limit voice with compressor - how?
I'm starting to get the feeling that I don't exactly know how compressors work. My voice has a lot of ups and downs and I am already adjusting the audio by hand, cutting each part and adjusting the audio. Thats a lot of work and I am sure there is a better way.
Can't I say that the lowest point/audio peak of my voice track is -12 dB and the highest peak is -3? I don't exactly know how to do this in Premiere. I was assuming the dynamics filter would be my best choice and I get it to cut off if it goes over -3dB but I don't know how to tell it to not get any softer than -12 dB. Also how does it work? Does it really get CUT OFF? That way clipping would occur which isn't good either.
So what do I have to do if I want the voice to be just in a range of -12 dB - -3 dB?
Essentially you use the threshold to set the point at which compression will start occuring.
You then set the ratio, to determine how sharp the knee of the compression will occur. The higher the ratio, the sharper the knee (harsher) the lower the ratio the gentler the curve of the knee.
Say you noticed your audio was moving from a range of -20db to -3db on a VO (that'd be quite strange to start with as if you had your mic setup right, and the VO artist was giving a fairly consistent performance you wouldn't really see this much change in volume)
But for examples sake, lets say thats how much you had.
One of the uses of compression is to reduce the range of the signal so that you can increase the overall loudness.
So for our example you would wish to say set your threshold at say -10db and have your ratio set to say 3:1. Then at worst when your audio's original volume was at -3db, which is say 7db higher than the set threshold, it would compress it down by 3:1. So thats a 1/3rd of what it was orginally, so rather than 7db higher it would only be 2.3db higher, which would put it at -7.8db final output. SO its more of a ratio thingy rather than an absolute range. Absolute ranges can be achieved but you will get a much harsher knee, and basically you end up with a limiter, where if the signal goes to -10db, it aint going to go over, and the limiter harshly reduces the volume quickly.
Does that help?
Mac Pro 2.4Ghz 8 core, 24GB RAM, GTX 670
It's very difficult thing to explain compression in just a few words. There’s a lot going on and how you use a compressor depends on what kind of content you are trying to compress. There are many audio professionals that are still trying to get the hang of compression, so be prepared to put a bit of time into it…
First of all if you want to achieve quicker results without having to understand too much about compression, then I will post another message after this one with links to some third party compressors that are designed for simpler operation.
Keep in mind though, in keeping things simple, there is always trade offs. So ultimately, if you're serious enough, you will want to learn as much as you can about compressors.
Second of all, read what Craig has to say above, he has some good tips. if you find that not detailed enough then read on...
I’d like to apologise in advance for the length of this post, but its bit like trying to explain how a car engine works, so some level of detail is required.
I hope you find this more helpful than confusing!?!?
This screenshot is of the standard compressor that comes with Logic Pro X, not sure if Final Cut has this new version yet. Either way, its most likely you're running Premiere so none of that will matter. I have chosen this particular compressor for demonstration purposes, because its very visual and aids in explaining some of the key concepts...
In my attempt at trying to dispel a few mysteries about compression, I will focus on the most commonly found controls, using this image as a reference. The terminology or control names are generally the same across compressors, though some may vary. There are some compressors that offer a more basic set of controls, and there are others that include extra features, like the one from Logic. So don’t let that throw you.
Here are some important things that you should try to understand…
Otherwise known as the Power or On switch. This control does what it says, it turns the effect of the compressor off or on. This can be useful at listening to before and after versions of your audio so you can tell what type of impact it is having and whether or not you need more or less of it.
It’s also useful for balancing your input and output signals so it sound roughly the same level going out as it does coming in. This also makes it more easy to compare your before and after.
Sometimes its simply referred to as Input. Not every compressor has this control, but it can be useful. This sets how loud you want your volume (gain) to be before you start processing. Use this control to set your desired level, so that it’s peaking towards the top.
