Realistic Arrangements: Vocals (Part 2)
Our series continues with the voice, whether synthetic, sampled or the real thing.
Continuing our series on realistic arrangements, here's what to do with the vocals.
Welcome back for this month's episode, and I hope you've all had fun with the drum ideas from last month. This time around, we'll look at ideas on arranging vocal parts, whether played on sampler or synthesiser, or produced by the real thing - the human voice. Luckily enough for H&SR readers, you've had the delights of Michael Cooper's articles on recording vocals, but again we come back to the point that, arrangement wise, vocals have (with a few honourable exceptions) been very much the poor relation in recorded works.
Think back through the history of Rock Music, and remember the vocal greats. The Beach Boys, obviously. Queen? Of course. Who else? The Eagles, Manhattan Transfer, yes. But then the choice thins out. The Flying Pickets? (Hardly - listen to their records and play spot the Emulator 2.) Capability Brown? Gentle Giant?
Now while you're scratching your head, I'll tell you about the last two. They were both British rock bands of the early 70s, Capability Brown featuring six-part harmony vocals (and up to three lead guitars) which on a good night, and live, could put the Beach Boys to shame; Gentle Giant were the band that could and did play anything and everything instrumentally and pioneered the use, in rock, of contrapuntal part writing for vocals. Now before I go any further, let's get the term contrapuntal - from the word counterpoint, in case you've never come across it before - right. We all know that a polyphonic keyboard means that it will play more than one note at a time. Counterpoint, here referring to vocal parts, is used in its classical term meaning more than one independent vocal line. So, you'll readily appreciate that there's a world of difference between four voices singing exactly the same note and rhythmic phrase and four voices singing a four-part chord and the same rhythmic phrase. Now take the idea one step further and visualise four voices each singing different (but interlocking) vocal and rhythmic phrases and you get the idea. It's counterpoint! If you're still lost, tune to Radio 3 when the composer of the week happens to be JS Bach and cop an earful of his chorale and fugue settings - OK?
It seems rather curious, after spending weeks (months even?) working on drum patterns and fills, honing the interplay between bass and drums to get that perfect groove, carefully layering keyboard textures and arranging speed-of-light guitar solos, and then when it comes to vocals, the Spinal Tap Heartbreak Hotel approach is substituted. Listen to most of the pre-Trevor-Horn-era Yes albums, and you'll realise that it's all happening in the instrumental department, while the backing vocals are definitely dodgy in parts. It's not enough to merely throw together two or three harmonies that sound 'close enough' when heard in the context of a full arrangement.
Now, how can we sort out vocal arrangements that are at least the equal of the instrumental arrangement? A variety of approaches can make your vocal department light up a track. For starters, let's just presume for a moment that you're a good lead singer but have a mental block over harmonies and/or backing vocals. Technology is at hand in the shape of the sampler, obviously. You may like the thought of being backed by the BBC singers, and have a nice line in mixed choruses ready for loading into whatever machine you use, from Emax to Akai and back again. But - have you thought of sampling your own voice? If not, why not? Because you think it sounds naff? Really, that's not as important as you might imagine. The most important point is that (supposedly - and all these judgements are subjective in the end) even a weak voice will sound good 'en masse', singing chords behind a lead voice. To illustrate: Peter Skellern is no Placido Domingo (thank God), but just check out his ensemble vocal harmony work on either his earlier rock albums or the more recent jazz standards (try 'Astaire' for good measure). One voice in isolation may sound weak, but put enough weak voices together and what do you have? No, not 24 seven-stone weaklings who can't sing, but a rather nice laid-back choral texture.
