Compressing sound

October 30, 2022
 by Paul McGowan

In recordings, it is quite common for an engineer to use a compressor to better fit dynamics. Especially those of a singer.

I like to imagine the art of compression is like the proper setup of a subwoofer. Properly done you shouldn't notice it is there.

Unfortunately, that's not the norm. So many of today's recordings have enough compression on the voice that we hear it: the singer gets loud but the sound level does not.

Another side of sound compression is when it is used to bring up the level of an instrument or voice that is too soft (more accurately, this is known as expansion but since it is done with a compressor the name remains).

When we use compression to make something louder the same rules apply. It works only if it sounds natural.

Sounding natural is the point of any recording tricks and techniques from EQ to compression.

In the right hands, it can makes magic.

Subscribe to Paul's Posts

43 comments on “Compressing sound”

  1. Natural?

    These are all interesting pie-in-the-sky adjectives that we use when trying to convince ourselves that
    these are achievable from a 2-channel home-audio rig.

    "Sounding natural" from a recording, through a home-audio rig, is different things to different listeners.
    I've heard some, what I would call 'magic recordings', but do they sound natural, or would they sound natural to another listener?
    Has someone got a coin that I can toss?
    Heads or tails?

    Compression is like any other recording tool...use it sparingly & only if you absolutely have to.

    1. Is it “natural” when a single sound source (a voice) is reproduced by two or more loudspeakers each having some 5 or more drivers each being single sound sources and feature different designs (ribbon, dipole, ….)? 🙂 You will always and inherently get an unnatural artificial sound! I recommend: back to basics! Hey, what about crosstalk cancellation and recording techniques matching the intended goal of “natural”? 🙂 🙂

  2. According to global industry data, only 8% of music listening is done on hifi systems. Radio is out front at 29%, phones 27%, computers 19%. Hifi snobs may turn their noses up at the compression used in Adele’s latest album, but it was released around November and was the top selling album of 2021, selling twice as many units as any other release, almost 5million units and 862,000 on vinyl. It is ridiculous to judge music that is most likely to be listened to on a smartphone or laptop using a $100,000 audio system. Recordings must be engineered to take account of how people listen to them, and mostly it is not in high fidelity systems. The global recording industry is very strong at the moment, changes rapidly and seems very well attuned to the needs of artists and their audiences.

      1. A mate said,”come and listen to my new car. “.
        The hifi was as good as Frankie’s home superfi system.
        The car was junk. He knew it But gee that car rocked.
        Let’s us not be snooty about all ICE (In Car Entertainment)

  3. Thank goodness the old toolbox photos have been found. 🙂

    Meanwhile, back to the topic. It’s unfortunate the compressor so often ends up in the wrong hands and the way it is utilised today. Too many times I have heard a song on the radio, bought it and then been disappointed with the results on my system. Another one never to be played again but as Steven points out, this music is made for its audience. Interesting how audience and audio start with the same four letters, audience originally meaning a group of listeners.

    As audiophiles we are a group but tend to listen solo. From the figures quoted we are clearly under threat, an endangered species. Mind you, I’d like to know what criteria that global industry data used to define a hifi system. So many say “I’ll put it on the hifi” yet when you see the system it’s anything but. We need protection or we could become extinct.

    1. The use of compression and listener preferences is well researched.
      Here is a programme on the BBC "Compression versus Art"
      You might be interested in Tevor Cox's book "Sonic Wonderland: A Scientific Odyssey of Sound".

      Trevor Cox refers to a famous instance where there were two versions of a Mettalica album available and fans complained about the compression on the CD release. You can hear the with/without compression on a laptop.
      The 8% does not include smartspeakers or bluetooth portable speakers. So the hifi 8% will probably include any quality of 2-channel wired speaker system, so compression is probably equally important to many of them. So for the hifi snobs, the figure might be far less than 8%.

      1. Thanks for posting these links SNTBCWS. I started to listen to the first discussion and it didn’t take long to realize it was a real eye-opener and that spilled over into an ear opener.

  4. Human hearing does not like large dynamic jumps in living rooms. That's why compressors have their justification. When used properly, these devices can work wonders. Who wants dynamic jumps of 40-60 dB? Most recordings are in the 10-20dB range, and not without reason. This is different for live performances, e.g. in stadiums.

    1. Opa-Alli,
      True...I listen mainly to Rock 'n Roll at home & when I'm pointing my dB meter at my rig it's
      usually indicating between 88 - 108 there's that 20dB range that you mentioned.
      And this meter reading is fairly constant for original Redbook CDs & remastered ones.

    2. I have observed the following with respect to dynamic range and my wife. On our main audio system if the first track of the music has loud passages she tells me to turn down the volume. Then as she adjust to the music by the third track she will tell me to turn up the volume.

      I have a decent 2.1 audio system that supports our TV watching. First, when watching broadcast or streamed programs that have loud commercials I always mute them. Second, when we watch bluray videos of rock concerts she never complains about loud passages. However, third, when we watch action movies that have loud crash-bam scenes she always complains about the increase in the volume.

      1. Tony,
        When I start my listening sessions, I start at around (in the range of) 77 - 90dB but by three-quarters of the way through (Rock 'n Roll) I've settled into my usual 92 - 106dB range.
        For jazz & acoustic female/male vocals I'm usually in the 80 - 94dB range ✌

  5. Anti Compression

    It would be cool if component like a DAC or preamp had an Anti Compression button that could decode or strip the compression from a recording?

