Antares AVOX Evo user manual

							
5

Chapter 2: Introducing AVOX
T
he heart of any great song is a great 
vocal sound. With the AVOX Antares 
Vocal Toolkit, we’ve combined five 
state-of-the-art vocal processing 
modules that open up an entirely new world of 
vocal processing capabilities. AVOX gives you 
the power you need to create stunning vocal 
tracks in any musical style as well as design 
unique vocal effects for audio post-production 
applications.
In the following chapters we’ll deal with each 
AVOX tool in turn, but first let’s take a brief 
look at what’s in AVOX and how you might use 
the tools in combination.
AVOX Overview
The AVOX Antares Vocal Toolkit includes:
THROAT Physical Modeling Vocal 
Designer   THROAT is a revolutionary new 
vocal tool that lets you process a vocal through 
a meticulously crafted physical model of 
the human vocal tract. THROAT begins by 
neutralizing the effect of the original singer’s 
vocal tract and then gives you the ability to 
specify the characteristics of the modeled 
vocal tract. 
THROAT’s controls allow you to modify the 
voice’s glottal waveform as well as globally 
stretch, shorten, widen or constrict the 
modeled vocal tract. For even more detailed 
control, THROAT’s graphical Throat Shaping 
display allows you to individually adjust the 
position and width of five points in the vocal 
tract model, from the vocal chords, through 
the throat, mouth and out to the lips. Finally, 
THROAT’s Breathiness controls let you  add variable frequency noise to the model, 
resulting in a range of vocal effects from subtle 
breathiness, to raspiness, to a full whisper.
While THROAT has been designed to allow 
subtle modifications to a voice’s vocal quality, 
the range of the controls also allows the 
creation of vocal tract models well beyond 
the limits of physical human anatomy, offering 
the possibility of vocal characteristics that are 
simply unattainable by any other means.
DUO Vocal Modeling Auto-Doubler
  DUO 
automatically generates a doubled vocal part 
from an existing vocal. In addition to allowing 
programmable variation in pitch, timing and 
vibrato depth, for even more realism, DUO 
also includes a simplified version of THROAT’s 
vocal modeling to provide timbral variation for 
the doubled part. DUO’s output section gives 
you independent control of the level and stereo 
position of the original and doubled voices.
CHOIR Vocal Multiplier   Neither a 
harmonizer nor a conventional chorus effect, 
CHOIR is a unique processor that actually 
turns a single voice into 4, 8, 16, or 32 distinct 
individual unison voices, each with its own 
pitch, timing and vibrato variations. The voices 
can, if desired, be spread across the stereo 
spectrum. Assign an instance of CHOIR to 
each of four voices singing four-part harmony 
and voilà, instant choir.
PUNCH Vocal Impact Enhancer   PUNCH is 
a unique processor that, as its name implies, 
lets you give your vocal more dynamic impact, 
allowing it to cut through a dense mix with 
clarity and power. 
						

							
6

SYBIL Variable Frequency De-Esser SYBIL 
tames vocal sibilance with threshold, ratio, 
attack and decay controls as well as a variable 
sidechain high pass frequency to match any 
vocal performance.
AVOX  Processing Guidelines
While it might seem to be convenient to 
consolidate all of the AVOX functions in 
one integrated plug-in, some of the AVOX 
functions use quite a bit of processing power. 
By organizing them as separate plug-ins, you 
can choose to instantiate only those you need 
for each particular vocal track, thereby using 
only as much of your computer’s DSP power 
as you absolutely need. 
While it’s not likely that you’ll be using all five 
plug-ins on every track, it is likely that you’ll 
often be using more than one. When that’s the 
case, keep the following in mind:
Normal Practice
•  In general, you should progress from the 
cleanest possible source vocal through 
successive stages of processing. THROAT, 
DUO and CHOIR in particular are designed 
to process pitched monophonic voices and 
will function much more effectively with 
clean input signals. Effects like reverb and 
conventional chorus or flanging should 
typically be applied after processing by the 
desired AVOX plug-ins.
•  If a track requires de-essing, use SYBIL as  the first of the AVOX plug-ins.
•  THROAT should be either first in line (fed by  a nice clean signal) or second after SYBIL.
•  If you will be using PUNCH, it should come  after SYBIL and/or THROAT. 
•  If you will be using DUO or CHOIR, they  should typically be at the end of the AVOX 
processing chain.
Special Cases:
•  While you will usually choose between DUO or CHOIR, you can use them together for 
a really huge vocal section. Assign DUO to 
your track and pan the original and doubled 
voices to opposite tracks. Then assign 
separate instances of CHOIR to each of the 
two tracks. Humongous!
•  If you will be using a harmonizer to generate  harmonies from a single vocal line, start 
with SYBIL (if necessary) and then THROAT 
and then feed its output to your harmonizer. 
Assign each of the harmonizer’s outputs to 
individual tracks and use DUO or CHOIR 
on each track to create a vocal group of the 
size of your choice. If desired, PUNCH can 
be used on THROAT’s output or on the 
individual harmonizer outputs prior to DUO 
or CHOIR.
•  If you have a vocal with problematic vibrato  (either too much or too little), you can use 
DUO for vibrato modification. Assign DUO 
to the track and use only the doubled output. 
Set all parameters to their minimum effect 
and use the Vibrato function to adjust the 
performance’s vibrato depth. See Chapter 4 
for more details.
•  For maximum control of the timbre of  doubled voices (at the cost of increased 
CPU requirements), assign DUO to your 
vocal and pan the original and doubled 
voices to opposite tracks. Then assign an 
independent instance of THROAT to each 
part and create a unique timbre for each 
voice.
But all that being said, always feel free to 
ignore any of the above suggestions. The 
AVOX tools offer entirely new realms of 
creative possibilities. There is no wrong way.
Now on to the details. 
						

