Voxos: Epic Virtual Choir | Sound On Sound

Alleluia! Cinesamples' Voxos surround‑ready choir delivers a near‑religious experience. Dave Stewart

After a few years of inactivity, sampled choral collections now seem to be popping up all over the place. Whether this is due to the deep religious feelings we all share or simply a reflection of the Hollywood film industry's obsession with satanic forces, exorcism, Biblical epics and Holy Grail mystery thrillers is not clear, but whatever the underlying cause, today's media composers are keen to get some liturgical choir samples on their hard drives, and sound companies are equally anxious to supply them with the goods. Headed by Long Island natives Mike Barry and Mike Patti (who better to helm a company that sells stereo samples than a pair of mikes?), LA‑based Cinesamples are the latest to offer a sampled choir simultaneously geared to the big screen and tailored to the needs of the modern media composer. Their Voxos Epic Virtual Choirs features a full SATB (soprano, (contr)alto, tenor and bass) line‑up of 40 singers plus a boys' choir and three soloists, recorded from multiple mic positions in St Thomas Chapel, Bastyr University, Seattle. The adult singers regularly perform with the Seattle Symphony and were hand‑picked for their recording experience. Despite their tender age, most of the boys (all members of Seattle's Northwest Choirs) are also soundtrack veterans.

The 24‑bit/48kHz samples are formatted exclusively for Native Instruments Kontakt 4, with free Kontakt Player 4 software included. The DVD version ships on three discs; alternatively, you can help save the planet and (more importantly) avoid cluttering up your music-room shelves by buying the download version. Surprisingly, this costs the same as the DVDs, confirming the widely‑held belief that big, shiny packaging has no intrinsic value. One cautionary note: if you do opt to download Voxos' 35,000 samples, be advised that unless you have a reasonably fast broadband connection it could take a very long time indeed.


Sections & Soloists

Voxos' mixer screen, showing the four microphone positions plus the 'full mix' option, which is a blend of the close, stage and far mics. A set of global three‑way, fixed‑frequency EQ controls is also available here.
Voxos' sopranos, altos, tenors and basses sections contain 10 singers each, while the boys' choir has 15. These section vocalists don't use a heavy vibrato and there's no trace of the pompous, rather stuffy quality you hear in some classical vocal ensembles - the deliveries are natural and fairly plain, which means they'll sound equally at home performing simple Gregorian chants or complex modern choral arrangements. I liked the sopranos' clear, clean timbre and the breathy quality of the altos' low notes: the considerable overlap of their ranges (sopranos C4 (Middle C) to D6, altos G3 to E5) affords plenty of choice for part‑writing. The 20 women team up to perform some nice multisampled 'Mr Sandman'‑style 'bum‑bums' (for want of a better word) which could also be handy for vintage Swingle Singers‑esque Bach vocal arrangements. The male singers match the women's high vocal quality: the three‑octave range of the tenors is performed with impressive accuracy and poise right up to their top, larynx‑busting note of B4. Moving down the scale, if you play loud, low‑pitched fifths, the basses sound like sinister monks, but quiet, closely‑voiced chords pitched higher in their register have a lovely warm and comforting timbre. One day the boys' choir will be able to emulate these manly tones, but for the time being this team of sub‑12‑year olds (uncannily professional and well‑behaved, according to the producers) can move audiences to tears with their purity of tone, which reaches angelic dimensions in their low register (Middle C to G4).
Augmenting these straight articulations is a collection of effects sung by the women (rising shrieks, breathy noises, random pitches, and so on) and men (monosyllabic, staccato, guttural exclamations, quasi-demonic low-pitched 'ehs', and the like). Both sexes also tonelessly chant 'Voxos' in the manner of a group of brainwashed zombies, which I'm sure is a testament to the singers' acting skill rather than the voodoo powers of the production team - however, its usefulness as a 'cinematic effect' is questionable (unless, of course, the film in question was about the making of Voxos). Included in the effects section is a set of atonal and diatonic clusters. The exotic, beautifully‑voiced latter type made me want to hear this choir performing conventional major and minor chords, but there are currently none of those in the library - perhaps this will be remedied in the major free upgrade Cinesamples say is in the pipeline.