What is important when sampling yourself is the practice of multi-sampling. Unless you're a dyed-in-the-wool Pinky and Perky fan, or perhaps you've pirated old Chipmunks recordings from the 50s, or you happen to think that 'The Laughing Gnome' was the best thing that David Bowie ever recorded (somebody out there probably does!), you'll find that your voice sped up three or more octaves is going to sound avant-garde to say the least. Most singers who have a three-octave range are usually very proud of that fact, and anything over four and a bit octaves is definitely into the realms of fantasy, or heavy metal for that matter. The same principles that I dealt with in the earlier article on string sampling apply here: depending on the memory time of your sampler, you can try intervals of a fifth or less for each sampling point. If you're in the Emax class and above, for maximum fidelity, try samples a major third apart. For instance, go up from C (your first sampling point), to E (your second one), to G sharp/A flat (third point), and up to the next octave C, and so on - but keep in mind the physical limits of the human voice when it comes to range. Don't expect vocals to sound good burping away at 32Hz, or screaming away at 32kHz! Some exponents of the art recommend multi-sampling vocals downwards to avoid quantisation and aliasing problems. When in doubt, experiment!
In addition, why stick to the standard 'Aah' vowel sound? You might like to consider 'Ooohs' and 'Eeehs'; even Hums can be useful. The only time you're likely to come unstuck is in sampling words (unless you're using percussion-type interjections of vocal 'Hey', 'Whoohoo' and 'Yeehaa' sounds), as the dreaded aliasing effect comes into play. Despite what manufacturers' blurb might tell you, all the antialiasing filters in the world aren't going to stop certain kinds of sounds turning out inaccurately and robot-like. We'll deal with a way round this in a moment.
So much for sampling (he says in his most dismissive tone, to hoots of derision - and worse - from the assembled readership). Now if you're not blessed/cursed/lumbered (delete whichever you like) with a sampler or three, how are you going to assemble your choral sounds? Fear not. As any DX librarian owner will tell you, there are a whole raft of vocal-type sounds available for digital synths (you'll find them between 'Rhodes Mk.9,576' and 'Kablooey!' on most DX banks). Oberheim built up a whole reputation on the fact that their OB range's Pulse waveforms could come up with a very pleasing analogue vocal sound, and, to really plumb the mists of time, the Korg PS modular range weighed in at about 9.5 on the Richter scale of cosmicity with their choir sound - and 48-voice poly to boot! The secret of successful synthesis here (as in fact with nearly all types of instrumental sounds) is to try and combine analogue and digital together. DX and Oberheim voice sounds, Roland and Casio perhaps, mix and match to taste and then to top it all off, combine synthesised and sampled voices - and see what you come up with.
Now I can imagine you looking at that subheading and thinking 'Oi - what's all this?' It's the one factor that makes vocals come to life, it sorts out mediocre samples, and it also provides you with a sure-fire way to get your keyboard to sing the lyrics with you. Yes, brudders and sistas, the wonder of the age, a real contribution to road safety, guaranteed to supercede the last piece of gear that was guaranteed to supercede everything else, it's the humble: vocoder...
What? That old piece of history? Herbie Hancock, ELO, 'Mr Blue Sky', and 'Sparky's Magic Piano'? You've gotta be kidding!
No - I'm not. If you think back to the days when the vocoder was the trendy, cosmic, far out machine to use, then 99% of the time one put a rather horrible mono synth through the circuitry; just the sort of stuff for all that 'robotic' tosh of the late 70s and early 80s (Devo et al). In Mr Hancock's case, it was probably an ARP or Minimoog, ELO used an EMS vocoder, while EMI built their own for Abbey Road studios, making its appearance on the very first Alan Parsons Project album (although they had the imagination to put what sounds like a Wurlitzer electric piano through it). The best pre-sampling era choir ever on record was achieved by Japanese synth whiz Tomita, who put everyone else to shame by the simple expedient of combining a Mellotron choir and Roland vocoder. Easy! But at the time of writing, no-one's followed the idea up (don't all rush at once). Now do you see how you can get your vocal samples to sing backing lyrics?