If you don’t have an input gain control and your volume isn’t in the desired region, then you may need to insert a Gain plugin to adjust the volume going into the compressor. If you do this and the audio sounds too noisy or distorted, chances are you may need to re-record your audio at a better level if possible.
Input Gain doesn't always have a meter to indicate how loud the signal is, but generally you will want it to peak anywhere between -1 and -8dB, ideally. That helps set a good general loudness for the rest of the compressor to do its part.
This acts kind of like a reference point for the audio you want to process. Getting this set right is quite important in how well overall the compressor does its job. Any audio that exceeds this reference point or threshold will get processed. Any audio that is lower then the threshold, will pass through unprocessed.
This is important because, generally speaking, you only want to tame your loudest levels so you can turn up your quietest levels. You must also be aware that too much processing on noisy or ambience sounds will only make that noise or ambience sound even louder which is usually undesirable. The only way to beat noisy signals is to either record it again, or use some noise reduction software, which might improve things but can’t work magic.
For the most part, you want to aim at setting the threshold somewhere below the level of your loudest parts. Above it and the compressor will have no effect. The lower the threshold is the more of the signal that gets compressed. Too much compression can lead to audible side effects, like breathing, pumping, elevated noise levels, loss of high-end and intelligibility, stuff like that. So it’s important to set this right.
For voice aim for a gain reduction of somewhere between -2dB to -6dB. I’ll explain how to do this shortly.
Notice that in the middle of image is the Graph. On the left side pane there is a bend in the diagonal line, that is the point at which the threshold is set. In this example it's just under -20dB, this is also indicated by the threshold control beneath. Everything above this point gets processed. The little white marker indicates how loud the signal is at this current moment. Since it’s above the bend, compression is currently being applied.
The threshold states where to start compressing the audio, but the amount of gain reduction to be applied is dictated by the Ratio control. The higher the ratio, the more the loudest parts of the signal gets squashed, or reduced in volume. The ratio is always measured on a n:1dB scale. Meaning that for every number of decibels that exceed the threshold (n) the audio is compressed to within a 1dB range.
For example, a ratio of 2:1 means that for every 2dB of audio increase it will be reduced to a maximum of 1dB above the threshold. A setting of 5:1 means that every 5dB that exceeds the threshold will be compressed to the space of 1dB.
In looking at the left side of the Graph again, you can see that above that point in the line that indicates the threshold, the line begins to slope at another angle. That is because the ratio is indicating a change of about 2.5:1, in this case. A ratio of 1:1 would indicate no change, therefor no compression will be applied and it would look like straight line on the graph.
In general, the lower the ratio the lower you can get away with setting your threshold. The lower your threshold, the more processing is applied to your audio, so you would want a lighter setting usually. This technique is commonly done if you just want to subtly lift the overall energy of the audio.
On the right side of the Graph pane shows the history of the processed audio. The area below represents the actual audio signal, while the white line above it indicates how much gain reduction is being applied to that part of the audio. Noticed the louder the signal, the more reduction was applied.
To the right of that is a vertical white meter, this is the gain reduction meter, most compressors have these. Its probably your most important meter. Many compressors have just one meter with a selector switch that will show the Input, the Output and also the Gain Reduction levels. Make sure you select Gain Reduction (GR) whenever you setting up the plugin to apply compression.
Using the GR Meter is the best way to tell how much overall compression is being applied. As stated earlier, you will be wanting to set a level of -2 to -6dB in reduction, this meter is how you can tell how much reduction you are getting.
Lower the threshold and adjust the ratio until you get the desired reduction. Listen closely, however, so you don’t ruin the integrity of the audio by overdoing it. For your purpose a relatively high threshold that captures most of your loudest parts, and a ratio of between 2-4:1 would be your best starting point.
Not every compressor has a Knee setting Sometimes its a dial, sometimes it’s a switch and sometimes it's just a button. In general terms the knee setting implies how gradual the gain changes will start to apply once it passes the threshold.