The 'Legato Sections' patch brings together all of the choir's five legato sections. Powered by complex, musically intelligent under‑the‑bonnet Kontakt MIDI scripting, Cinesamples' Voice Leading Engine handles voice assignments between the sections and allows their legato intervals to be played polyphonically as well as monophonically.
Taking the sections as a whole, the overall impression is of a medium‑sized choir capable of both intimacy and grandeur, though the grandiosity is not on the scale of the jet‑engine roar one experiences when hearing a full‑scale choral ensemble belting out hits of yesteryear such as The Messiah (ah, they don't write them like that any more).
The library's three soloists are spearheaded by a soprano with a pronounced, operatic vibrato - not really my cup of tea, though fans of the bel canto style should enjoy it. The combination of throbbing vibrato and uninhibited portamento pitch slides (made possible by Cinesamples' legato sampling, more of which shortly) gives the soprano an almost Theremin‑like quality as she glides through her intervals. Non‑vibrato performances would have been a nice option here, but I'm told that 'trained' vocalists of this type often find that near‑impossible.
Happily, the boy soloist has no such difficulty, and his naturalistic, vibrato‑free 'oohs' sound sweet and innocent, a lovely sound to use for a top line in an arrangement. When we come to the solo alto, the library takes a sharp but not unwelcome swing to the left: this singer performs her straight multisamples with a hard 'eh' sound and developing vibrato, and also contributes a set of extemporised phrases sung in a quasi‑Hebraic language of her own invention (I could swear she sings "someone get me a mini‑cab" at one point), utilising the melismas and pitch slides one associates with Indian vocal styles. A nod to the ethnic‑sounding female vocal on the soundtrack of Gladiator, perhaps?


Surrounded

Voxos' samples were recorded simultaneously from four microphone positions: close, stage, far and 'surround'. Patches load with all four mikings in place plus an additional 'full mix' option whose samples are a blend of the first three positions. An on‑screen mixer allows you to mute, solo and balance the mics and assign them to different outputs, but has no panning or stereo width controls. These facilities make the creation of four‑channel surround mixes straightforward: having added a second stereo output in Kontakt, you can simply pan the 'full mix' samples to the front speakers and the 'surround' set to the rear. Route the 'stage' or 'far' miking to a third stereo output pair and you'll be all set to create 7.0 surround mixes.
As is usually the case in this sort of library, no specific provision is made for the fifth, mono, centre speaker, nor for the 'point one' LFE (Low Frequency Effects) channel used in cinematic 5.1 mixes. This is actually a good thing: in a film mix, the centre speaker is usually reserved for dialogue, and the precise nature of the subsonic racket booming out of the LFE channel is best left to the film's sound mixer, who will prefer to create his/her own mono music low‑frequency mix in order to balance it with the bowel‑loosening subsonic rumblings and explosions that are such an important part of the modern cinematic experience.
The high ceiling, hard surfaces and stained glass of Bastyr Chapel create a clean, lively acoustic and a natural reverb that can be heard as a kind of audio 'halo' in all the samples. As you move from the close miking to the more distant positions the reverb effect comes more to the fore, eventually reaching a peak in the 'surround' miking, which reveals a nice, full reverb of just over a second in length.


