I’ve noticed that there were previously questions about Compressor filter but all threads are closed. Thus please let me return to this subject.
Do you often use Compressor for lector type audio? If yes, when? If not, why?
Where could I find good description of parameters? In some threads I found this link: Compressor - Audacity Manual However here not all parameters are described. Ex. for example what means and how works Knee radius and Makeup gain?
Could you give also some examples of your Compressor settings you use everyday with short explanation why these settings?
Is the order of filters important? Ex. when having Normalise Two Pass, High Pass, Equalizer Parametric, Limiter, which order seems to be the most reasonable in your opinion?
Thanks for all remarks.
I very often (almost always) use the audio compressor filter on a track with a human voice, very often on the recording the announcer speaks either very quietly or very loudly. The normalization filter cannot help if the audio file is long and has a very wide dynamic range.
Compressor settings cannot be universal due to the fact that there are so many scenarios for using this compressor. But there are some general guidelines. If you want to do subtle and soft compression, set the ratio to 1:4 or 1:5, and the compressor in RMS mode, it will do soft compression. But if you want to get a radio sound like podcast bloggers, you need more aggressive compression, ratio 1:10 and higher, compressor in 50% rms/peak mode, or even 100%, in peak mode, reduce the attack, reduce the release. This is where you get the “brick” sound that bloggers, podcasters and radio hosts love so much. Well, the main thing is to guess the acceptable level of treshold by ear, it depends on what volume level the compressor will work with. If the “gain reduction” monitoring jumps above -12 dB, your sound is in danger of becoming a brick and artifacts may appear in it. Keep the compression at about -3 -6 dB.
As for the equalizer, depending on the timbre of the voice, it is recommended to cut off ultra low frequencies, which may be due to the lack of a foam filter on the microphone itself. Depending on the timbre of the voice, these frequencies may be different. But as a rule, everything below 60 hertz is garbage on the voice track and it is desirable to reduce these frequencies. I usually put the equalizer BEFORE the compressor, so that the already filtered signal without parasitic frequencies goes to the compressor.
Thanks for this @dimadjdocent
Thank you for this explaination.
What is the role of this RMS parameter? If I change it from 0 to 100, the difference is almost not noticable or even not noticable. As for now I set it to 0.
Is that somehow related to Threshold parameter? Or it’s only very indirectly basing on many experiments?
What is the role of Makeup gain parameter? If I set it to 0dB I have impression that voice is a bit “flat”. If I set it to higher values, suddenly sound monitoring (audio peak meter) becomes quite often red.
If the compressor is running in RMS mode (in shotcut this is 0), the compression is softer, since in this mode the compressor’s task is to compress the average volume of the signal, the compressor will skip short peaks. The compressor reacts differently to attack and release. If the compressor is in PEAK mode (shotcut set to 100), it will try to compress the peaks of the signal and this will be more noticeable compression, especially at low attack and release values. As a rule, RMS compressors compress voice, bass guitar and any other sources with a smooth sound. PEAK compressors compress a more aggressive signal such as drums.
The Treshold parameter of the compressor is needed in order for it to understand at what volume it is necessary to work with the signal. If set to 0, the compressor will never run. If the value is set to -10, then if the audio track has bursts of volume above -10 dB, the compressor will respond and compress them. If the track is much quieter than -10 dB, the compressor will also be idle and you need to lower the Treshold parameter even lower until the compressor begins to respond to the sound.
After the compressor has done its job, the volume of the signal will inevitably be quieter than before compression, “Makeup” is needed in order to compensate for the loss of volume. As a rule, this parameter is just a volume knob, if you put a gain filter after the compressor and add volume in it, the result will be the same, the Makeup parameter is needed in order to make the signal louder and not use third-party plugins and filters.
The Makeup parameter is adjusted after the basic parameters of the compressor are set and the compressor has compressed the signal. In order to strictly limit the sound and prevent it from flying into the stratosphere, it makes sense to put a “limiter” filter after the compressor, this plugin will severely limit the sound. Most often, I set the limiter filter on the output channel, so that the limiter already limits the sum of all channels and not just a single one.
In this video, I demonstrate in what sequence I use audio filters to even out my voice. I specifically found an example of a voice in a wide dynamic range on the freesound site to show what a compressor is for and how this voice can be processed using the built-in audio filters in shotcut. Please note that the woman pronounces the second part of the phrase much quieter, and at the beginning of the recording there are obvious bursts in sound. After the compressor and limiter, the sound became almost equally even.
p.s. I almost do not understand English (I communicate through a translator) and I don’t know what this woman is talking about, if she says something indecent, I apologize for this entry.
Big thank you. It makes the situation much more clear.
The only thing which is still not fully clear is this “knee radius” parameter. How would you explain its role?
This is one of the most difficult parameters to explain. But I’ll try. The Treshold parameter specifies at what volume the compressor will operate, for example -10 dB. Using Knee, you can force the knee to strictly follow this parameter if Knee = 0, which corresponds to a “hard knee”, or soften it by increasing the Knee parameter, this is called a “soft knee”. If the parameter is closer to the soft knee, then the treshold value is not so strict and the compressor can operate already in a certain range, for example, from -12 to -8, operate unevenly, at -12 it begins to operate but is barely noticeable and up to -8 it operates at full strength, previously specified parameters.
Thus it’s probably better to set knee to few dB to get result sounding more natural.
Big thank you for all these explanations! Very helpful!
The default knee value that shotcut offers can be considered universal, as practice has shown.