Oh dear, very well, the theory goes like this: Feed your sample/synthesiser/anything else sound into the 'instrument line in' input of your vocoder, and modulate the sound with your own lead vocal. As you sing the lyrics, the vocoder gets the choir sound to do the same!! Roland's VP330 Vocoder Plus does a fine job in the 'synthesiser and vocoder in a box' stakes, as anyone who's a fan of jazz-rockers Mezzoforte will tell you. In fart, their keyboard player, who goes by the Icelandic name of Eythor Gunnarson, gets my vote as the great undiscovered keyboard hero of the 80s, and a master genius in the art of arranging keyboard parts. Check him (and the band) out now!
Aficionados of DJ Kenny Everett will have worked out by now that he used a vocoder over Vaughan Williams 'Tallis Fantasia' for some of his jingles and links on his various radio shows. Back to ye vocoder: your sound source to be modulated does not necessarily have to be direct from sampler/synth, but can also come from your multitrack. You may prefer to do it this way if you're baulking at the thought of playing and singing at the same time, and it will also give those twits who say that the multitrack recorder is on the way out something to consider. Direct to disk? A snip at 34Kiloquid plus VAT? Phooey!
For super realism you may want to try laying down each part of your arrangement line by line, so that the minute differences of phrasing, articulation, etc, end up sounding just like a massed choir (singing notes wrongly, arguing amongst themselves, discussing last night's TV, you know the type of thing). For MIDI fans, the Korg DVP1 will be worth checking out, with its host of extra facilities, but most vocoders will do the job, be they Electro-Harmonix, Moog, EMS, Roland or whatever. The last two companies still manufacture rack mount units that are well worth investigation.
Now we come to the nub and gist of vocal arranging, and this means Harmony. As an exercise in how not to harmonise vocals, listen (if you dare) to some traditional sea shanties, as murdered by some deservedly obscure trad singers. The melody line is in there somewhere, but the harmony lines are often either nonexistent, chaotic, or reveal the approach of 'it seemed like a good idea at the time'. In a six-voice group, a bass line of sorts may be present, thirds and sixths are mostly absent, and droning fifths only turn up at the end of the song - at which point everyone adjourns to the bar. The whole problem lies in the fact that a lot of people don't use their ears when it comes down to working out harmonies. Listen to a four-part chord sung by a slightly suspect vocal group: Who is singing the wrong note? Are they flat or sharp? - and possibly most importantly, is anyone else in the group unconsciously trying to compensate for another's flat note by singing sharp, and vice versa? Such problems can and will occur in the studio with backing vocals. As arranger, it's your job to sort it all out.
Let's look at part writing. Again, some of the principles of string arranging will apply here; unless you're arranging for a massive choir and using at least eight-part harmony, then close voicing the bass and tenor lines will sound muddy. As a general rule, in a four-part harmonisation, the top three parts will be written close together, and the bass line can quite happily be an octave (in fact up to a tenth or more, depending on key) below your tenor line. Have a look at Figures 1, 2 and 3. If you are setting a melody line up for a full vocal arrangement - think about chordal progressions. Do you have to place a chord under each note, or would passing notes sound more classy? To give an example: suppose your melody line is C, D, and E in the key of C major (or Do, Re and Mi for those of you who remember your school singing lessons). Now, a fairly standard way of harmonising that line would be the chords of C, G7 and C again. Figures 4, 5 and 6 show you suggested voicings. Elementary, my dear Swingle. But try C and then a chord of E7 for both D and E. The D resolves up to E (Figure 7) and you're ready to modulate your way through as many keys as you like. Another approach might be C to A11 (Em7 over A bass), then resolving to A7+9 (Figures 8 and 9). In all cases, the secret of successful vocal part writing is to make the changes as smooth as possible. Write a part with lots of flattened 5th and 9th jumps as well as 200 chord changes, and you're asking for trouble. (Unless you're into Gentle Giant, in which case let's experiment!)
Why not? I've never forgiven Brian Ferry for reducing 'Smoke gets in your eyes' to three chords - and neither should you. I would even go so far as to suggest that the use of subtle and ingenious chord changes can have a distinct effect on the emotions, as well as the critical faculties. But how does this relate to vocal arrangements, I hear you groan? All will be revealed.
First principles: and we go back to the Beach Boys. The guv'nors of vocal harmony, of course. But - and this is a big but - where did Brian Wilson get his influences from to inspire him to write those sublime harmonies? The answer is unusual. A variety of sources, obviously, but one of the main ones was a vocal group called the Four Freshmen (and not doowop groups, as all the histories of Rock would have you believe), whose trademark was a jazzy variation on Barbershop quartet singing. Look back even further than that, and you're back to the black Gospel quartets of the forties, in turn taking their cue from Stephen Foster. There's more to this vocal lark than you think! Add Barbershop quartet writing to jazz harmonies, add a surfing beat, and you have the raison d'etre of the Beach Boys' harmonies. Basically, you could break down the construction of their vocal sound into: a lead voice that also doubled in high falsetto lines, three-part harmony and a bass voice. Remember that the BBs were a five-part harmony vocal group, often interchanging parts in the same song. For those of you with long memories: did you know that 'Help me Rhonda', the hit single, exists in two distinct forms? An album version with three-part harmony singing the words 'Help me Rhonda, help, help me Rhonda'; and the better known single version with added vocal bass line and extra falsetto top line ('Help me Rhonda, yeah!'), which suggests that Brian Wilson was readily looking to broaden his arrangements.
From a recording point of view, he used plate reverb extensively on massed voices, which were often double tracked. I'll also hazard a guess that vocal bass lines, as well as being EQ'd to bring up the bass resonances in the voice (acoustically most people's voices are bass light unless they smoke 60 a day, and then they can't sing), were the first thing to be doubled. You can very often hear a doubled vocal bass line over single (ie non ADT'd) higher harmony parts. This also gives you some ideas for your own vocal recordings if you're seeking to recreate that Beach Boys' sound.
The idea of doubling bass lines and adding high falsetto parts to our three-part harmonies leads us today to the current state of vocal groups, typified by the six-part harmonies of black gospel group Take Six. Now a six part vocal group is an extremely flexible thing. Going back to my earlier plug for the legendary Capability Brown, they tended to use the six voices either in straight three-part harmony, or doubled in octaves (Figure 10). Occasionally they would split the part writing in the manner of a sustained triad (three-part chord) answered by another inversion sung by the other three voices. Now if we lend an ear to Take Six, they will do a similar thing, but being a jazz influenced group, the upper triad may be something highly exotic like an 11th, or a 13th with a sharpened 11th. What's that? For example, from the bottom: C (1st voice), G above (2nd voice), B flat (3rd voice), D (4th voice), F sharp (sharpened 11th-5th voice), and A at the top (6th voice). Or Figure 11, to make it apparent to keyboardists. Other stylistic techniques might include doubling the melody line in octaves, still giving you a nice three-part harmony and bass line accompaniment.
Another good way of getting an arrangement to have a sense of movement is to add passing notes and resolutions - for example, add the 9th to a chord (say C), and get it to resolve to the octave before turning it into a flattened 7th. Then change to another chord (say F), making it a major 7th, before turning it to a minor 7th. Then add a new bass note a fifth below that (B flat) and you're talking 11ths (Fm7 over B flat which is B flat 11) and so on. Try it out yourself!
Another wheeze is to try the vocal equivalent of MIDI's 'channel stealing' - giving the impression that there are more musical parts than there actually are. In a polyphonically written vocal arrangement, the bass vocal line may be providing a pulse, say a two-beat bass line (beats 1 and 3), with the other vocal parts off the beat (beats 2 and 4). If you write your bass line to include off the beat notes, your bass singer can add an extra part to the accompanying chord. Figure 12 shows the idea.