Syllabub

The Phrase Builder screen allows users to construct words and phrases from a menu of 30 syllables. Cinesamples have addressed the issue of 'wordbuilding' by having their male (tenors and basses) and female (sopranos and altos) sections sing 30 syllables, which can be strung together into virtual words using an integral Phrase Builder. The producers decided against the 'micro-splicing' technique pioneered in EastWest/Quantum Leap's Symphonic Choirs, which allows individual consonants, vowels and diphthongs to be joined together to make composite sounds; instead, they opted to record the open vowels 'ah', 'ay', 'eeh', 'oh' and 'ooh' along with 26 other common consonant‑vowel compounds ('doh', 'sah', 'nay', 'mee' etc.) The diphthongs (compound vowels) which occur in words such as 'die', 'boy', 'loud', and 'hair' are not provided. (See the 'Sections & Articulations' box for a complete list of the syllables used in the library.) All syllables are sung sustained and staccato across the choir's full range, along with unpitched shouted and whispered versions.
A single phrase can hold up to 20 syllables at a time and a maximum of 12 phrases are selectable via a set of keyswitches, enabling you to construct 'lyric sheets' of up to 240 syllables. The Phrase Builder is extremely easy to use; clicking on a syllable adds it to a playlist (or 'string'), the contents of which are clearly displayed on screen. The syllables move automatically through the playlist as each new note or chord is played, starting again from the top when they reach the end of the sequence. Individual syllables can be edited or deleted at the touch of a button, and phrases can be stored as part of a user patch. The syllables are precisely articulated and the timing of their delivery extremely well co‑ordinated across their respective ranges by the men and women. The sustains are looped, so both short rhythmic phrases and ultra‑long notes can be performed.
I admired the Phrase Builder's ability to stretch vowels over multiple notes. If you play staccato, you get a sequence of short, distinct syllables; if you play with the sustain pedal held down, you'll hear a sequence of syllables with no gaps in between. But if you play a vowel sound with notes overlapping slightly (as if playing a legato instrument), the vowel is automatically elongated over those notes. Another neat touch is the provision of release triggers for some syllables, so that their final consonant sounds only when the note is released. All this helps the legibility of the phrases.
In an attempt to wrest some kind of linguistic sense from the syllabic syllabub, I used Phrase Builder to construct the immortal lyric 'video killed the radio star'. Due to gaps in the syllabic menu, this came out as 'vee‑dee‑oh kee dee ray‑dee‑oh sah‑tah', the words clearly recognisable despite the unwittingly amusing Afro‑Mexican rendition. Liturgical chants such as 'kyrie' (kee‑ree‑ay) and 'domine' (doh‑mee‑nay) sounded much more realistic, but there are simply not enough variants to cover all requirements: for example, the nearest you can get to the words 'sanctus' and 'Christus' is 'sahct‑tus' and 'kree‑sah‑tus'. Still, the Phrase Builder gives an unmistakable impression that real words are being sung, which is a big improvement over an unvarying diet of 'oohs' and 'ahs'.


