Midrange, compression and the history of technology

If you do any music pro­duc­tion, even at the tyro level like me, you real­ize that there’s a char­ac­ter­is­tic “sound” to most pop music: mid scooped and com­pressed. It’s a very odd sound, highly un-natural, and it’s com­pletely ubiq­ui­tous. But it was  not the char­ac­ter­is­tic sound of other eras. In the 20s, music sounded the oppo­site, with a pro­nounced mid-hump. In the 50s, the eq pro­file was “flat.” Why the changes?

The mid scoop is what hap­pens when you boost the highs and lows and cut the mid­dle range. It’s the char­ac­ter­is­tic eq pro­file of most mod­ern music. It’s com­mon in gui­tar tones, it’s com­mon in bass, espe­cially slap-style bass.  I took a bass line from George Porter of the Meters, (Funkify your Life)and played it on my bass with the eq set flat:

bass flat.mp3

Here’s the same line with a pretty typ­i­cal midrange “scoop,” the “smile eq.”

bass smiley.mp3

Midrange is the range of the human voice: it’s where our ears are most sen­si­tive. Most peo­ple will hear the sec­ond bass line as pret­tier, and the first one, the one that’s flat, as harsher sound­ing. It’s espe­cially pro­nounced when you hear it through good speak­ers. Bass play­ers know that it sounds pretty by itself, but in a mix, with a band, it tends to vanish.

Now onto com­pres­sion. Live music typ­i­cally has high vol­ume parts and low vol­ume parts. If you sing, you will typ­i­cally start a note low, bring it up to full, vol­ume and then taper down. All struck or plucked or bowed notes have an “enve­lope:” an ini­tial attack fol­lowed by a slow or fast decay as the note fades. Musi­cians typ­i­cally hit some notes harder than oth­ers: it’s part of being expressive.

Com­pres­sion elim­i­nates those vari­ables. It’s an effect that make the quiet parts loud and brings the loud parts down. Or really, in most music, what it does is take the quiet parts and make them as loud as the loud parts. Com­pres­sion makes the music “punchy,” because instead of com­ing at you with the nor­mal attack/decay pat­tern, the notes start at full vol­ume  and end at full vol­ume. Lit­er­ally all pop music today is mas­sively com­pressed, at every stage. That’s how they make the com­mer­cials louder than the show–they mas­sively com­press them. There’s a very good expla­na­tion on Wikipedia.

A lit­tle loop, bass, drums, gui­tar uncom­pressed and unscooped:

allflat.mp3

And the same loop, mid scooped and compressed

alltreated.mp3

You should hear a pretty dra­matic dif­fer­ence. The sec­ond one  has been com­pressed and mid scooped on every track: it will prob­a­bly sound brighter and louder: the play­ing may sound more pro­fes­sional and less fumbly. To my ears, the first one is way way bet­ter, because there’s more there, but what’s there has more human mis­takes in it.

Com­pres­sion seems to have orig­i­nated with AM radio, way back in the day. Radio broad­cast­ers wanted to make their sta­tions jump out when users twid­dled the dial. They fig­ured out that if they com­pressed the sig­nal, it would sound loud and direct and “in your face.”

Com­pres­sion also makes up for weak play­ing. A really good musi­cian has con­trol of the dynam­ics. A weaker musi­cian, like myself, does not. Com­pres­sion makes the things I play much more even, “tighter,” more uni­form. I’ve often sus­pected com­pres­sion came into use around the same time as rock music, when there were more and more bands who really could not play that well. But now it’s the default taste. Lis­ten to a Lady Gaga song–there are almost no dynam­ics, every­thing comes at you with the same punchy attack, heav­ily, heav­ily com­pressed, not like actual untreated live music at all.

Just to try to make it clear, here’s that same bass part, uncompressed:

uncom­pressed

and com­pressed:

bass com­pressed

All expla­na­tions for the mid scoop depend on psychoa­coustics, and specif­i­cally the fact that our ears tend to “fill in” what isn’t actu­ally there. It’s some­times argued that the taste for mid scoop comes from the rise of earphones/earbuds. Ear­buds have tiny speak­ers and they really can’t repro­duce bass fre­quen­cies very well. Open ear­phones, the kind that don’t seal against your head, are even worse. To make them work,  music pro­duc­ers pumped up the bass and tre­ble, so you heard bright sparkly highs and whompy lows, and your ear sort of filled in the rest. A good set of speak­ers, on the other hand, will give you the full range, and the smi­ley eq can sound fatigu­ing and arti­fi­cial and hollow.

The other expla­na­tion may be the mp3 for­mat. A music track on a cd, a three minute song, will be around 30–40 megabytes. The same song as an mp file will come in at 3–4 megabytes–smaller by a fac­tor of ten. Whoa, dude, where’d all my infor­ma­tion go?