Off beat (and off beam) part writing for vocals leads us to the approach taken by the legendary group Gentle Giant; one of whose trademarks were riffs and musical phrases built like a jigsaw, with each instrument (or indeed vocal) playing just three or four notes of a line. The resultant patchwork quilt of sound would have the musical phrase bouncing like a table-tennis ball between musicians. With up to five-part vocal harmony available in the band, phrases would be split lead/two backing vocals/two extra backing vocals, with a choice of voicing. One pair of vocals may sing a phrase in thirds, be answered by the other pair in sixths, fifths or octaves, and then to cap it all, a classical fugue with odd numbered bars and full development. This is all just to show you that the range and complexity of your vocal arrangements are only limited by your imagination.
Queen get their big choral sound by the simple method of getting three voices to sing one vocal line (bassist John Deacon is the one missing on 'Bohemian Rhapsody'), so that by the time you get to a six note chord you're actually hearing eighteen voices! Add a few of Roger Taylor's superhigh top notes, all together now... 'I see a little silhouetto of a man, Scaramouche, Scaramouche, will you do the fandango, hey ho hay nonny no,' etc. 10cc achieved their vocal backwash on the immortal 'I'm not in love', again by recording three voices on each note, but then building up huge clusters of notes and mixing them down to one track at a time on a multitrack tape machine. They were then able to fade them up and down in the mix. Eric Stewart has reported that there are at least 256 voices on the backing tracks, and that building up the textures took three weeks of recording before anything else was put down.
1988's technology can bring this kind of sound within your own reach by the use of sampling delay units (Bel do a nice one) onto which you can overdub many sounds at once, and freeze the lot. You may not, however, quite achieve the tape saturated ethereal quality of the original approach - it's probably too low-tech for a lot of people!
When overdubbing all your own vocals onto multitrack, if you are going to put lots of harmonies on you'll have to consider how much track bouncing you'll have to do. All the usual rules about trying not to bounce onto adjacent tracks, and thus avoiding crosstalk and spill, apply here. In addition, you'll have to take into account how many generations of sound you're losing in your stereo vocal mastermix. If you're going for a 24-part vocal wall of sound plus full instrumental arrangement, and you're recording on eight-track with no MIDI, the chances are the sound quality of some of your backing vocals will be suffering quite spectacularly if you don't take care. Try to tweak some extra top end onto the vocals at the recording stage. You may think that it sounds to bright and harsh (just like some CD mixes, in fact) but by the time you've bounced down two or three times and Mr Dolby's circuitry has done its stuff, things should sound more natural and subdued.
Many people (myself included) use the varispeed on their tape machines with impunity in order to get those notes that are just outside their singing range. Now if you're doing a similar job with a MIDI setup, all the textbooks tell you to transpose your relevant synths up or down to the same amount as the varispeed is running. I never actually manage to varispeed up or down a whole number of semitones, there's always an Eastern style microtone stuck in there somewhere. Then you may have fun and games if you're using an FSK synchroniser system (which is speed related). To avoid brain strain, I would recommend the following: simply put a mono mix of all your MIDI gear onto the edge track that doesn't use the timecode on your multitrack. You could even try the adjacent track to your timecode track if it's unlikely to interfere with the reading of the code at a later time. Simply listen back to the mono mix, speed up or slow down to taste, then when you've finished simply reconnect your sequencer and away you go. If that seems like a waste of an extra tape track, then always remember that you can use the track for something else when all the vocals have been added.
One final point. Leave recording your top harmonies until another day. If you're like me you'll build your choral arrangements from the bass line upwards. The chances are that by the time you get to the top two parts your voice will be knackered. Have a good night's kip, then go back into the studio on the morrow with yourself and your voice refreshed and it'll go down to tape in no time.
Suggested Listening: Any Beach Boys albums; Queen's Night at the Opera; 10cc's The Original Soundtrack; Any albums by Capability Brown and Gentle Giant; ELO's Out of the Blue; any Mezzoforte albums; and last but not least, Take Six's album - Take Six.
This is the only part of this series active so far.
Feature by David Etheridge
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!