Silky

The opulent interior of St Thomas Chapel, Bastyr University, Seattle, where the choir was recorded. Legato‑interval sampling - in short, the technique of sampling all upward and downward intervals of (usually) up to an octave for every note - ensures that each played interval has its own unique sample, thus enabling consecutive melody notes to run together smoothly. This practice has now practically become an industry standard, so it's no surprise that Cinesamples have put in the requisite superhuman effort and created legato sections for their sopranos, altos, tenors, basses and boys' choir. That's a lot of intervals, readers - if you can't be bothered to do the maths, take my word that it requires an enormous number of samples. The legato sections default to singing 'ooh', which can be switched to 'ah' by sending a MIDI Continuous Controller command (I feel it might have been better the other way round, but only because I feel an overwhelming urge to do a little nit‑picking). You can morph seamlessly back and forth between the two vowels on the fly by varying the CC value. Taken to extremes, this creates a funny ooo‑wa‑ooo‑wa effect, but if used more sparingly it can introduce subtle and unusual timbral shifts. The 'vowel‑morph' control defaults to CC2 (breath control), but that can be changed to any CC of your choice. Transitions between the legatos' two dynamics are controlled by the mod wheel (leaving keyboard velocity to control the hardness of the attack) and sound as smooth as silk.
Voxos' legato section patches benefit from some clever, highly intensive Kontakt scripting that Cinesamples have christened the 'Voice Leading Engine'. When it's disabled, legato patches are entirely monophonic (as is normally the case with this technique). With the VLE turned on you can play chords of up to four notes and also perform monophonic legato melodies while simultaneously sustaining up to three other notes, a very cool innovation that greatly increases the playability of Voxos' legato patches.
When applied to the full choir legatos (five sections combined within one 'Legato sections' patch), VLE automatically handles voice assignments between whichever sections you've selected ('auto divisi', as it were), and also lives up to its name by analysing chord changes and making sure that the individual note movements from chord to chord are as closely adjacent as possible, thus avoiding awkward, unmusical jumps. Though the results are sometimes unpredictable, I was impressed by the musical intelligence of the scripting. (By the way, the three soloists' legatos are strictly monophonic, which makes sense, since very few people can sing more than one note at a time.)
It's worth noting that the type of monophonic legato used in Voxos is the same as that pioneered by Vienna Symphonic Library: when two notes overlap, the second note always cuts off the first, but if you continue to hold down the first note while playing the second, the original note will automatically re‑sound after you release the second note. This makes it very easy to play trills and fast repeated intervals, since instead of having to rapidly alternate two keys while making sure they always overlap slightly (not easy), you can simply hold down a key and play another fast, repeated note with a spare digit. As well as making it easy to execute conventional trills, this opens up nice possibilities for folky melodic gracings and ornaments. I used Voxos to mock up a Ligeti‑style atonal choir for a somewhat avant‑garde arrangement. On paper this sounds like a grand undertaking, but in practice all I had to do was load a single Phrase Builder patch (which contains both the men's and women's performances), select the 'ee' open vowel sound and simply play the parts. For more conventional four‑part writing, you can dial up the Legato Sections patch (which comprises the five legato sections) and the Voice Leading Engine will make intelligent musical choices about note assignments and polyphony. In both cases, there's no need to laboriously fiddle around creating splits or changing MIDI channels: the whole choir is there at your fingertips ready to play.


Conclusion

It could be argued that the current focus on liturgical‑sounding sampled choirs is too conformist - we certainly don't seem to be experiencing a corresponding rash of gospel, African, Asian or pop/rock vocal collections. That said, a good sampled church choir is a staple of the modern media composer's sound library, and Voxos is certainly one of the best I've heard. I'm happy to have this superior choral collection on my hard drive and can unhesitatingly recommend it to my fellow MIDI maestros.

Sections & Articulations

Legato Sections
Sopranos (10).
Altos (10).
Tenors (10).
Basses (10).
Boys (15).
Soloists
Solo soprano.
Solo alto.
Solo boy.
Effects
Men clusters.
Women clusters.
Men effects.
Women effects.
Women 'bum bums'.
Pitched Syllables
Ah, Cah, Mah, Rah, Sah, Tah.
Ay (as in 'ray'), Nay, Ray, Tay.
Eee, Dee, Kee, Khree, Mee, Ree, Tee, Vee.
Oh, Doh, Voh.
Oo, Noo, Too.
Nus ('noos').*
Tis ('tees')*
Tus ('toos').*
Vos ('vohs).*
Frayst.*
Sact ('sahct')*
Unpitched Syllables
Whisper.
Shout.
* Final consonant sounds as a release trigger.


System Requirements

Despite its relatively large size, Voxos will run on fairly modest systems: the minimum requirements are 1GB of RAM and a Mac Intel Core Duo 1.66GHz or Windows Pentium or Athlon XP 1.4GHz machine. Mac owners need Mac OS 10.5 or higher, while Windows users can use XP SP2, Vista (32‑bit/64‑bit) or Windows 7 (32‑bit/64‑bit). When installed, the samples (which were losslessly compressed from the original 35GB of data) use 17.3GB of disk space.
I found I had to increase my soundcard's latency to 512 samples to prevent the legato patches (which use a lot of samples) from glitching. The higher setting theoretically introduces a latency of around 10 to 12 milliseconds. Were we dealing with percussion samples, this might be considered a minor issue, but it certainly wasn't noticeable when playing the choir samples. Nevertheless, it's clear that to get the optimum performance from Voxos one needs a fast, powerful processor.