Mp3s work partly by suck­ing out the midrange fre­quen­cies, because  your ear’s ten­dency to will fill in what isn’t actu­ally there. The clas­sic illus­tra­tion of this is a phone call. Your iphone can­not phys­i­cally repro­duce the actual fre­quen­cies of your male friend’s bari­tone voice.  It only actu­ally repro­duces the midrange fre­quen­cies, and your mind/ear “fills in” what’s miss­ing and tell you it’s Bob, with the deep voice.

If you com­pare a good cd track with a typ­i­cal mp3 of the same track, side by side on a decent repro­duc­tion sys­tem, the mp3 will sound “hol­lowed out.”  Try, for exam­ple, a Frank Sina­tra record­ing from the late 50s/early 60s., back when they did not do much fid­dling. But if it’s a mod­ern song, designed to be played on ear­buds, it won’t sound that much different.

In the 20s, all recorded music had a midrange hump–it empha­sized the midrange. That’s because record­ing equip­ment could not cap­ture the extreme fre­quen­cies, either the lows or the highs, very well. Lis­ten, for exam­ple, to a Louis Arm­strong Hot Five or Hot Seven record­ing. By the 1950s, record­ing equip­ment had become extremely sen­si­tive, and the eq pro­file was mostly “flat,” even in all fre­quen­cies, much as you would hear it live. Lis­ten to the clas­sic Miles Davis album Kind of Blue for an example.

In the 1990s, the taste for mid scoop­ing and com­pres­sion became really pro­nounced. On the dance floor, or in a car dri­ving by, peo­ple wanted that whomp­ing bass, and the sparkly top end: a Madonna song like Ray of Light is a good exam­ple. It’s a sonic con­fec­tion: all sugar and salt and no other flavors.

Leo Fender, the great instru­ment and amp maker who founded Fender Musi­cal Instru­ments, often com­pared his sonic goals to lemon­ade. When you drink lemon­ade, he said, you want to taste the sour lemon and the sweet sugar–everything else is “fluff.” He wanted, he said, to get rid of the fluff, and fender’s gui­tars and amps were designed with a scooped eq in mind.

It’s easy to argue this is kind of a juve­nile taste, a junk food taste, a taste for extremes. Is it accu­rate? I’m not sure. It’s also pos­si­ble what what appeals about lemon­ade is the impos­si­bil­ity of two oppo­site tastes, sour and sweet, com­bined: that we like to have nat­ural oppo­si­tions over­come, and so we like a track that is both all bass and all tre­ble for the same rea­son we like foods that com­bine salty and sweet, or dra­mas that depict love and hate in equal measure.

Update: I got a cou­ple com­ments that saw this post as a cri­tique of mod­ern music–it’s really not meant that way. It’s meant as a pos­si­ble account of changes in taste, and a pos­si­ble expla­na­tion of why. I sus­pect lady gaga would be delighted to have her music described as arti­fi­cial and a snack food con­fec­tion, and I like snack foods as much as the next guy.

Update two: Sev­eral peo­ple have sug­gested that mp3 does not reduce mids. This is not what I had read in ear­lier research, but  I’m not an acousti­cian or a com­puter guy–so if any­one has some links or source with bet­ter infor­ma­tion, post away!

Update three: “mp3” can mean dif­fer­ent things–mp3 files can be encoded at dif­fer­ent sam­ples rates. The small­est files have typ­i­cally suf­fered the biggest losses and will sound worse. At higher sam­ple rates, it’s extremely dif­fi­cult to tell mp3 from other for­mats. It’s pos­si­ble that the mid scoop is an arti­fact of the early days of mp3, when more “lossy” tracks were com­mon. Its also pos­si­ble that the pre­v­e­lance of mid scoop comes from the fact that mp3’s some­times strip out the very low and very high fre­quen­cies, the ones we hear less. In that case, the mid scoop would be a way to restore “flatness.”

It is also true that while mp3 was designed to be uncol­ored, many peo­ple hear a dif­fer­ence: Wikipedia’s entry for “mp3” includes “test given to new stu­dents by Stan­ford Uni­ver­sity Music Pro­fes­sor Jonathan Berger showed that stu­dent pref­er­ence for MP3 qual­ity music has risen each year. Berger said the stu­dents seem to pre­fer the ‘siz­zle’ sounds that MP3s bring to music.[44] Oth­ers have reached the same con­clu­sion, and some record pro­duc­ers have begun to mix music specif­i­cally to be heard on iPods and mobile phones.[45]

It’s still the case that Leo Fender, who helped devise the sonic sig­na­ture of post WWII pop music, wanted a scooped midrange sound. He saw it as pure and essen­tial and clean, unclut­tered by “fluff” in the middle.

6 Comments

  • I wore big-ass head­phones walk­ing around my col­lege cam­pus back in the day. Years later I installed an out­ra­geous car stereo when I lived out in the burbs and drove every­where. It was so fancy that I couldn’t lis­ten to the radio any­more because the broad­cast com­pres­sion was too obvi­ous and too fatigu­ing com­pared to CDs.

