Compression and frequencies

Question: does compression affect different frequencies differently?  More specifically: if a compressor set at the same settings for attack and release were used to attenuate a 100 Hz signal and then a 5000 Hz signal, would the frequencies react differently to the same settings on the same compressor? And if they do, what are the implications for sound, like a vocal, which is composed of many frequencies, some high and some low? Thanks for your thoughts.

The simple answer is “no”, compression does not affect different frequencies differently.

However, compressors do REACT to different frequencies differently.

Bass frequencies have more energy/power for the same perceived loudness, so lower frequencies will trigger more compression than higher frequencies.  Thus, you’ll see many compressors with different filters in their side-chain or detector circuits that allow you to change how they react to lower frequencies.  For example, my hardware Drawmer 1968 compressor has a “BIG” switch that keeps filters some of the low-end from the detector circuit, thus keeping the compressor from reducing the gain as much for low frequency content.  My API 2500 hardware compressor (as well as the Waves plugin emulation) has a three position “thrust detector” switch with “loud”, “medium” and “normal” settings, which offer different amounts of low end roll off to the detector circuit.  These are both still single band compressors, so the actual compression action affects all frequencies the same… those detector circuits just vary how the compressor reactors to low frequency content.

Now, there are other things in compressors (mostly analog compressors, or proper digital emulations of analog compressors) that can affect different frequencies different.  These are things in the analog signal path such as tubes and transformers.  Many of the classic compressors are “classic” because of the sound that they impart to the audio that passes through them.  Good transformers add a bit of saturation and definitely change the EQ balance of audio that passes through them… same with tubes, depending on how hard you drive them.  Many times compressor are purposely driven hard to add some distortion to the sound.

Another way that a compressor can affect frequencies differently is when you are using a multiband compressor.  These split the audio into several different frequency ranges and the compression can be adjusted on each range separately.

Also, you can use the side-chain input on a compressor to have the compressor react to just a certain range of frequencies.  A good example of this is a De-Esser.  For De-Essing something like a sibilant vocal track, you would set up a filter that emphasizes the troubling frequencies (you could use a High Pass or Band Pass filter) and you feed that filtered signal into the side-chain input of the compressor, while the full audio passes through the regular input of the compressor.  Whenever there is extra sibilance (above the threshold you set) the compressor reacts by bringing down the audio level.  Again, this is still usually just a single band compressor, so the compression is still affecting all frequencies the same (pulling down ALL the audio), but, in the case of a De-Esser, gain is only reduced when there is excess sibilant frequencies.  Dedicated De-Essers, as well as plugin versions, contain the filter circuits internally, so you don’t have to use separate EQs and patching to make it work.  A multiband De-Esser would be better for something like a drum kit where you want to control the brightness of loud cymbal crashes without affecting the lower frequencies of any other drums playing at the same time (just use one band of the compressor for the high frequencies only and compress as needed).

Hopefully that helps!

I have query, Please explain Which is the best low-bitrate audio compression algorithm? Guide me on the topic how to select the best one.

You’re talking about a different kind of compression than what we are talking about in this particular message thread.

You are talking about compressing the data stream to reduce file size and make it faster/easier to stream audio files over the internet, or just smaller files for email or download.  Most of the audio formats that do this, such as MP3 or MP4/AAC, use a lossy form of compression, meaning that they try to analyze the audio and throw away parts of the audio that the algorithm thinks you won’t notice (such as softer content that is masked out by some other louder content).  Obviously, the more you try to reduce the bit rate (to make the file size smaller), the more audio gets thrown out and the more you hear the affect on the audio.

There are several different encoders for each of these various compressed audio formats, and some work better than others at getting a low bit rate while still maintaining a somewhat acceptable sound quality.  A lot of it depends on the type of audio you are encoding as well… some encoders work better for certain types of music, while different encoders will sound better on other kinds of music.  The only way to know for sure is to test them out and compare using the music you want to encode, and pick the best option for that particular piece of music.

A new plug-in was recently released which allows you to compare various codecs and algorithms in real time to see which works best for your music.  You can read about it here:
http://www.sonnoxplugins.com/pub/plugins/products/pro-codec.htm

You may also be interested in looking up FLAC, which is a lossless audio compression algorithm.  It will reduce the size of audio files without any loss of quality.  You can’t get the huge data size reductions with it that you can with MP3 or AAC, but FLAC doesn’t change the audio at all… it just finds a way to compress the data in a way that can be fully restored on playback (just like ZIP codecs do lossless compression on other types of data files).

Posted in Ask MusicTECH! Tagged with: , , ,