							
7

Chapter 3: THROAT Physical Modeling 
Vocal Designer
T
HROAT is a revolutionary new vocal tool 
that lets you process a vocal through a 
meticulously crafted physical model of 
the human vocal tract. THROAT begins 
by neutralizing the effect of the original singer’s 
vocal tract and then gives you the ability to 
specify the characteristics of the modeled 
vocal tract. 
THROAT’s controls allow you to modify the 
voice’s glottal waveform as well as globally 
stretch, shorten, widen or constrict the 
modeled vocal tract. For even more detailed  control, THROAT’s graphical Throat Shaping 
display allows you to individually adjust the 
position and width of five points in the vocal 
tract model, from the vocal chords, through 
the throat, mouth and out to the lips. Finally, 
THROAT’s Breathiness controls let you 
add variable frequency noise to the model, 
resulting in a range of vocal effects from subtle 
breathiness, to raspiness, to a full whisper.
While THROAT has been designed to allow 
subtle modifications to a voice’s vocal quality, 
the range of the controls also allows the  
						

							
8

creation of vocal tract models well beyond 
the limits of physical human anatomy, offering 
the possibility of vocal characteristics that are 
simply unattainable by any other means.
In order to understand what THROAT is 
doing and how you can use it to process your 
vocal tracks, it helps to know how our throats 
actually work to create what we perceive as 
unique vocal qualities. 
Vocal production starts with the vocal chords. 
Air from our lungs is forced through the vocal 
chords, causing them to vibrate. The contour 
of this vibration is the glottal waveform. The 
actual shape of the waveform is affected 
by each individual’s anatomy as well as the 
pressure applied to the vocal chords. From 
there, the voice is propagated through the 
throat, the mouth and out through the lips. It 
is the shape of these structures, both their 
length and width, that create the resonant 
characteristics that combine with the glottal 
waveform to define a unique vocal identity.
With THROAT, for the first time, you have 
individual control over each of the elements 
that go into creating a distinct vocal character. 
Whether you are a producer or engineer 
looking to subtly enhance a singer’s 
performance, or a sound designer in pursuit of 
a totally new vocal effect, THROAT will give 
you creative capabilities that have simply never 
before existed.
Controls
In order for THROAT to do the best possible 
job of modeling, it needs to know some basic 
things about the source audio. The following 
three controls are used to characterize the 
vocal that you will be processing:
Vocal Range
Use this control to select the range of the 
track you will be processing. Choices include Soprano Voice, Alto/ Tenor Voice, Bass/
Baritone Voice and Instrument (a general 
setting for anything that isn’t actually a vocal). 
Matching the appropriate algorithm to the 
input results in faster and more accurate pitch 
detection and more accurate modeling.
To select vocal range, click on the Vocal Range 
pop-up and then select the desired range from 
the pop-up list. 
Note: Choosing the wrong Vocal 
Range (or just forgetting to set it at 
all) can result in compromised performance. 
Pay attention.
Source Glottal Waveform
The glottal waveform is the waveform 
produced by the vibration of the vocal 
chords. The range of an individual’s possible 
waveforms is defined by their particular 
anatomy. Within that range, the waveform 
can change pretty dramatically depending 
on the performer’s singing style. Imagine, 
for example, the progression from a breathy 
whisper, to a straight-ahead pop vocal, to hard 
rock or full-on operatic aria. 
A lot of subtle factors influence the glottal 
waveform, but for ease of setting, we have 
characterized the choices as loudness. 
Typically, as loudness increases, so does 
the pressure applied to the vocal chords and 
with that change in pressure a corresponding 
change in glottal waveform. 
When setting this control, select the loudness 
level (soft, medium, loud, intense) that most 
closely matches the performance you are 
processing. If you’re not sure where your 
performance lies on this scale, don’t worry. 
There is no need to obsess over subtle 
distinctions. If in doubt, just pick one. Once 
you have set up a model, you can always 
come back and try another setting and see if it 
makes a (positive) difference. 
						