    Still, I think your MP3 and iPhone-related asser­tions really under­cut your other obser­va­tions in this post. Also, you are a Luddite.

    It con­fuses mat­ters to con­flate the lossy encod­ing of MP3 files with the dynamic range com­pres­sion in stu­dio pro­duc­tion. MP3s, ear­buds and iPhones play no arguable role at all in your indict­ment of mod­ern day pro­duc­tion tech­niques. MP3s sound great, ear­buds are just handy for kids walk­ing to class, and iPhones sound as good as any MP3 player if you jack them into decent speakers.

    There’s no way either of us could tell the dif­fer­ence between some nice MP3s of Bitches Brew and the loss­less FLAC files. Our ears aren’t keen enough and our ref­er­ence mon­i­tors aren’t fancy enough.

    Pro­duc­tion value is as high as it has ever been. Music tar­geted for younger audi­ences might cer­tainly suf­fer to our ears from pop or club mix­ing and mas­ter­ing tech­niques. But just as you wouldn’t buy a bedaz­zled Dora The Explorer back­pack or sneak­ers with flash­ing lights in them you need to let Gaga and Bey­once do their thing. Those tracks are for places we don’t go filled with peo­ple we don’t know. They sound just right for that con­text, and bet­ter than Louis or Miles would sound through a wall of sub boxes. Music is just application-specific like that, always has been.

    Noth­ing sounds as good through ear­buds and most folks don’t care — or at least don’t care enough — I’ll grant you that.

    But maybe you just need some good new music rec­om­men­da­tions. Have you heard Voodoo by D’Angelo? It’s awe­some. And well-produced.

  • Inter­est­ing that you should read it this way–it’s not at ALL intended as some kind of “all mod­ern music is no good” post. This is some­thing I think peo­ple read into it–I like lemon­ade, I like fender amps, I sort of like lady gaga, but not in the same way as I like Miles, and why would I–they are work­ing totally dif­fer­ent terrains.

    I think this is an exam­ple of ageism–I’m older than you, so you assume this is a “watsa matta with kids today” post. It’s just an obser­va­tion about dif­fer­ence and some spec­u­la­tion about what it might mean.

    You might be right about con­flat­ing two things–in the his­tory of tech­nol­ogy this is the argu­ment for treat­ing all spe­cific pieces of tech­nol­ogy as “tech­no­log­i­cal sys­tems,” in the way that light­bulbs require power gen­er­a­tion require mass pro­duced stranded wire require rub­ber insu­la­tion etc etc. I prob­a­bly need to re-examine that part.

    Now get off my lawn!

  • Yeah, we’re on the same page now: it’s only the poor fit of both the tech exam­ples (MP3s, ear­buds, and iPhones) and the music exam­ples (Gaga vs some Blue Note jazz giants) that make you sound like Clint Eastwood.

    There’s no such thing as “music designed to be played on ear­buds”. The novel look at the midrange scoop vs. hump through­out his­tory is the fun and inter­est­ing angle in this post. I’m just say­ing that the anec­do­tal stuff you pulled together to con­tex­tu­al­ize it under­cut the value of that angle.

    It’s hid­den in aca­d­e­mic lan­guage, but the crit­i­cism of Gaga and the exal­ta­tion of Miles and Louis is in there. It’s not some­thing I’m read­ing in. You’re not com­par­ing any of our beloved jazz play­ers to salt water taffy or what­ever rot­ted your teeth out back then. The terms “juve­nile” and “junk food” are pinned on the likes of Madonna, while lan­guage like “extremely sen­si­tive” and “even in all fre­quen­cies, much as you would hear it live” is reserved for Kind Of Blue.

    And yes, I still haven’t for­given you for mis­tak­ing Pom­paloose for being supe­rior to Bey­oncé. I think you need new speak­ers to go with your new D’Angelo album. =P

  • […] get­ting a lot of gen­er­ous and smart com­ments on my ini­tial post about com­pres­sion and mid scoop­ing I decided to take another crack at it. The basic fact, described here, is that modern […]

  • I believe that when Fraun­hof­fer orig­i­nally devel­oped the mp3 codec, they worked under the premise that we are most sen­si­tive to the midrange, least sen­si­tive to tre­ble above 16 Hhz, and deep bass below 60 Hz. Mp3 com­pres­sion does the exact oppo­site of what this arti­cle spec­u­lates, it pre­serves the midrange and cuts tre­ble and bass with fil­ters. The higher the bitrate the higher the low-pass cut­off is (192 kbs usu­ally cut it off at 20 kHz).

    The article’s dis­cus­sion of the mp3 for­mat should be sim­ply deleted for being incorrect.

Leave a Reply

Your email is never shared.Required fields are marked *