Compression vs. Compression? Part I
Compression vs. Compression? Part I
The issue of compression can be confusing. We certainly saw that in the Harman sponsored film “The Distortion of Sound”, which seem to dwell on the ills of MP3 data compression as a primary culprit in the inability of artists to get the “soul” of their music to their fans and then provided a very weak demonstration of audio compression instead. I know the engineers at Harman and JBL are some of the smartest in the business so how did the message in the film hit the wall and burst into flames? I’m still waiting to hear back from some of my friends at Harman…but the best I can say so far is the filmmakers were asked to exaggerate the facts to make their point.
So today, I thought I would provide the facts about audio compression. I’ll follow this up with a post on data compression soon (the lossy kind that was referred to in the 22-minute film).
Audio compression is a pretty simple concept. I usually explain it to my students as an automatic “turner downer” mechanism built into the signal path. If the compression/limiter circuit detects a signal that it knows will cause an overload at some subsequent stage in the signal path, it automatically attenuates (turns down) the amplitude of the signal by a specific ratio or amount. Light compression uses ratios from 2:1 to around 8:1. For ratios of 12:1 or 20:1 it’s referred to as limiting. Seems simple enough, right?
In the days of analog recording on tape, one of the first things a good engineer would do is ask the ensemble, conductor or band to run the loudest section of the piece to be recorded (for a live concert this could be the loudest section of the loudest piece in the program). He or she would then set the microphone preamplifier and/or the recording level of the analog deck so that those peaks would remain safely below the overload level of the meters…usually 3-6 dB below to guarantee that the performance adrenaline wouldn’t kick it up past an acceptable level.
Knowing that the dynamic range of analog tape is around 65-72 dB, pieces with very loud and very soft sections could be problematic. In order to ensure that the loudest sections were safely recorded meant that the softest ones might find themselves in the noise floor of the tape. I hate to admit it but there have been times during a live concert recording when I realized the level that I set was too high and that I was going to overload the tape if I didn’t turn down the record level (which definitely a no-no in the world of concert recording…after all I’m screwing with the dynamics that the conductor or performer is producing). So I would very slowly turn down the recording level during the recording. This is same concept as a compressor, although a compression circuit does this automatically and much more quickly.
To be continued.
It still seems to me that the true evil of MP-3 is simply the large percentage of “data reduction.” To use a simple phrase as did you, the MP-3 process ‘throws away’ 90 odd % of the musical content. Overtones, undertones, ambiance, dynamic range, all the things that put the flesh and blood on and in the musical bones is gone, replaced by built-in, insidious, inevitable, time-based distortions that are the reasons folks commonly say to me, “We love music, but we don’t listen like we used to.” Yes, because far too much mental processing must occur separating out the good from not so good, real musical listening concentration and reaping the full range concomitant rewards become impossible too. Put it this way, we used to complain about dynamic compression years ago, nothing new there. Two other facts; the MP-3 encode/decode does much less harm to small acoustic ensembles; the machine can more easily tell what’s important and what is not. But fuzz, distorto-rock, studio layering in popular music, the very market the process most principally addresses, is in fact the type of material that MP-3 mangles the most!! Last but not least, the analog SN’s you mentioned are in fact not improved upon by the majority of CDs. 96 db of dynamic range in the home doesn’t fly; too quiet or way too loud . The theoretical 144 db of available hi-res dynamic range would have woofers and tweeters whizzing by us in the air as our amplifiers tossed their cookies! Best,Craig
MP3s at very low bitrates sound terrible…but when you move up to 320 kbps, things get much, much better. But there’s no reason to have MP3s at all anymore.
I agree mark; 320 passes muster if done well, though it is yet far short of CD or better. Browsing internet radio, I found Radio Afghanistan at 32kbps!
Mark,
Sorry, but I think that saying the facts were exaggerated is too generous. The 256Kbps downloads are not what is stealing the soul from the music and to imply that it is the root cause is deceptive. If the film concluded that thief is audio compression and not data compression, then the fix is not a simple upgrade of the bitrate but requires remastering of the recordings themselves. That’s big bucks and high risk. No, I think that the film is a marketing tool, not a report. The “solution” is to buy a bigger bucket that we already have waiting for you. Maybe it will fly for a while…
Blaine
They blew it for sure! They started with an agenda and produced the film to support it. Too bad.