A knee of 0 means there is nothing gradual about it, compression will take place immediately the audio passes the threshold. While a higher setting will more subtly introduce the compression until it is applied at its fullest, somewhere above the threshold. The higher the knee, the more gradual and smooth the process, but it does change the over character of the sound as well. So a gradual or soft knee isn't always the best approach.
In the image, the control indicates a knee of 0, or a hard knee, meaning there is nothing gradual about it. This is indicated in the graph above as a sharp bend in the line at the threshold point. Higher settings would be represented by a more curved, or soft, bend in the line indicating a more gradual process.
You might want to try compression with a softer knee, that way the gain reduction won’t sound as aggressive or punchy on your voiceovers.
This is how quickly the compressor jumps on any audio that exceeds the thresholds, and beats it into submission. Its usually measured in milliseconds. A slower attack setting will mean that it will let more of the initial attack transient pass through before the compressor begins to get to work.
In many cases letting a bit of the initial audio past the compressor helps with intelligibility, also making for a less dull sound. You only normally need a fast attack if you really have to manage your peaks quite aggressively. At that point, though, you may want to turn to a limiter. Effectively its just a glorified compressor, but fine tuned to handle peak processing a bit more elegantly.
Try starting with an Attack setting of between 4-15ms for voice overs.
This control is also measured in milliseconds, usually. It indicates how quickly you want the compressor to stop reducing the level of the audio once it passes back under the threshold. You might think the quicker the better, but that could introduce some nasty side affects such as distortion. Having it set too long, on the other hand, could also diminish the transients that follow. The idea is that you need a balance somewhere in the middle.
Firstly, if you have an Auto Release option, then use that to begin with. On the image it is indicated by the Auto button next to the release control.
When this is engaged the compressor will perform a bit of mathematics and try to calculate the most optimal release time for you. It can be quite effective, but not optimal if you are really into sculpting your own tone.
To set the Release manually, usually requires careful listening and intense studying of the Gain Reduction Meter. Use the meter to observe how quickly the compressor lets go of the audio and returns to the state of no compression, or 0. The idea is you will want the compressor to reach roughly 0 before it begins processing again. In the case of speech it should relatively quick, but not too quick.
If you feel the audio is too punchy or possibly distorted or maybe it could do with more overall energy you could try increasing this parameter a little. If the audio is too dull, the meter too slow and almost never returns to 0, and you have an Attack longer than 5-10ms, then chances are the release is too slow.
Speech and vocals can be quite difficult to determine a good release setting, since they behave quite unpredictably. Other peoples opinions may vary, but try a setting of about 50 - 500ms. Start low and try increasing it until it sounds smooth and energetic, but without it sounding dull or less intelligible.
This control is sometimes referred to as Blend or a Wet/Dry knob, but not many compressors have a control like this. What it does is enable you to blend a bit of the original signal (Dry) with the processed signal (Wet). This is also what’s known as Parallel Processing.
In many cases you would want this control set to full so that it passes through 100% off the effected audio and none of the dry. At times, though, it still may prove useful. Particularly in the case where you may have been a bit heavy handed with your use of compression. If this were true you can try blending some of the original back in by adjusting the control and see if you can get a nice balance between the two.
If your compressor does have a control like this then start with it at 100% wet and apply your processing. At some point later when you come back to your project with fresh ears, have another listen to both the compressed and uncompressed version, use the Bypass button for this. Then try blending the two together until you find a happy medium. This method works quite well actually.
Sometimes Output, or just Out. This a simple output volume level. Any work done with compression ultimately affects the overall volume so you would use this control to compensate for the changes in level. If the input level is set so that it peaks between -1 and -8dB without it being adjusted by the compressor, then ideally it should leave the compressor with roughly the same overall reading.
You may have to flick between meter modes to get your input and output levels right. You can also use the Bypass switch to turn the effect off and on using your ears for an accurate judgement
If there is an Auto Gain or a Auto Make Up control, you may like to try it. It might help in balancing the output level for you. More often than not they tend to not work too well, sadly. At least, that’s been my experience.
As you can tell by the image above, there can be a lot more to compressors than meets the eye. Although, I hope I’ve detailed enough of the basics so that it can be of use to you, but not too detailed so that it becomes confusing or hard to read.
If you have any questions, don’t be afraid to ask. If, however, you would like to try something simpler first, have a look at the following post…
Here are a few compressors designed for simple use. Just because they are simple doesn’t make them less useful. However, the will have limits to what they can achieve.
I chose these particular compressors because they are currently on sale at Waves. So if you found yourself in desperate need of something here, at least it would be more affordable.
One Knob Louder
This is the simplest of all three. As the name suggests it only has one knob. Hard to screw up!! The higher you jack it, the more compression it applies, in an intelligent way.
This one might be the most useful to you. It has two controls plus an output level. The Low Level control will bring up the quieter parts of the audio, while the High Level will manage the loudest parts.
This plugin works great for post-production, although you might want to be weary that you don’t raise the noise floor too much, or that you over use it in general.
Butch Vic Vocals
Admittedly, this plugin was designed with vocals in mind, but I have also found it to be just as effective on dialogue as long as you don’t over do it. It offers more than just compression and makes for some really nice lo-fi effects for emulating sound sources like telephones, radios, loudspeakers, etc.
It comes with a basic eq, that you can use to help tune the tone of your dialogue. There is a Deesser to help manage anything to sibilant. It has a single knob Compressor, careful not to push it to hard or it will go gritty. There is a Focus control to help push the voice to the foreground, thats great to use if it has to cut through music or effects. Finally, this plugin also features Saturation controls with its own dedicated with Hi-Cut and Lo-Cut, this is where you can really distort your sound. When used in conjunction with the Hi-Cut and Lo-Cut found in the equaliser you can produce all sorts of interesting lo-fi effects.
This plugin is a bit more complex than the other two, but you get some great creative features along with it, plus lots of flexibility.
Hope this helps.
First of all I would like to say a BIG THANK YOU two you guys for the great help and detailed reply. Sorry for my late reply.
I have to admit that I have learned things but still I am VERY lost and don't fully understand how to tackle my task.
Here is my problem: I get a lot of street interviews / vox pops and the DB will constantly be at around -21DB and going up to -3 DB when ONE person is talking. So its ONE and the SAME person and his DB has side wide range. Now people will complain about it and what I am doing is: I am adding points in adobe premiere (using that pen tool) and moving those points up and down so that the audio stays in the same level - somewhat. 1) I feel like this is a terrible work flow 2) I don't feel like this is the right and best way to do it 3) I guess the audio that I am given is just not good (really going from -21 up to -3DB...isn't that a lot? But that's also what I get when voice actors do audio for me but god knows if that's right anyways...).
Anyways I am told that the max for my audio is -3DB, it can also go up to 0 if that happens for a split second or really shortly but what is prefered is an area of -6DB and -3DB- in that range.
Now I still do NOT find settings where i say: Lowest and Highest like minimum DB and maximum DB.
I do not understand why you give me ratios like 2:1. Well I mean I understand it ratiowise but Adobe Premiere doesn't put it like 2:1 (like it's a scale by 1 to 5). In adobe Premiere it's like: Ratio: 2.19. So they write a dot. I am wondering if 2.1 in adobe premiere is just like a ratio of 2:1).
Also I am using a filter called Dynamics. Here are the settings I have got:
AutoGate (I don't think I want to fiddle around with this because I don't want to filter anything- what would i even want to filter? I mean its often noisy because it's street interviews so maybe I can filter something with that AutoGate?)
Threshold/Attack/Release/Hold ; and a Soft Clip Button
Then I have my Compressor:
and an Expander and a Limiter.
you have described what does what above already so thanks. Still I do NOT know how to achive what I have described above. What would the settings have to be?
I am TERRIBLY sorry for asking like an idiot.