The signal is already almost to zero.
Try setting the compressor to peak rather than full RMS. It might engage in a slightly different way (reacting to the peaks of the signal rather than the average level of the signal). In either case, though, the attack can’t be set to true-zero because it would then need to “pre-compress” the signal before it could even see what the signal was. It needs time to see the signal and react.
I’ll shut up now, though. There are people here FAR more qualified than me on this subject…
Yes, thank you setting the analysis to peak does produce what I expect, more or less. I’ve experimented with a few different samples, and full RMS seems to make the compressor just reduce the volume of loud sections… I guess that is what RMS is.
I am disappointed with my own expectations, though, and how I thought it would be so easy to improve the power of kicks/snares. My ultimate aim is to be able to make an extremely punchy kick and snare. My original belief was that EQing peaks and squeezing as much volume out of a sample with compression would enable this, but it hasn’t.
I see what you mean, but the “live” setting says that it stops “look ahead”. It is disabled by default, and I had it disabled for my previous examples. The setting says that it does look ahead slightly by delaying the signal or something, so it can see into the future, as it were. So attack and “live” seem to have a connection.
Try researching a technique called “parallel compression.” Can be very effective.
The short story is you’re mixing an uncompressed version with an extremely compressed version, but at a lower level (to taste). The effect being the uncompressed version gives you the transient you want but when that decays the heavily compressed version “takes over” and provides body and sustain.
Well, to properly adjust a compressor is not the easiest thing in the world. Each instrument or sound requires a particular setting, which is also dependent on the particular effect you are looking for. Moreover, the matter varies from one compressor to another and finally you’ll have to check the final effect in the mix, because what we want is not that a particular good sound, but a sound that works well in the mix.
The solution: experience or to test, test and test.
For not all remain theoretical I dare to give you some compressor settings for kick drum, even at the risk of being strictly useless:
Threshold: to taste (depends on the sample level and the type of compressor)
Attack: rather slow
Release: rather fast (0.2 sec?)
Ratio: 6-8
You might want to listen how the BD sounds in my mixes clicking on my signature. May be it’s not what you are looking for.
If you’re using commercial samples, these are often already quite compressed. SLD Music mentioned using reverb, I too think that would give you closer results with a bit of additional EQing and comp.
Maybe try some subtle tube/tape saturation or something to beef up the samples without completely mangling them.
You’re definitely on the right track with compression and EQ-ing, and I think playing around with attack and release and listening very carefully to the result will pay off.
After that you’ve got to try psychoacoustic things to fool your ear into making the sounds seem ‘bigger’ (such as distortion/saturation).
Saying that, sometimes you’ve just got to bite the bullet and try different samples/sounds before you get the effect you’re looking for.
…which features a before/after comparison, although I’m sure there’s many detailed tutorials on ‘YouTube’ with a step by step approach which would be really useful to you so that you can gradually hear the difference between each stage etc.
I found the quick overview/advice above inspiring as its forced me to think more creatively about ways in which to use compression (i.e. such as using different types and settings of compression at different stages in the signal path) rather than just overall compression on drums period.
What you are trying to achieve is much easier with something like SSD as it has heaps of gorgeous room ambiance (if you want it) which makes the compression effect(s) much more obvious to pick out, but as others have mentioned it only takes a moment to process yous samples with a decent ‘room’ IR/verb to gain access to a similar sound.
I’m a noob when it comes to doing anything clever/sophisticated with compression, but having recently snagged a copy of SSD I’m downright intrigued now as there seems to be so much you can do if you have a pile of funky drums sounds to play with!
Reading some good quality theory is mega helpful too (if you’re a book worm like me) but considering the depth of information available online these days a good search through say the ‘Sound On Sound’ archives should give you lots of ideas (they have great articles/tutorials).
Speaking of SOS, Mike Seniors book Mixing Secrets For The Small Studio is packed full of useful information (including why/when/how to use compression etc) and as a literary fiend I couldn’t recommend that enough! Good luck with the compression experimentation anyway, I agree totally that exploring and listening is the way to go!
The way I think of compressors when used with drums is that the Attack value controls the peak of the start of the sample, thus the punch. A longer attack value lets through more of the original drum level before clamping down the volume of the rest of the sample. The release value is then used to allow the compressor to ‘reset’ itself ready for the next drum hit, rather than to make the tail become louder (though it will if set appropriately).
If you want to elongate the tail of a drum then you should try a limiter because it will clamp down all signal above a certain level. So, you clamp everything down until the tail arrives. An extreme type of limiter would be a guitar amp or distortion simulator!
But if you’re finely sculpting individual samples then you could use clip automation to get a more controlled shape.
And an envelope shaper works very well too if you have one available.
This is a great thread. Thanks everyone. There are resources ad nauseum about what a compressor does and what the controls do. It’s much harder to get information on exactly what settings to try on what sounds, and practical things you can try yourself to hear the differences. The advice here has been a great kick start for that kind of thing.
Regarding the original issue, it might be wise to really think whether it might be more profitable to look for or make a better sample. Ask yourself whether you’ may be trying to polish the proverbial turd.
The signal is already almost to zero.
setting the analysis to peak does produce what I expect, more or less. I’ve experimented with a few different samples, and full RMS seems to make the compressor just reduce the volume of loud sections… I guess that is what RMS is.