These notes will help you get the most out of Clone Ensemble.
The Clones dial controls how many clones are generated in the ensemble. The range is 1 to 32, and the more you have, the more CPU is used. On a 2GHz P4, 32 clones with a 3-way Sex Machine split (Bass:Natural:Alto) uses a little over 30% of the CPU. For a natural choir sound, the more clones you can afford to run, the better the results. For ADT (automatic double-tracking) a single clone (along with some of the dry signal using the Mix control, below) may suffice.
The Timing dial scales the small delays between each of the clones. The ratios are fixed, this dial merely stretches them all. Tight values are best for thickening synth sounds, middle values are usually appropriate for realistic vocal ensembles and acoustic instruments, and loose is a kind of chaotic delay effect.
There are three Vibrato controls:
The Depth dial controls the average depth of vibrato for all the clones. Each clone has its own vibrato rate, phase, and depth, carefully chosen to work together smoothly. A low value is good for thickening synth sounds, middle values are best for acoustic instruments, high values work well for vocals.
The Rate dial controls the average vibrato speed for all the clones. As mentioned above, each clone has its own variation centred around the average that you can control. If you dial up a very slow vibrato, Clone Ensemble will use a little more memory.
The Random switch causes random variations in the vibrato rate and depth of each clone. When Random is off, the sound will be smooth and even, switching it on may add a little character and realism to the sound. For ADT, you can achieve good results with a single Clone, vibrato Depth quite low, Rate at its slowest, Section D, and the mix halfway.
The three Varying Dynamics controls are quite similar to the vibrato set:
The Depth dial controls the average depth of dynamic variation between the clones. Varying Dynamics is a slow tremolo – a varying volume, so that some clones are louder than their neighbours from time to time. As for vibrato, each clone has its own rate, phase, and depth of dynamic variation.
The Rate dial controls the average speed of the tremolo. As for vibrato, each clone has slight differences in speed, you control the average. The range of speeds of tremolo is much slower than that of vibrato.
The Random switch causes random variations in the rate and depth of the variation in dynamics.
The titillatingly named Sex Machine control lets you split up the clones into groups, and change (or enhance) the sex of some of them. This involves an octave pitch shift, and an associated formant (vocal characteristics) shift:
The Bass group are transposed one octave down, and the vocal formants are deepened. For a natural tenor voice, this gives a very deep bass part – a la Barry White; for a female voice, it will approximate a male baritone sound. You can control the degree to which the formants are shifted with the Bass Formants control, from normal (no shift) to unnaturally deep, and adjust their mix with the Bass Level control.
The Unison group are not transposed – they are the exact same pitch as the input signal. However, you can adjust the formants with the Unison Formants control, to make the sound either more masculine or more feminine, and adjust their mix with the Unison Level control.
The Alto group are transposed one octave up, and the vocal formants are raised. For a natural tenor voice, this can give a convincing female voice; when applied to a female voice it will deliver an altissimo, child-like sound. You can control the degree to which the formants are shifted with the Alto Formants control, from normal (no shift) to unnaturally child-like, and adjust their mix with the Alto Level control.
NB: if you select only one Clone, and turn the timing and vibrato controls to the minimum, you can solo the Sex Machine effect!
If you are creating a multi-part harmony, the Section control provides four variations in the way the clones are organised - each clone has a different vibrato, timing delay, comb filter and position in the stereo image. The A, B, C and D sections shuffle these attributes, so that if you process each harmony part separately, they won't all end up sounding the same.
The next two controls determine how the clones are positioned in the stereo image. Focus controls the spread - they can fill the stereo image evenly from left to right, or you can group them closer together. Balance is similar to a pan control, except that it is controlling the panning of each individual clone rather than the entire mix. Together with the Section control, these two are very useful when processing harmony parts separately.
The Mix control lets you balance the dry unprocessed sound with the wet ensemble sound. For a realistic choir, this should be set close to the maximum. For ADT (automatic double-tracking) try somewhere in the middle. The Dry Delay control let you appy a delay to the unprocessed signal, to help it sit in the middle better with all the clones (which of course have varying delays).
Finally, the Gain control adjusts the final output volume. Some synth sounds get very loud when lots of clones are active – the peaks add up fast. Vocal however tend to “pack together”, and the volume does not build as clones are added – so you might need to boost the levels.
Visual Display Diagram
Near the top of the Clone Ensemble window is a visual display
diagram. This may help you see at a glance the effect of some of the
controls. The audio field is represented by a semicircle with the listener
at the centre. Each voice in the ensemble (clones plus dry signal if you have
any mixed in) is displayed as a small oval. The color indicates the type:
white for the dry signal, green for a natural clone, blue for bass and pink for
alto. The distance from the listener represents that voice’s delay,
and the angle around the semicircle represents its panning
position. Note that this is a diagram of control settings, not a floor plan
of standing positions! In particular, a clone further out simply means it
has a longer delay.