    For instance the Patricia Barber Cafe Blue has two different versions. The “stripped”version sounds significantly better than the original release. If you’ve not heard the “stripped” version your in for a real treat.

    I would guess this would have to be in the digital domain. It would seem as simple as programming an algorithm that could detect compression then “uncompress” the recording. This would allow audiophiles to combat the “loudness” wars that is so prevalent in today’s recordings.

    1. Too bad it isn't "simple" to design such an algorithm. I have long pressed on my brightest programming friends for such a thing and have been assured it is anything but simple.

      And, f you think about it that makes perfect sense. How does a program "know" when something has been compressed?

      The best compression I know of in the recording arts is manual. I use it myself when mixing for Octave Records. It's called riding the faders. It works because you're listening to the end result and you work until you get what you want without being heard.

      1. Paul, just like the engineer at Sheffield Lab used to “ride” the lathe pitch to allow for extra loud and/or dynamic passages to have “groove room”.

        Since “compression”is not *encoded* into anything, no genuine “decoder” is really possible.

        One would end up with a sort of a bastardized “reverse AGC” with likely horrid results 😉

    2. sgrowan, that was the goal of the dBx 3BX dynamic range expander back in the early 80s. It had 3 bands and actually worked quite well helping to add some life back into over compressed recordings.

      But as Paul mentioned, compression is a non-linear, lossy process, so there is no way to accurately "decode" it. The processor can't know what was compressed vs what the natural dynamics in the original recording were. So it has to guess, with varying results.

      1. I had the dbx 118 back in the last 1970s. It was also a dynamic range expander, but less complicated than the 3bx (read dumber). That was the age of vinyl as that's what there was. Not a CD to be seen on the distant horizon. It did bring some albums to life and they sounded great. Then again, some sounded worse. There was no free lunch here.

  6. All this talk of the recording process is 😎 and educational. So thanks.
    Yet that process is something few of us have absolutely no control over.

    I pose this question…. What makes a recording technically and sonically superior? The format it’s recorded in or the amount of processing (mixing / mastering)? (I’m guessing it’s neither of the above, but more to do with the human factor)

    1. It's a great question, Mike and one I have been contemplating as of late. I would say that assuming the recording sample rate and resolution is up to our standards (as a given) then the factors that control it seem to be one of simplicity in miking and getting the right mikes in the right places.

      I would further suggest that the use a stereo microphone for the main recording with a few spot mikes to fill in those soft instruments makes for an exceptionally live sounding recording.

      I just recorded a local swing band in the studio. There were 4 musicians and we had them circle around a single Blumlein stereo microphone. I then spot miked the stand up bass. That was it. The work sound incredible. As if they are right there in the room with you.

      In speaking with "the regular kind" of engineers, they would have separately miked each instrument and then reassembled in the mix.

      That artificial building of the sound field is rife with problems.

      1. Thanks Paul!

        You used a term that we (me?) don’t hear 😀 much. Sound field. I’ll have to dwell on that and what it means, but my gut says that’s one of the keys.

        1. Gottcha covered, Jazznut. One of our upcoming releases is Gabriel Mervine Quartet live at Nocturn jazz club in Denver. He partners with a brilliant young lady playing (of all things) the accordion and along with a drummer and bassist, they rock the house with Parisian jazz like you've never heard before. It's a killer release I am just finishing up with now.

  7. I wonder how they accomplished the stripped version of Cafe Blue? I know it wasn’t a one button job.

    But someone accomplished it or maybe a version was saved before all the compression and other mastering techniques were used?

    Has anyone else heard the two different versions of Cafe Blue?

  8. I own all three versions of Cafe Blue. To quote Gus Skinas “It is well known that the original of Cafe Blue is digital. The mix of the recording heard here was done in 2011 at Hollywood’s Capitol Studios where the multitrack was mixed in the analog domain and analog chambers were used to produce ambience and reverb. This SACD nicely portrays the impact of Capitol Studios on Cafe Blue. Actually, that is a bit of an understatement. One of the great benefits of DSD is the ability to preserve the integrity of the very quiet parts of the music.” He goes on, but that’s the gist.

    I like the LP and DSD versions over PCM for sound, but of course they all sound amazing.

  9. Here in Germany there are currently television broadcasts that are broadcast with very dynamic sound. You have to constantly adjust the volume with the remote control. Thousands of people are already complaining about this dynamic every evening and want to go back to the compressed sound recordings. That's how it is with large dynamic leaps in music via our hi-fi systems, right?

    1. The first DSD album I ever bought (Shostakovich 2 / 15 Royal Liverpool on Naxos) was so quiet in the first movement I thought it was faulty.

      It put me off DSD from the very beginning that DSD producers would use too much dynamic range.

      Play Mahler 5 / Barbirolli on vinyl from 1969 and it has more than enough dynamic range. Some modern recordings have less.

  10. Hi Paul. Actually, expansion (often called downward expansion) is different from compression. It is used to make quieter sounds even quieter, and is often used in single and double ended noise reduction systems. In double ended systems such as dBx, it's called compansion - compression on the front end and expansion on the back end.

    Making quieter sounds louder is still a form of compression - you are reducing the dynamic range. Think of it as compressing the loud sounds, but then adding make-up gain so the the loud sounds are as loud as they were before, so now the soft sounds are louder.

    Just wanted to clarify the terminology so readers aren't confused.

  11. When compression isn’t used liberally or tastefully it can be the single greatest killer to suck the life out of your music.
    It is so sad, really. I also think over compression should be a crime against humanity. lol.
    God. I hate it.

Leave a Reply

© 2022 PS Audio, Inc.

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram