Jump to content
  • entries
  • comments
  • views

Compressors, Limiters, Loudness War, and Mastering Inception


Years ago, I was lucky enough to spend a decade as a live sound mixer and recording studio engineer in Western Canada. While most of my sound engineering experience was with rock bands, I had the opportunity to work with many talented musicians, recording/mixing many different types of music, from folk, country, jazz, choirs, and classical. I spent a quite a bit of time working with compressors and limiters, so I thought I would share some of my experiences with them, along with mixing and mastering. This summarized info may assist folks in their decision criteria for selecting high resolution downloads.


Here is a story to illustrate why compression and limiting is inevitable in rock (actually most multi-track recorded) music. I remember recording a bass player in the studio using a direct box. I wish it were this one: http://www.musicvalve.com/directbox.html This would be your audiophile direct box - a work of art and beautiful sound. I figure Computer Audiophile’s could appreciate that even in the pro sound category, some manufacturers are obsessed with sound quality. Back to the story, the point is when the bass player played his part and depending on how hard or soft he played on the strings, produced a wide dynamic range.


I was using a Sony 24 track analog tape recorder with 2" tape, Quantegy (Ampex) 456 brand: http://www.quantegy.com/specsheets/PDF/456.pdf . The general idea is to record as "hot" as possible on the tape so you get the best signal to noise ratio. The tape had quite a bit of headroom, (+10db) so it was popular to "saturate" the tape just a bit (ok sometimes slammed) that also gave the music a "hot" sound. A VU meter is used to measure the level of audio signal going to tape. Side note, that's the short version. The long version is that each strip in the mixing console had a VU meter and that was calibrated along with a corresponding VU meter on the tape recorder, x 24 tracks or more - the calibration took a long time). The idea is, on average, to record around 0db or +3db is hot, and the tape had over 10db of headroom.




The issue is that when the bass player played softly, the level on the VU meter was too low. So if I increased the gain on the input preamp on the console, I could not control the amount of headroom and sometimes when the bass player played harder on the strings, that would "pin" the meter and the result would be preamp and/or tape distortion. Further, given the wide dynamic range, the bass really did not "sit" well in the overall mix.


You may be surprised to learn how sensitive the VU meters are. It does not take a wide variation to swing from -20db to +3 on the meter. While I used the bass player to illustrate a point, recording music of any type, whether it be with a direct box or mics naturally have a wide dynamic range.




Even the most consistent musicians I have worked with, in the sense of even playing, still produced a wide dynamic range. So what choices are there to get the best signal to noise ratio, without distorting the tape (or direct box, input preamp, digital recorder, etc.)? I can't very tell the bass player to change his playing style or sound as you can imagine how nervous people get when they are being recorded in the first place. I have a lot of respect for muscican's to go on tape.


Enter the compressor. I think Wikipedia does an excellent job of describing how a compressor works: http://en.wikipedia.org/wiki/Dynamic_range_compression I am not going to rehash the operation or details here. From a critical listening perspective, depending on the compression threshold, ratio, attack, and release, the compressor will “shape” the waveform passing through it. Folks with an ear for the compressor can set it just enough to maintain good dynamic range, but not distort the tape or digital overload. Judiciously used, the idea is that the (critical) listener would not hear the "compression" working.


The point being, in order to get a good (and even) level on tape (analog or digital), with good signal to noise ratio, a compressor is more often times than not used on every single channel of a 24 track tape recording. Especially drums, voices, or instruments that have wide dynamic range and are closely mic’d typical of any studio/multi-track recording. High end mixing consoles like Neve and Solid State Logic (SSL) have compressor/limiter plugins for every channel "strip" on the console. There is usually a line/mic preamp on the strip, along with EQ, usually parametric with full range of adjustment over multiple bands, headphone mix level, effects level, and then the channel fader or volume control that then feeds the master buss or subgroup. A lot of electronics in the chain even before it hits the tape electronics and tape itself.


After spending many years recording and mixing sound, one of the side effects is that I can easily hear compression/limiting on virtually every single song I listen to. I am so intimate with it that I can probably estimate the compressor settings used and in some cases I can even tell the brand of compressor being used. Here is an example of someone that intimately knows the sound of UREI's line of compressor/limiters: http://www.gearslutz.com/board/568476-post2.html


How does this effect mastering?


So the compressor is not only used, most of the time, on a track by track basis for recording, but sometimes even on certain tracks during mix down. This happens when the individual track still has too wide a dynamic range to "sit" in the mix when it was put to tape. Additionally, there is usually a 2 channel compressor/limiter used on the mix down buss as well, that then is also going to a 2 track (i.e. stereo) recorder, whether analog or digital.


So 24 track mix down --> 2 track master


Even after the final mix has been created, sometime folks want to fiddle with the mix again and rather than mix down from the 24 track, they use the 2 track mix and feed it through the console again, apply whatever "processing" is required and generate another 2 track mix down (whether analog or digital). I am guilty of that myself.


2 track master --> processed 2 track master


That then goes to the mastering lab (or a copy for fear of the original being lost if the engineer did not mix down from 24 track to 2 masters) to be mastered onto whatever media for distribution. Even during the mastering process may repeat processing as described above. The Wikipedia article on mastering does a decent job on describing all of the steps: http://en.wikipedia.org/wiki/Audio_mastering As you can see by the time the music gets to you, the actual copy may be several generations away from the original master. This is especially true when discussing analog generations.


So what about remastering?


Again, I feel Wikipedia does a good job here: http://en.wikipedia.org/wiki/Remaster and I am not going to rehash. But this is the point of Mastering Inception, it is not clear as to the source of the material, where did it come from? How many generations is it? What was done to remaster it? Is it a different re-mix direct from the original 24 (48 track, whatever) track tape? Or is it a direct transfer from an analog 2 track master with no processing? Or maybe direct from 24 tracks to digital. Or is it all digital? Or did it get killed by the loudness war? These are excellent write ups on “The Death of Dynamic Range”: http://www.cdmasteringservices.com/dynamicdeath.htm and “What Happened to Dynamic Range”: http://www.cdmasteringservices.com/dynamicrange.htm


What is really disappointing to me is that even as far back as 1982, digital recording pioneers like Peter Gabriel and his all-digital album Security, that was recorded in his home http://en.wikipedia.org/wiki/Peter_Gabriel_(1982_album) has awesome dynamic range.


Mastering inception? Remaster? Master? Premaster? You may wake up one morning and come to Computer Audiophile to find the thread on HDTracks have come clean and they were are all CD copies that have been upsampled ;-) Just kidding of course. However, I searched all over HDTrack's site and if you read the wording in their Mission Statement: "It is our purpose to allow our customers access to the largest online library of DRM-free CD and DVD-Audio quality downloads complete with liner notes in a PDF format." And read from their About page, "Finally, audiophiles take note. HDtracks offers select titles in ultra-high resolution 96khz/24bit files. This is true DVD-audio sound quality for music lovers that demand the very best! " It almost sounds that their sources are from CD's and DVD-A and who knows, have they been upsampled?


As a side note, it would be nice to see Fleetwood Mac Rumours remixed from the 24 track master. http://www.soundonsound.com/sos/aug07/articles/classictracks_0807.htm But if you read closely, given what was going on, and the tricks used by assembling bits and pieces from multiple 24 tracks is likely never to be reproduced again. Also consider that produce/engineers use track sheets to keep track of all the eq, effects, compressors, etc., being used. SSL consoles and others used computers to automatically keep track of fader positions, mutes, and the like so that the mixing engineer had more hands. Finally, consider that most pre- 90’s recordings were recorded, mixed on gear that simply does not exist or exist in a working fashion. So most material from the past, is likely to have come from an analog 2 track source. But on the other hand, it is a snapshot of history, never to be repeated again, but can be played over and over. Pretty cool if you ask me, provided of course that I am getting the best possible transfer.


Recently, I downloaded Tom Petty and The Heartbreakers - Damn the Torpedoes. Along with the download was a note from the Engineer and Producer. I have attached the note to this post, but here are the magic words I like to hear:


“We’re committed to finding the highest quality way to get music from Tom Petty and the Heartbreakers to you. We want you to hear it at home the same way we hear it in the studio.

We think the best audio option for your computer or media server is FLAC. FLAC is a high quality file that is the best sounding format for downloads. Unlike mp3, which discards elements of the audio to make the file size smaller, FLAC is a “lossless” format, which sounds the same as the source files it was created from. We made the FLAC files from the same high-resolution uncompressed 24-bit 96K master stereo files we used for the vinyl and Blu-Ray versions of Damn The Torpedoes: Deluxe Edition. When we compared those files to the FLAC’s, the waveforms tested out to be virtually identical.

FLAC captures the full dynamic range of the music from the quietest to the loudest sounds. Because of this (and because we are adding no digital compression) it will not sound as “loud” as a standard CD or mp3. To compensate for this, turn up the volume!”


So not all is lost. If we the consumer keep putting pressure on hi-resolution publishers of music to include where the source of the material came from and what the remastering process was, the quality should improve. And that really is the trick to discerning the sound quality of a new hi-res download – where did the source come from for the hi-res version and how was it remastered. Sounds simple enough (no pun intended), but trying to get that info is like pulling teeth and entering Mastering Inception.


A couple of caveats. I glossed over a LOT here. Hopefully, it is just enough info to give a you a flavor of compressors, limiters (the latter I really did not touch on, but virtually every LP that was mastered always had a limiter in the chain to prevent over modulating the disc while cutting) and mastering, premastering, remastering, and in a lot of respects, all the same thing, but at different stages between the 24 track analog/digital tape and you.


Another caveat is that Classical or other 2 channel recordings are typically recorded/mastered without compressors, limiters, or as little as possible in the signal chain and using as high-end mics and components as possible. I used to own a Sony PCM F1 Digital 2 track recorder when it first came out along with a pair of Crown PZM microphones (plugged directly into the unit) that I got to use recoding a number of choirs, orchestras, chamber music, including a cappella. Lots of fun and most of the time was spent getting the mics positioned in a stereo array and watching nervously at the level meter in hopes of getting a good signal to noise ration without clipping or having to ask the artists to perform yet another take.




Finally, Digital Audio Workstations (DAW) has revolutionized the recording industry. http://en.wikipedia.org/wiki/Digital_audio_workstation It has dramatically reduced the cost of having a “recording studio”. I have even heard marvelous uncompressed sound out of Garage Band http://www.apple.com/ilife/garageband/ with some good DI boxes and one or two decent mics. To a large degree, DAW’s being digital, don’t suffer near as bad when it comes to copying as it is all in the digital domain. Theoretically the same. However, I have heard of mastering engineers that will put in an analog loop so they can use their favorite tube compressor to get “that sound”.


My point in all of this is that it really makes a difference to find out where the source of the hi-res copy came from and how it was remastered. Those two decision criteria should assist in your hi-res download choices. Easier said than done :-)


Happy Listening!


Mitch<p><a href="/monthly_2012_05/why_flac_dtt_pdf.f13313415b8a33dff447f16dcf18d719" class="ipsAttachLink ipsAttachLink_image"><img data-fileid="28082" src="/monthly_2012_05/why_flac_dtt_pdf.f13313415b8a33dff447f16dcf18d719" class="ipsImage ipsImage_thumbnailed" alt=""></a></p><p><a href="/monthly_2012_05/why_flac_dtt_pdf.974a6576f853f21558b75261871fc6ca" class="ipsAttachLink ipsAttachLink_image"><img data-fileid="28332" src="/monthly_2012_05/why_flac_dtt_pdf.974a6576f853f21558b75261871fc6ca" class="ipsImage ipsImage_thumbnailed" alt=""></a></p>


Recommended Comments

This is really good stuff, Mitch. As with your other posts, written in very easy to understand "layman's" terms.




Thanks for taking the time to write and post this.

Link to comment

Man, the best I read to date this year.




Mitch, really great (stuff) and thank you so much to put this in the prespective showing some real merits. I can only hope that you don't disagree all *that* much with what you did or had to do yourself, and I am looking forward to a next blog on how you would do it today (hint-hint). Not different from before ? also good. But mighty interesting to know for sure.




What I wondered (unknowledged in this field obviously) ...


You talk so much about "downmix" from 24 to 2 tracks. From the technical point of view to me that seems a lot of compression. Undoable actually.


Am I wrong ?




Super, really !



Link to comment

rancew, Peter, Chris, I really appreciate that.




re: lots of compression and what would I do different? Great questions. It was common practice to use a soft compression ratio (2 to1) at a fairly high threshold, with a slow attack and fast release. The resulting shape on whatever waveform passing though it, (i.e. kick drum, bass, vocals, overall mix, etc.), would gently kick in near the peak of the waveform and ever so slightly, compress at a 2 to 1 ratio. It was just enough to keep whatever was being recorded from pegging the VU meter, but still a high enough overall level to get a good signal to noise ratio for recording to analog or digital.




To answer your question on too much compression, yes, absolutely, if it is slammed on tape and/or distorted or anything other than how it is supposed to sound, it is very difficult to fix that, you are basically stuck. The basic procedure to get a good signal to noise ratio, without distortion was you listened and adjusted the preamp gain, any eq, the fader, and the threshold on the compressor. Listen, watch the levels, repeat as many times until your happy with the sound. Once it is on tape...




How do you hear a compressor? You can especially hear the gentle (and sometimes not) "squeeze" on the chorus of a rock/pop song where there are back up vocals or a guiter solo or anything in the song that increases the overall mix level, will likely trigger the compressor , whether on an individual track or a two track mixdown or remaster, ...




If you listen you can hear it. The compressor reacts very fast and is basically smoothing the top peaks of the waveform, so it usually follows the beat of the music (triggering usually on the drum kit). Over done by mistake, it sounds like the music is breathing or pumping with the beat or whatever individual track that is too high in the mix and "modulating" the compressor. Then again, if you like the song "When the Levee Breaks" by Led Zep, then this is a desirable effect to have the biggest possible drum sound (the huge vertical foyer that John Bonham's drums were recorded in helped immensely as well - the first to get "that sound").




You can especially hear a compressor when the threshold has been dialed in too low and the compressor is on most of the time. When there is a pause in the music, as the music decays, you can hear the compressor release and restoring to its original level. What that sounds like is a rapid increase in gain just when the sound is fading way, all of a sudden it gets louder. That's the sound of the compressor "releasing"




It is likely that the majority of what we listen to, especially if it came from a "close mic'd" studio environment, has been compressed in some way. Which brings me to what I would do different.




In 1982, Peter Gabriel recorded Security in his home and all digital. That's just shy 30 years ago. I have the the original (I think) Virgin release on CD. You remember the song, Shock the Monkey? It got quite a bit of air play over the years. The Dynamic Range is 15 if I am using this meter right.








The trend over the years seems to further compress down the dynamic range (I am trying not to include the audiophile labels). I thought it would have gone the other way, i.e. an increase of dynamic range. We seem to have better audio dynamic range capabilities in Blu ray movies than we do for recorded music.




With the dynamic range capabilities of today’s high-end digital audio workstations, and our high resolution audiophile playback systems, we should be at 20db dynamic range or listening to no compression at all.




Short story, a buddy of mine has Garage Band, and a real electric guitar, and composes/performs Groove Salad style of music for fun. He was shocked to hear the full dynamic range of what he recorded. I heard it myself on his Paradigm bookshelf/sub combo with a vintage 100 watt NAD power amp and separate NAD preamp driven off his Mac. He asked why doesn't our CD's sound like this? My answer, compressors.




Given the computing power, storage, and bandwidth capabilities a "software" studio can run on these days for a commodity price. 20 years ago, an outfitted recording studio that sounds good, cost $5 million. Today for $5,000, in computer and software with a handful of peripherals (mics and direct boxes) and a band, and with a little experience, you can sound better than the $5 million facility.




That's what I would do different, turn the compressors off, fully utilize as much dynamic range, focus on mic techniques, and record the band as live as possible.




Cheers, Mitch

Link to comment

Of course I read it with great interest again Mitch. It just fills some gaps again I'm kind of looking for.


FYI, like : http://www.computeraudiophile.com/content/Recording-Engineers-You-molest-our-recordings




... so from there we might be able to improve (software, hardware).




I found it refreshing to learn about how the "timing" can get off.




About increasing dynamic range ... I talked about this before, but I think something like 72dB can't be utilized anyway, so there's (IMO) a good reason to compress (I don't mean the dynamic type here).




What makes it possibly more interesting for myself, is the "counter productive" something can be like sustaining the impulse response 1:1 - like I do. So, if you'd imagine an instant pulse from 0 to max and this is completely decently followed by the electronics, what would happen when in the base this is wrong (like the release from the compressor you talked about) and now that gets emphasized. Attack the same of course. I'd call this a typicle example of playback devices possibly being too good for the material they play. Running into limits (like speaker drivers may not cope).




In the end it is (also) how the various types of anomalies combine. Like the used long postringing filter in a DAC will be able to let sound cymbals infinitely long, in the mean time showing cymbals as those with nails in them, just because it is about ringing - sheer echoing in this case - recursive to itself.




Slowly but clearly I now learn how it is possible that my own recordings sound infinitly better, just because I don't know how to destroy them. I already thought it would be something like that, but maybe now it is a fact. Sounds sad btw.


Oh, this is different from being able to take a good recording of playing musicians of course (the link I referred to nicely makes that clear).




Again, great stuff !



Link to comment

To my ears, compression is one of the final frontiers (although well behind proper microphone use) in making recordings that sound convincing.




Over the years, I've heard many engineers talk about making players more "even". To my mind, this is "music in the service of technology" instead of the other way 'round. Music and sound do not exist to make the recording/mixing/mastering engineer's life easier.




In real life, music and players are not "even". That is a big part of the emotional communication in music. To alter this, is to alter the emotional message, perhaps miss it completely, leaving only the shell of mere notes (if that).




It is unfortunate that most trainees and folks in "audio engineering school" are taught which knobs to turn but not taught to ask why (and have a good answer before proceeding).




And while it is true most mastering engineers will routinely have a compressor and/or limiter in their mastering chain, it is not universal and is certainly not (to my ears) true where the best sounding recordings are concerned. The great George Piros, when I knew him at Atlantic -long after his work with Bob Fine and Wilma Cozart on the Mercury Living Presence series- never used compressors or limiters. (Of course, his concern was musical sound and not mere level, which he knew is much better achieved with the playback volume control.)




I too never use compression. Not in original recordings, not in mixes and not in mastering. Oh, I like it as a "special effect" such as used to achieve the "whoosh" of Ringo's cymbals on certain Beatles songs but outside of that, never.




(Okay, I've actually used it twice, both times at the insistence of the producer and both times many years ago, when I was younger and much more shy about saying "no". One instance was for the cassette master for Pete Townshend's "Psychoderelict", because Pete wanted it that way. The other instance was for a CD I did with a producer (who shall remain nameless) who not only wanted the compression but also told me I couldn't add enough treble to make him say there was too much. (!?!?!?))




Compression does a lot more than eviscerate dynamic life. It tends to smear temporal response too and harm pitch definition in the bass. Some will say it "adds punch" to which I always ask how one can add punch when removing dynamics, where the punch lives.




One thing that is often overlooked in discussions like this one is the economic side of compression. Compression makes the engineer's job easier. When things are "even", the engineer does not have to be as watchful. More jobs can be done per day, often resulting in greater billing.




This is not unrelated to why vocals and bass in pop mixes tend to be in the center all the time. (Anyone ever wonder about this?) It is much easier to cut a vinyl record when most of the energy in the music is in the center. Once again, an example of "music in the service of technology".




Compression as an effect is one thing. Used to achieve level, in my experience, it will always do sonic and musical harm. There are many reasons why level, when desired, is best achieved with the playback volume control.




Best regards,







Link to comment

Hi Barry - nice place to meet again.




Some will say it "adds punch" to which I always ask how one can add punch when removing dynamics, where the punch lives.




I think I can explain this, but please feel free to openly reject my thinking (I only grant myself opportunities to learn !).




Take the example of The Leevee breaks, like Mitch explained it. Of course this (Zep) is a little (if not all) related to you, and looking at my previous post and title, I already there wanted to continue on it.




If one is able to let excel the transient response way better (like I can do that with my "software + hardware"), the first thing what happens is that the smear you talk about is no smear anymore. So, not from the filtering angle (please ask if this needs eloboration, but I think you will know). This makes especially bass waves more origin, or better - without distortion. More pure. LESS DEEP.




The "less deep" may not be easy to grasp in a glance, unless you know what I'm talking about in advance. So, imagine the (in 2D) tops of the wave to be smeared "sideways" a little. Or let them echo a little. Now you have two tops where one whould be. The amplitute won't be the same, but the merit is that this creates a slower freqency if we'd work this out. A resonating one (to itself). This is why the less accurate wave sounds lower (of frequency).




Whether this is caused by (smearing) filtering, or time-lagging compression, doesn't make much of a difference I think. The snappyness will be gone, and fatness will be there instead. More "lumpy" - less refined.




In my experience, when the artificial slam is removed (less ringing etc.), the real sounds surface, like all the different veils from the different toms. Less jitter does the same btw.




Of course removing that nuance-impeeding phenomenon (never mind how it is done) and you'll have it the other way around : slam is added. Or punch if you want.




Going somewhat further, punch as such is not a really dynamical thing. It is too slow for that. What I mean is, a hit on a floor tom will show punch (ahd it's a fairly slow transient) while hitting the rim of that same tom won't have punch at all. In the middle there's the snare. Does it have punch ? no. But it's faster, that is why.


(of course the sound pressure of punch needs a certain "range" in the headroom, but I think this can be looked at as from zero to max anyway, if it were about a drum hit)




So it could be a conclusion that the faster we make the recording (or improve on it at playback) the less punch there will be. The means seems less important (but following compression could be one of it), but less smear = less punch looks obvious. And the other way around of course.




Btw, I fully agree that compression does a lot more than taking away dynamic range only, but it seems (worthy) another subject (rather new).




Regards and thanks,



Link to comment

Hi Peter,




The Zep track is an interesting example. I mastered the first CD release for all the original Zep albums (with the exception of "IV" which had already been done by the time Atlantic started doing CD masters).




Though many fans still consider these to be the best of the CD releases of those albums, how much better they might have sounded if two things were true:


1. Atlantic gave me original mixes to work with instead of copies. (Without the originals, I could not even vouch for the accuracy of the "flat" copies.)


2. More importantly, if the original mixes had not been compressed like they were.




While some may like the sound of compression (I would not argue with what makes folks happy in their listening), in my experience, there is no way in the Universe compressing drums or any other instrument adds punch. Invariably in recordings, particularly those that utilize compression, the punch (that "in your chest" feeling one gets in the presence of a good rock drummer -in the presence of a good rock band for that matter) is lost.




My label just released a recording ("Confluence" by Jason Vitelli) that contains a few all out electric rock songs. These, like the rest, were done with no compression whatsoever and to my ears, as well as those of the players and the folks who have commented so far, there is a depth and indeed a "slam" that is lacking on other rock recordings I've heard.




My personal perspective is that it would be illogical to expect to increase the amount of slam by curtailing the dynamic swings. I very much prefer the dynamics unfettered and have been wanting to record rock like this for years.




Best regards,







Link to comment

What a great read! I dont pretend to know much about audio or what goes on behind the scenes so it is really nice to have a reference like this to understand why I am hearing what I hear in modern recordings. There seem to be a lot of schools of thought in terms of what constitutes a good recording. I have heard some things that don't have a counterpart in real life and are very interesting soundscapes. I wouldn't call them realistic in the least. Some of these effects can be very interesting. I was comparing two very different albums that sound similar to me in an odd way. They were both released this year. The first Gillian Welch's new one, sounds about as analog as analog gets in the digital realm. Fine recording where all the instruments sound like one would expect them to, with lots of space for breathing room around all the instruments. The other Opeth's Heritage, couldn't be further from that. It is an extremely well made album (especially the DVD version) but I wouldn't say it sounds 'realistic' and in fact probably a whole arsenal of studio wizardry went into it. However, the dynamic range remains intact, extremely so actually. It certainly doesn't sound real in the sense that Welch's album sounds but it certainly sounds good in the same way that her album sounds if that makes sense. Two completely different takes on sound both done extremely well. I think there is something to be said for both approaches, ie using no limiters, compressors, ect or at least the judicious use, and the other that piles on effects without an inherent loss in musical quality or dynamics. I hope I didn't get too far off subject. I just think there are times when one approach seems appropriate and there are times when lots of studio tricks can sound great. These two albums are the extreme means of the same end.

Link to comment

Hey Peter,




For sure, suppressing drum peaks reduces the dynamic range and the transient impact of the kit. No question. However, there is a well-known trick in the recording industry to “fool” the listener into believing that the drums are “punchier” using a compressor/limiter. It is a psychoacoustic effect




Here is how it works, you set the compressor‘s attack and release times to very fast to purposely take out the leading edge of the transient impact and releases very quickly (in milliseconds) to raise the overall sound level as the drum sound decays. It is the simple math approach (remember calculus and integrals: http://hyperphysics.phy-astr.gsu.edu/hbase/integ.html ) of increasing the area under the curve, which increases the overall “loudness” of the drum. This is the classic “loudness war” but at the individual instrument level.




The psychoacoustic effect is that it “sounds like” the drum is “rounder”, “fatter”, “bigger” and the “punch” is actually the sound of the compressor trying to squash the transient impact. In other words, because drum transients are so fast, the compressor takes time to catch the impact and with the fast release, “it sounds like” it is punchier. In a lot of ways, it is a cheap parlor trick.




Remember, all of this occurs in milliseconds and takes a lot of practice with a compressor/limiter to get it right. Whether right or wrong, this happens all the time. Personally, like I mentioned in my response, I would drop all use of compression and preserve the natural dynamic range of the instruments, especially with DAW’s and their wide dynamic range capabilites.




Have you ever heard The Sheffield Track/Drum Record? http://www.sheffieldlab.com/sheffield.pl?detail=SL10081 The drum solos are worth listening to as a reference to a “live” drum sound. Most people from the rock world are surprised to hear what a “real” live uncompressed drum kit sounds like.




Hope that helps!





Link to comment

Once again excellent post as expected....




I look forward to seeing more of your posts.




Many Thanks!





Link to comment

Thanks, lots of great inforamtion here. BTW, Mitch, I believe Peter is a drummer, so he is is quite attuned to the sound of drums...


Great info on using compression on drums, my understanding is that what you describe would stretch the peak in time, resulting in "more" (area under the curve) sound, hence more punch-but I suspect also a change to the timbre of th drum.


Peter, thanks for the correlation in subjective listening to jitter effects with time distortions of compression. Very enlightening: with lower jitter I always seem to hear better distinction in timbre between different drums and how they are hit. Now I see how the use of compression in certain ways can have similar effects, in addition to its nominal effect on dynamic range...

Link to comment

To my ears, the idea of compression adding "punch" has much in common with the idea of adding harmonic distortion to give the impression of greater loudness.




Both are artifacts of not-so-great systems, pushed beyond their capabilities. Playing a recording at high volume on such a system will engender these affects. So, I can see that to some folks distorting the program by either adding spurious harmonics or by curtailing dynamics could fool them into the reverse, i.e., thinking the program is louder... until they hear the program when it actually is louder.




Compression "adds punch" in the same exact way increasing distortion "adds volume". That is, both create affects. And these affects are clearly revealed for what they are when compared with a real increase in dynamics and/or a real increase in volume.




Put another way, try comparing a compressed recording of drums with an uncompressed version of the same recording and see if the compressed version sounds more or less like being in the presence of the drummer. See if the compressed version kicks more or less than the one without compression.




I had an interesting experience in this regard in the mid-80's. When I was at Atlantic, Phil Collins' first two solo albums had come out on CD and were doing very well. One day, I found copies of the original, unmastered mixes in the tape library and out of curiosity, gave a listen.




Lots of folks know the big drum break in "In The Air Tonight". Well on the original mixes (i.e. before the original mastering), those drums really broke out of the speakers (and nearly broke the speakers ;-}). I proceeded to create a CD master of my own (for both of the first two solo albums), with of course, no compression. The next day, I compared these with the CDs for studio management, who responded by immediately sending copies of what I'd done to Phil. He declared the difference "night and day" and the masters I created were put into production.




All that said, while I know folks who like the sound of compression, more often than not, I personally find it harms a recording much more than it helps it.




Just my perspective, of course.




Best regards,







Link to comment

Peter, Barry, Caner, bleedink, barrows, thanks for your kind words and contributions.




Barry, I get your passion, that comes through loud and clear :-) Love your uncomprimised approach to stereo recordings. I had a listen to the Equinox recordings and took a look at the session photos. Nice stereo sound. Looks like the guitar player is using a compressor pedal: http://www.soundkeeperrecordings.com/confluence.htm#photos




Thanks for the mastering job on Phil Collins, "In the Air Tonight". But as you know, the drums are heavily compressed and noise gated on the multi-track tape recording. This Mix article has the story of how Phil's drums (along with everything else) were recorded and mixed by Hugh Padgham:




"The famous drum fill, Collins contends, could have been anything. What is on the record is what came out at the moment. “When people talk about the ‘Phil Collins drum sound,’ that is actually a huge variety of drum sounds,” Collins says. “We never left the setup; we always broke it down and started again so we could end up somewhere different. The Townhouse Studio actually wasn't that live. It was quite tall, but not really a big room — probably smaller than most people's bedroom. The Genesis studio we designed had a much livelier, bigger room, glass and reflective surface. So when you listen to “In the Air Tonight,” it is not really that live, it's big. The snare drum and tom toms kind of bark, but it is made from a lot of compression with ambient mics as far away from the drums as possible, and those are noise-gated.”




"The drums were recorded with two Neumann U87s compressed with UREI 1176s, 12 or 15 feet from the drums, the reverse talkback room mic [called a Ball and Biscuit, made by a company called STC] with the heavy-duty compressor and a U47 close on the bass drum and a Shure SM57 close on the snare drum. “But 90 percent of the sound was the live room mics and then I'd add a touch of close bass drum and close snare drum to give it a little more snap,” Padgham says."








barrows, thanks for pointing out that Peter is a drummer, awesome! With respect to your understanding, yes and yes it does change the timbre. However, if done judiciously, “most” people can’t tell. e.g. look at how Phil's drums were recorded above and who does not love that huge sound. If you regularly listen to rock/pop/blues/jazz/folk (on multi-track), you are listening to compressed instruments, vocals, mixes, masters, remasters “most” of the time. In other words, unless you regularly go and see live music of the same type (likely compressed if amplified), our ears have been trained for decades listening to compression.




I love music of any kind and if it happens to be really well recorded, mixed, and mastered, and put directly into my hands as a hi res copy, with nothing done to it, other than meticulously transferred with the full dr from the master intact, that's a huge bonus to me. As Tom Petty's producer/engineer note says, "We want you to hear it at home the same way we hear it in the studio." I am willing to pay good money for that.




bleedink, yes, compressors and other effects, like from my vintage effects rack below, can be used in many different ways. Always with the intent to represent the artist/band/mix in the best possible light or the artist/band/producer asking for a particular sound like from the classic Roland Dimension D http://www.uaudio.com/store/delay-modulation/roland-dimensiond.html in the rack used a lot on guitars from yesteryear.








Now all of these hardware/firmware solutions are now available as software plugins into DAW's for a fraction of the cost.




In the awesome sounding, “It Might Get Loud” http://sonyclassics.com/itmightgetloud/IMGL_presskit.pdf on Blu ray, you see Jimmy Page calling The Edge from U2 a sonic architect. For good reason, as the Edge has a dedicated set of effects , one per song! Anyone that likes guitars would likely enjoy this Blu ray recording.




Here is another example of trying to represent the artist in the best possible light (ah, sound) by using sound tools at my disposal. I once recorded a wonderful artist that could play a 12 string acoustic like no other I have heard. The issue was that he played quietly and too evenly, at least according to the producer. No matter what we did with, pics, mics (different types and positioning), riding the fader, etc., the sound did not convey his emotional playing enough.




Enter in the compressor again. With a small ratio of 2:1, a slow attack and medium release, the compressor would let the pic or finger picking through (i.e. the pluck of the string) and then ever so gently compress the note and raise the decay, just a wee bit. The result was bringing that wonderful dynamic plucking sound to the forefront with a little bit of sustain that the Producer wanted.




When the artist heard his acoustic sound, he thought it was a miracle and kept thanking us. With renewed vigor and confidence in his sound, we got him right back into the studio and he knocked off one tune after the other thoroughly enjoying himself and that feeling came through recorded on tape.




Just saying that compression and processing isn’t all bad and is widely used in the recording industry as a standard for decades: Urei 1176 peak limiter gets into the audio technology hall of fame: http://mixonline.com/mag/audio_tecnology_hall_fame_4/




Cheers, Mitch

Link to comment

Totally agreed that all these approaches are just tools to an end. Also understand that most recordings of rock/pop will include compression, on a track by track basis as a means to achieve a certain sound.


I do have pretty good references for what live sounds of most instruments are, thanks to lots of concerts, small, large, amplified, and not, and having some very good friends in college who were/are musicians (prog/rock/jazz).


Honestly though, I do not think anyone I know would even for a minute think that the Gabriel/Collins drum sound made famous in the 80s was the way an actual drum sounded without any processing...


In any case, as we understand, compression is a dirty word amongst audiophiles these days, because of its overuse as well documented over the last ten to fifteen years or so, what everyone now refers to as "loudness". The good news is that the pendulum has begun its slow swing back, with some musicians, and some producers, demanding more dynamic range...

Link to comment

Hi again Mitch an others,




Ok, careful now, because I sure don't want to degrade this beautiful blog into some stupid yes/no discussion/estimating/not-knowing and it certainly shouldn't be about Audacity etc. graphs. Still it may look like that just me starts such a down-detailed thing. However, I do certainly not and it will hopefully merely be about the professional judgement I would like to challenge from all of you. I try to count myself in.




For Mitch : I develop playback software which should be the best there is for SQ; same with a D/A converter. "the best" may always be under discussion, but it certainly is the explicit aim, cost no object.


Both are subject to exactly what is talked about in this blog, because it is do or die because of these things. It is, for example, die for even the smallest smell of "remaster", which will be destroyed immediately. This is no exaggeration. And please people, this is only to indicate my "listening" and system perhaps, and not to come up with piles of "good remasters indeed". We do that elsewhere.




After this introduction, here is what I wanted to put forward : That Drum and Track Record album does not sound at all.




Yea, well, I thought to emphasize it, to make the distance even further from what is and should not be. That challenge.


Now, next we should be able to talk about how it happened. Not how it happens that it sounds so poor in my system, but what will have happened to the recording(s) so that it sounds like it does. Of course this needs the description from my side, which I will present below.




Before I forget, theoretically (!) this is indeed an album with a fair amount of dynamic range. But, I only derive(d) this from it being some 10dB softer than my (set !) average. Notice that this -10dB is for the whole album, while half of it is about drumming (that encouraging for less average SPL output). I forgot to look a the maximum used in the album, and I can't reach that data from where I am at this moment. With this I want to say : when the maximum used is at -10dBFS this says nothing for *used* DR. So this is a disclaimer for further real merits (and possibly my judgement, may related stuff popup hereunder).




A few things for your reference :


- Most albums sound the best in my system (one better than the other of course);


- This album can be called Rock;


- My system follows transients 1:1 within the audio band easily (genuine impulse response for a frequency beyond the audio band (IIRC I at least measured it for 29KHz amps-out).




To keep in mind the little subject here :


This album is advised to listen to for its good dynamic range, plus it was implied, I think, that this would be a better one. I judge it as a worse, and I only listened to it throughout because I liked to judge it and report back (after hearing a first 10 seconds). Would it have been ok - no big deal. But since it is not, why ?


Why, and why someone with the real knowledge thinks the opposite.




So, what do I perceive from it ?




First off, it is completely dead. There's nothing alive in it. A robot.




Next, I do not perceive a high DR at all. So, it should be, but it looks like not to be at all. Of course I should have looked into that data, but I rather be free of that at first, because it makes me (or us) think better.


I must notice (again) that the 10dB being softer (could be 8, could be 12, I didn't watch close) counts for the whole album on avegare, while half of it is about drumming (last two tracks). To me it looks like it could be that the engineer forgot to pump up the levels, and the drumming never reaches close to max. So, the whole album is tuned into the first more "heavy" tracks, and now the drumming doesn't work out (but see towards the end; I virtually did the same and *that* just works).




Down to the real merits of what I hear, it is distortion all over. Each hit on a tom, snare and even base drum carries the exact same distortion. It makes the drum sticks stick to the veils. This is easy to understand when you "saw" something like Yamato using long flexible bended sticks. See this video [video=youtube_share;Dark883eH3s]

at 5:40 into it, and the guy in the middle with the red long sticks. These sticks won't imply a short attack, but a long one because the hit itself takes long (the sticks bend into flat onto the veil).




If I may call it like that, the release sound is always the same. Think like a too loose snare which sings along on every hit onto something else. But it is not that, unless this was so and is EQd out or something.




It is way way too dry (part of the being dead I think), and therefore there is no realism at all. I tried to imagine a dead room, but it is articial IMO.




It is the worse representation of toms etc. sounding different amongst eachother. Cymbals same story. This, while a great deal of the track(s) seems to be about explicit showoff of just that (differentiation).




I own more Sheffield recordings and think they are just fine up to great; Right after this I played The Sheffield Pop Experience, and I found no reason to turn that off (the opposite). Interestingly enough that is some 7dB louder than (my) average. That's some 17dB difference between the two albums, which is huge.




For analysis (well, if you want to do it anyway), this distortion I talked about is key;


To my own experience this must be something about harmonic distortion like from aliasing back or something. It is very similar to what happens with my own recording (I referred to earlier) and what will only show with the cowbell; give it false in-band harmonics by not enough upsampling and suddenly the cowbell starts to show some after-echo like a (very) short rattle. This is the difference between 352.8 and 176.4 in my case and the only exhibit that the 176.4 playback is not live. And thus, with that cowbell only.


With the Sheffield album the cowbell has the exact same exhibit, but, the 100% same is in all hits (not cymbals).


I envision that you can concentrate on the cowbell, just because I can let that go wrong too in my own recording. If you don't hear it, no worries, because it will be because of the general smear and the cowbell will not exhibit its very high transient (at lower frequency this time !) in the first place. This by itself won't allow harmonic distortion coming from just that (the high transient). -> Hey, sounds better !




One last less important thing (although we talked about it of course) :


- The Sheffield did not show a single bit of punch. Just nothing;


- I received the remark the other day from some good ears that the difference with playing the drums live, and my recording of it, was the measure of punch (feel in the stomache).




While this latter only occurred last weekend, I didn't pay much attention to it (never noticed it myself either), until of course this blog. So, most 100% probably this *will* be one of the dimensions of pass vs fail, were it for reality.


Of course I try to play back at realistic levels, which will be 110dB for some firm rock. So it shouldn't be about that. Most probably not enough juice in the amps.




May it be of help or general interest, my only three-attempt recording (setting the levels) was very much on the safe side, the general level not even consuming half of the headroom, peaks at some 80% of it (the graphs of it are in the link I gave earlier). In this case, the level is 16dB (!) softer than my set average.


And as said earlier, just 16/44.1.




Alright, before the question got totally lost in my too long post :


What could have happened to this Sheffield recording which should be a good one, but is not once it is played back on a system which should show it as a good recording ?







Link to comment

Hi Mitch,




The photo link points to the "Confluence" sessions. Here's a link to the "Equinox" session photos.




In all cases, the musicians are free make their instruments sound as they want them to sound in performance. I strongly doubt there is a compressor pedal in either case, based simply on the dynamics of the guitar solos. (On "Equinox", listen to the complete "Seremoni Tiga" and see if you think the guitar is using a compressor pedal. On the "Confluence" sessions, I had the bass player, who usually does use a compressor pedal, omit it for these sessions.)




With regard to Phil Collin's drums, I'm aware that Hugh Padgham compressed them in the recording and likely, in the mixing too. What I was showing was just how much better and more alive they sounded by removing the compression from the mastering. To be clear, I believe the drums could have sounded much better still, had the previous stages of compression been avoided too.




I know many folks, including many colleagues, who like what compression does for recordings. So, I can understand and respect if you feel similarly. My only point is to express a perspective that seeks the sound of the performance rather than the sound of a recording - two equally valid but very different approaches.




Best regards,







Link to comment

there needs to be a distinction made here regarding the difference between a musical piece which is created in the recording studio, using the many tools available in the studio (and as popularised in the 60s by the Beatles and Beach Boys), and a recording of a "live" performance.


Clearly Barry's approach with his Soundkeeper lable is to record a live performance, being as true to sound of the live performance as is possible. On the other, there are many studio "creations" which bare little resemblance to live performance.


To me, both approaches are equally valid, and both can achieve a significant artistic expression, they are just different things. I am huge fan of the studio creations of peter gabriel, and find they are very moving, powerful, expressions of his artistic intent. Just because peter gabriel takes full advantage of the technology available to him, does not mean that the result is not musical (but certainly, if the technology is used just for technology's sake, rather than in support of genuine artistic/musical expression, then the result can be nothing of value). There is plenty of empty, vacuous, "music" created in the studio, but studio creations do not have to be this way. Just as there are plenty of recordings of live musical performances which do not "make it" either. What really matters is first, does the musician have something relevant to say along with the chops to say it, and secondly, how well does the use/application of the recording technology support the communication.

Link to comment

Hi barrows,




Agreed 100%. As I mentioned in my last post, "My only point is to express a perspective that seeks the sound of the performance rather than the sound of a recording - two equally valid but very different approaches."




"Sgt. Pepper" could not have been made any way other than how it was made. And it has a magic all its own.




Agreed too that the real key is what the musician has to say. We've all heard plenty of recordings that accurately captured uninspired performances.




Dynamic range (vs. compression) is a subject very close to my heart. Even with studio creations that are not necessarily trying to emulate a real performance, much less capture one, I find that when aspects of the recording are more truly convincing, it becomes much easier to "suspend disbelief". And that, I find, is the key to becoming immersed in any work of art, be it a music recording or a motion picture.




Best regards,







Link to comment

Hey guys, I really appreciate your point of views.




The purpose of my blog post was to point out to audiophile listener’s that compression is used on the vast majority of recorded music we listen to. However, I did not offer a judgment if that was a good or bad thing. I think there may be an implicit assumption that I seem to support using compression. That is not the case. In fact, I did state that if I had my way, I would not use compression at all, at least a couple of times :-)




I also stated that I was surprised to see the industry trend over the years towards more compression, where I thought it would have gone the other way, just like how we have higher resolution video. As barrows points out, and what folks like Barry are doing, the pendulum is swinging the other way to less compression. My viewpoint on remastering to hi-res would be give me the closest copy you can make of the original master and put away the compressors, d'esser's, noise gates/reduction, just give me the best historical print.




barrows, just saw your post as I was posting this. Total agreement. I love live band sound recordings. Unfortunately, there are no too many of those around that also have audiophile sound quality. Barry's appoach being the few exceptions.




Barry, respect. Sound of the performance first. When I had the chance to use my Sony PCM-F1 digital recorder with a stereo pair, I felt like this is what real stereo recordings should be. At one time, I was trying to pursue binaural recordings using the Neumann dummy head http://en.wikipedia.org/wiki/Binaural_recording so you were "in the audience". It was pretty incredible, but at that time, listening on headphones, being the limitation of binaural recordings, was not popular. However, today, look at the proliferation of headphones. Maybe I missed an opportunity...




barrows, for sure, with respect to the famous Gabriel/Collins drum sound. But when “In the Air Tonight” recording was released, every drummer that came into my studio wanted their drums to sound like that. I had guys sampling the sound right off the disc and using http://simmonsdrums.net/ to trigger the samples to get that sound. Crazy!




I wish the folks at HDTracks and other purveyors of high resolution recordings would spend a bit more time indicating exactly where the source came from and what exactly was done to the source to produce the hi-res version. Similar to what Tom Pettty did with his "producer engineer note." I would have also expected a blog post on each of the transfers from the mastering engineer(s) going into the details of the process. Instead, there is very little to nothing. Even from a business perspective, HDTracks and others would get a lot more of my money if I knew these things.




Peter, I made no judgment as to what I thought of the sound quality of the Sheffield Drum/Track record. I said that it is a “live” drum recording, with no compression, minimalist mic technique, no processing or artificial reverb. I was contrasting the sound of a live drum recording to a processed sound like what is on “In the Air Tonight”




Listening to Ron Tutt and Jim Keltner, (with a measured DR of 22 and 19 respectively), there is a telltale sign of something wrong. Both recordings are clipped :-( at least according to the measurements using the dynamic range meter. I have the original CD transfer. Clipping does not help transients.




Just like you describe, others also feel that the original transfer to CD sounds totally dead: http://www.amazon.com/Sheffield-Drum-Track-Disc-Master/product-reviews/B003JG93DC/ref=dp_top_cm_cr_acr_txt?ie=UTF8&showViewpoints=1 Apparently, Bruce Brown's transfer to DXD sounds better. I have not heard it, so cannot comment.




Peter, I would be interested in hearing your thoughts with respect to what you think is a good drum sound. Perhaps you can point me to some sources?









Link to comment

"barrows, for sure, with respect to the famous Gabriel/Collins drum sound. But when “In the Air Tonight” recording was released, every drummer that came into my studio wanted their drums to sound like that. I had guys sampling the sound right off the disc and using http://simmonsdrums.net/ to trigger the samples to get that sound. Crazy!"




LOL! Yeah, because this drum sound will be appropriate (not) for every piece of music... I guess that is what a little bit of success can breed.

Link to comment

Peter, I would be interested in hearing your thoughts with respect to what you think is a good drum sound. Perhaps you can point me to some sources?




That would be difficult for me Mitch. Most just is, but it so much depends on the playback quality. What counted last month, is overruled today. I could divert you to fairly random Norwegian recordings, but although mostly superbly recorded, it will be Jazz and more easy to record. So, assuming you are more in the leage of "Cozy Powell" drums (I mean more commercial stuff), one which always jumped out of it is this album from Gruppo Sportivo and then the first 5 tracks or so (after that they become too "new" or something) :








This is not about the cymbals, put purely the drums. How honest they sound and how each tom can bee recognized.


I have hundreds of more examples, but they will only work in my system. This one will show it on anyone's, I think.




But about random others - which I can only take from my latest listenings, take the track Tusk from Tusk (mor Cosy Powell like :-). However, you probably won't get there easily, because of this : http://www.computeraudiophile.com/content/Tusk-HD-Tracks#comment-110883 and the further contents of that thread.




Also from last night's playback would be this one :






which can be considered "big band" like. Here too it is about the honesty, and further the for me superb rendition of the copper in there (man, what a snap).




But to please keep in mind : although these would be my examples from only yesterday's playback, they wouldn't have been examples a few years back. Again, it all depends on the playback chain, and mine changes weekly.


Btw, I am not telling you to get another DAC or something - just that comparison and real judgement is not easy at all.




Here is one which can be considered explicit test material :








This is merely percussion in general, but superbly recorded and with wild transients because of that (like the smallest bells, rimshots, clapping on to ...). However, it will completely fail just because of this on anything which smears (which wille most probably be your "OS" DAC).




Generally : try anything from Turtle Records (superb transients).


And a controversy : anything from Maple Shade is strange. Odd. BUT, they exhibit the best base drums - somehow.







Link to comment

  • Create New...