With that in mind, you may find this control very helpful in understanding how some of the controls affect Clone Ensemble’s sound – for example the A/B/C/D Section.
The octave and formant shifting algorithm used by the Sex Machine
only works on clean, monophonic sounds. For best results, sing with a clear,
smooth, pure tone. If you are going to apply any de-essing or compression, do it
before Clone Ensemble to smooth out the sound even more.
Under some circumstances, certain vocal sounds may cause audible glitches,
particularly for Alto clones. You may be able to reduce this by:
- Adjusting the Formants control
- Patching an EQ plugin before Clone Ensemble, with notch cut, somewhere between 500 and 2000 Hz
- Singing in a less strained, less raspy vioce, or with less pronounced vowels.
- Having lots of clones makes most glitches unnoticeable, especially in a mix
Some other plug-ins that use pitch shifting to generate vibrato, chorusing or flanging do so by skipping or repeating samples. This can lead to little artifacts if there is any high-frequency energy in the sound. Clone Ensemble uses interpolation for its pitch shifting, giving perfectly smooth sound even when there's lots of treble. This is one reason it is rather CPU-hungry. Part of that strategy to minimize CPU usage involved using up quite a lot of memory - so older machines with limited memory might be a problem.
How does it work?
Up to 32 clones of the original sound are generated, each with its own unique vibrato and a small delay. The parameters of each clone are carefully chosen to work together.
In addition to that, groups of clones can be transposed up or down an octave, and the vocal characteristics (formants) shifted. Together this can change the sex of the affected clones.
What sort of performance does it get?
Clone Ensemble is quite CPU-hungry. When generating 32 clones with a 3-way Sex Machine split, it will utilize around 30% of a 2GHz Pentium 4. It also requires quite a bit of memory. If you are using an older machine, you may not want to have more than one instance running at once.
What do I get for free?
The unregistered version is hardwired to a single preset - you can't twiddle the dials. The preset is 16 clones, with vibrato and timing optimised for vocals, and multitimbre set on. As a special dispensation, you can adjust the Sex Machine knob, to audition the sex transpose feature.
Will Clone Ensemble work with sample rates higher than 44100?
The code is written to work with any sample rate, but by far most of the testing has been done at 44100. Remember that at higher rates the CPU usage will go up, and the plug-in has to do more work for each second of audio. 32 clones at 96KHz might challenge some older computers.
I'm blind, my screen reader has trouble with the knobs and buttons.
In the standard versions on this plugin, the knobs and buttons are all graphical controls. If you prefer, I also have DirectX and VST versions available with plain, focusable controls. Please contact me and I will email them to you directly.
If you want to use Clone Ensemble to produce a realistic choir sound,
the most important thing is to sing like a choir member.
A pure tone of voice works best, not overly breathy of growly, and with little or no vibrato.
(Vibrato is one of Clone Ensemble's jobs, because in a choir everyone's vibrato is different.) Don't try to sing with an individual, expressive style - that's great for solo singers, but it sounds fake with Clone Ensemble. You simply couldn't get a real choir to sing with that expressivity... perfectly synchronised!
Clone Ensemble can accentuate sibilance in vocal tracks - the "S" sounds can get a bit heavy. Using a little bit of de-esser before Clone Ensemble fixes this.
With around 16 clones, hard voiced consonants like "G" and "D" can sometimes sound a bit buzzy if the timing control is loose. Tighten up the timing, use more clones, or "ghost" those consonants when singing. Alternately, substitute the corresponding nasal sound: G->Ng, D->N, B->M. It sounds weird while you're singing it, and unconvincing when you hear the solo track, but after Clone Ensemble has processed it you can hardly hear the difference.
If you are going to do any compression, do it before Clone Ensemble.
If you are going to use any reverb, do it after Clone Ensemble.
Clone Ensemble uses up a lot of memory and CPU power. If this causes problems (particularly on older machines), the best way to avoid frustration is to use it destructively (ie as an edit command), one track at a time. This means that you have to do your compressing and de-essing destructively first. Keep a copy of the raw recorded data in a separate muted track, so that you have the option of going back and reprocessing.
Clone Ensemble is most effective on vocals and sustained synth sounds (it pumps life into a single-oscillator sawtooth wave). Getting a natural sound from an acoustic instrument may be more difficult.
Going for a natural ensemble sound is a great way to use Clone Ensemble. So is doing the exact opposite - low vibrato and tight timing gives vocals a sound like the classic "doubling" effect - but smoother and more dramatic.
Save often. Lock-ups can happen if you over-tax the system, whether you are using Clone Ensemble or not.
All content and software Copyright 2007 Trevor Magnusson