ABBA IN THE STUDIO
As told by Michael Tretow
The sound of ABBA can basically be described as a mixture of the Phil Spector sound and modern technology - it is definitely a big sound. In concept they are a recording group who are quite happy to work in their home territory especially since they built their own studio, Polar.
This article has been written to describe the techniques I use to get the sound. The typical set-up for the basic tracks is as follows: drums, bass, piano or Polymoog, electric guitar, two or more ambience mics and one or two mics for the cue vocals.
The drums are nearly always recorded in the drum booth and very seldom out in the studio. I much prefer the relative dryness of a drum cage to the uncontrolled leakage which usually occurs in the studio. I try to compensate for the dead drum-sound by placing the ambience mics very carefully, the idea being to get a good overall picture of what's going on in the studio, even without using the closeup mics. (That is also the reason for using an amp for the bass guitar, although the amp is not miked up. The bass is recorded by direct injection into the desk, with a few dB of limiting.)
For drums I usually prefer to use AKG 414s on the tom-toms, RE-20s on the bass drum and under the snare, CK1 for the top snare, and whatever for the overall kit - I'm constantly trying out new mics for the overall kit, so I take whatever I find in the studio and try them. Sometimes I'm lucky, sometimes I'm not.
In most cases I don't use any mics on the hi-hat - I always seem to get enough, or even too much, hi-hat anyway. By the way, must be time for someone to say something about the annoying brass plates that every drummer carries around! I do not regard the cymbal as a musical instrument. I regard it as a tool with which the drummer can ruin a good drum sound in the overall kit mics. It's about time somebody put a stop to them. Put a tax on cymbals, that would do the trick! (See also Peter Gabriel's view on 'No-cymbals' in the SI Dec '79 interviews - Ed.)
So the best solution I've found is to place the overalls underall, down at the floor some 7½ ft apart. It sounds like a weird idea, and when I read about this technique in a US magazine I thought it must be totally wrong to do it that way, so I tried it - and it really works!
I never use limiting or compression for the drums on to the tracks, except for good old tape saturation.
I always try to place the electric guitar amp in a different room, if the studio has access to a storage room or something like that. I believe that to get a really loud sound you must play loud and literally let the sound fill the room. I use one close-up mic in front of the amp and one omni, out in the room, to pick up rattling windows and the like. You can control the amount of leakage into the ambience mics by closing or opening the door. However, if the studio isn't equipped with a separate room for the guitar we normally re-do the guitar parts out in the studio when the session is over. The room mic will almost certainly have a boomy character, so it is given rather heavy eq. Sometimes I also use the MXR Flanger on this mic to get a 'singing' sound, but the amplifier mic is always cut flat. These will usually be the U47 fets or maybe some good dynamic for the amp and a condenser for the room.
The grand piano is cut with three 414s in the sound holes. The middle mic is limited and fed through the MXR Flanger to the middle of the stereo pattern on the two piano tracks. One useful feature of the MXR, in fact, is the possibility to feed the signal through the bucket brigades only, without summing the direct sound to it. If the signal that's going via the bucket brigade has no phase relationship to the rest of the sound when you finally sum it, you will just get a pitch variation with no phasing. This is the reason for using a separate mic to feed the MXR.
And that is what makes it possible for the engineer to turn a $20,000 grand piano into a $100 honky-tonk upright. Amazing how you can make technology work for you!
By the way, you might be interested in how to make a 25-piece string section sound like a 25-stringed banjo. Well, the secret is to run the violins through a Kepex and key them with the hi-hat. Not a very useful feature, I'm afraid, but it's yours for free! I mean, if it worked the other way round - turning a 25-stringed banjo into a 25-piece string ensemble - it could have been very useful.
The Polymoog creates very special problems. The signal-to-noise ratio in the instrument is probably around 3dB or so, so I used to feed it through a Dolby 361 unit in the decode position. If you're careful with the level this will not affect the sound but you have to level out the dynamics, if there are any, and then compensate for the lost dynamics in the mix. Anyway, it's a good method of cleaning up noisy tracks. The Polymoog's got its own balanced line output so I use that, sometimes with a miked amp as well.
On one song (Arrival) the Moog was picked up by two ambience mics only. To get a 'natural' sound, as if it were a bunch of real instruments playing out in the studio, I moved the amp in the room for every overdub we made, and recorded each harmony in stereo on two tracks. If you listen to the record it's very hard to tell what instrument it is; it sounds like all-metal bagpipes - or something.
As I said earlier, I try to put the ambience mics where I can get a good sound on all the instruments. Naturally this is impossible. Usually you lose all the low end in the drums and all of the highs in the piano. However, these tracks are only used to beef up and bring a little 'air' to the recording.
A few years back I tried desperately to get a very dry and tight sound on my recordings. But nowadays as studios get tighter and tighter I try desperately to get further and further away from the dead sound. I believe it's all a question of how much control you have, and back in the old days you didn't want any leakage purely because you couldn't handle it. Now you can.
When we've recorded the final take, I usually try to convince the musicians to do an overdub of all the instruments and record them with the two ambience mics only. When doing so, I change the speed of the 24-track some 0.5% to get a very subtle pitch deviation. This broadens the sound and makes it really sound doubled.
Recording the basic tracks will take approximately one full day's work per song - that means 10 to 12 hours for just the drums, bass, piano and guitar.
Björn Ulvaeus and Benny Andersson are both very conscientious people and true pros - they never stop until they've found what they're looking for. They try every song in different tempos, moods and instrumentations until they get what they want. This of course leaves a lot of time for the engineer to try out new sounds - you have time to correct mic placements and time to optimise each track. No doubt every engineer is aware of the sad fact that most of the time the situation is quite the opposite. The vocalists are usually permitted to sing the songs hundreds of times, but when did you hear of an engineer who was allowed 40 takes in order to get the Kepexes right?
Either Björn, Frida Lyngstad or Agnetha Ulvaeus will sing a guide track with phoney lyrics like: 'I love you/Please love me too/Don't be untrue/Say "I do"/Cause I lo-o-o-o-ve you so.' Then...
Next day we start with the instrument overdubs; Björn and Benny are trying to fill up the tracks as fast as the machine will go. That means that there are two empty tracks left at lunchtime, and it's time to reduce the 16 Moog tracks to make room for the vocal track that the record-buyers so urgently demand. No-one's ever interested in leaving tracks open for the vocals, so it usually falls on me to erase one of my 'unnecessary' drum tracks to put the vocals on. Since I'm aware of this, I always put the hi-hat on a separate track, even though I never intend to use it. But it certainly gives me an unselfish image, to 'give up' one of my drum tracks; the effect on the producer is always most pleasing to watch.
If there are going to be acoustic guitars on the song, we usually re-do the guitars afterwards. I always try to record them on a hard surface, eg with the carpet rolled off. The U87 always seems to work fine on acoustic guitar, so that's mostly my choice; again I always try an AKG C34 stereo mic at some distance to get the spaciousness, together with close-ups for the 'body'.
Other overdubs which we do often involve a lot of glockenspiels, for which I prefer the old RCA DX66. It's a ribbon mic that's virtually impossible to overload with high energy treble instruments (like the glockenspiel). It is also my favourite trumpet mic for the same reason.
The instrument overdubs will almost certainly take up another full day, so the next day it's time for the vocals. All background harmonies and choir parts are sung by the group and are nearly always double tracked. For the vocals I use the C34 again, with Frida and Agnetha on each side, facing each other, with quite a lot of dbx 160 limiting on each capsule. By using a stereo mic you don't get the annoying phase cancellations you get with separate mics. Since Frida and Agnetha have to stand very close to each other for timing, you can't have them far enough apart to avoid leakage.
The separation between the capsules in the C34 is excellent and, in fact, you even get a better back-to-back rejection since the leakage does not become so obvious in character alterations.
For Benny and Björn, I use a U47 if they're doing tight harmonies. If it is more of a 'choir' character we're after we usually record out in the big isolation room at Polar Studio with just one stereo mic hanging from the ceiling.
We do a lot of ping-ponging with the voices to save tracks for the Moogs. (Every song can be improved by adding a few Moogs!)
Another day will pass before we're through with all the vocals, and the next day we start to mix. I would like to point out that although this seems to proceed very slowly, it is not because of any lack of energy. This whole time-consuming business is created because everything - I mean everything - is tried out. The group can arrange and record full four-part harmonies, double track them and, if it doesn't come out right, erase them and start over again! Move On, one of the songs on The Album LP, had four completely different sets of lyrics, with harmonies and all, before the final one was recorded.
The mixdown session usually starts with the boys leaving the control room and with me setting up a basic mix. After a while they can't keep away any more and come into the control room saying, 'That sounds great! Let's take it exactly like it is.' And then slowly they start to change everything.
Since the last album, Voulez-Vous, we've done all of our recordings at Polar Studio. That is a relief after all those years of carrying around back-breaking piles of 2in tapes. We used to work a lot at Metronome Studio with its superb Neve 24/8 console, but Leif Mases at Polar Studio has done a great job on the Harrison console there, so it almost sounds like the good old Neve!
Anyway, this very fine studio is also the proud owner of a pair of Universal Audio LN 176 valve limiters, which are a very important part of the ABBA sound. I use them on the rhythm tracks, drums, bass and percussion, which I assign to two of the outputs. I then compress them and send them back again. This gives a very thumping sound to the rhythm which can be controlled by sending part of the same signals straight to the mix output busses, without any compression.
The tracks on the '24' are normally recorded with as little eq as I can get away with, so that most of the equalising can take place in the mix. In the past it's been a matter of matching the different characteristics of all the studios in which we've been working, in order to make it sound as if all the tracks were recorded in the same studio. That's one reason (and just one) why I've given up Dolby, and recorded at 30in/s instead. Every studio seems to have their own Dolby level and due to all the track-bouncing I have to do, I've sometimes ended up with tracks which are impossible to decode. If, for instance, you decode five vocal tracks a few dB out and put them on to another track, there's no way to get it right afterwards.
I never limit or compress the total - it never works - so naturally I always try before I give up.
After 10 hours people don't know what they're hearing any more. 'I think the bass is too loud, or, wait a minute, no, it's too soft,' or 'Let's have some echo on the snare!'
'It's already there.'
'Are you sure?'
So sometimes we go back and remix, but not very often. Thus, an average ABBA song will take some four days to complete - or even five, if you include the cutting.
I always cut my own recordings and since I used to work at Metronome some years back, I'm very familiar with their system. They've got the Westrex 3 D II A cutter head and Westrex amps to drive it, and to my ear, nothing can match the Westrex. It's the last chance you've got to correct your miss-takes, and to even out the differences between the mixes you've made in the middle of the night by taking off some 15dB of treble, and the mixes you've done in the morning after a full night's sleep.
And then - take a listen!
For some more interesting reading go here: http://www.abbajustlikethat.comyr.com/1_4_Audiophile-Corner.html