							
9

To select vocal type, click on the Vocal Type 
pop-up and then select the closest stylistic 
character of the vocal performance from the 
pop-up list. 
Source Throat Precision
If you’ve already been poking around the 
THROAT interface, you might be asking 
yourself, “What the heck is throat precision?” 
And a good question it is. 
As it happens, this control works a bit 
differently than the previous two. Here, you’re 
not being asked to tell THROAT something 
about the input, but instead to tell THROAT 
how precise to be in its attempt to calculate 
the characteristics of the input based on the 
type of modeling you intend doing (the choices 
being subtle, medium, or extreme).
So why not always leave it at “subtle”? 
Another good question. The answer is that 
THROAT’s most precise analysis results in 
extremely accurate characterization of the 
source throat, which works well for models 
that are within the general range of human 
anatomy (i.e. subtle modifications). However, 
for more extreme models, the “subtle” 
setting can in some cases result in artifacts 
— most often a sort of “whistling.” In those 
cases, such artifacts can often be reduced or 
eliminated by choosing a different setting for 
this control.
As a result, the strategy for this control should 
be to always start with “subtle” (which is 
the default) and, if a particular model results 
in artifacts (and you don’t actually like those 
artifacts), change the precision one step at a 
time (i.e. to medium and then extreme) until 
you get the desired effect.
To select source throat precision, click on 
the Precision pop-up and then select the 
appropriate model type from the pop-up list. The next two controls allow you to add a 
variety of breathiness effects to your modeled 
voice:
Breathiness Mix
This control lets you select the amount of 
breathiness component mixed into your 
modeled voice. With a setting of 0, there will 
be no breathiness (apart from what might be 
in the original vocal). At a setting of 100, the 
model will be all breathiness, with none of 
the original vocal characteristic present at all. 
Intermediate settings will produce mixes of the 
original vocal and the breathiness component. 
The sonic character of the breathiness will 
depend both on the articulation of the original 
vocal and, more dramatically, the setting of the 
Frequency control described below.
Command (Mac)/Control (PC) click the control 
to reset it to its default value of 0.
Breathiness Frequency
This control lets you set the high pass 
frequency of the breathiness component (i.e., 
the frequency above which the breathiness will 
be present). This frequency determines the 
audio range and character of the breathiness 
effect. 
At high settings, the effect will be rather 
whispery and ethereal (depending, of course, 
on the original vocal and the Mix amount). 
At lower settings, the effect is more of a 
raspiness (again, depending on the mix). 
Experimentation is the best way to become 
familiar with the possibilities.
Command (Mac)/Control (PC) click the control 
to reset it to its default value of 4000 Hz.
The remaining controls are used to define the 
model vocal tract: 
						

							
10

Model Throat Length
The Throat Length control allows you to 
globally lengthen or shorten the modeled 
throat. Values above 1.00 represent a 
lengthening of the throat while values below 
1.00 represent a shortening of the throat. 
The actual values represent the percentage 
change in the throat length. For example, a 
value of 1.20 represents a 20% increase in 
throat length, while a value of 0.80 represents 
a 20% decrease in throat length.
Changes made to this control are reflected on 
the Graphic Throat Display described below. If 
you have used that display to create a custom 
throat contour, this control will preserve the 
overall contour while scaling it by the selected 
amount.Note: While this control gives you 
the ability to radically change the 
throat length, keep in mind that the variation in 
the length of human vocal tracts is rarely more 
than about 25% in either direction. If you are 
looking for a “realistic” vocal characteristic, 
start with modest settings of this control. (As a 
visual reference, this range is indicated by color 
on the control scale.) More extreme settings 
can produce dramatic results, but probably not 
what anyone would call “realistic.”
Command (Mac)/Control (PC) click the control 
to reset it to its default value of 1.00.
Model Throat Width
The Throat Width control allows you to globally 
widen or constrict the modeled throat. Values 
above 1.00 represent a widening of the throat 
while values below 1.00 represent a narrowing 
of the throat. 
The actual values represent the percentage 
change in the throat width. For example, a 
value of 1.20 represents a 20% increase in 
throat width, while a value of 0.80 represents a 
20% decrease in throat width. Changes made to this control are reflected on 
the Graphic Throat Display described below. If 
you have used that display to create a custom 
throat contour, this control will preserve the 
overall contour while scaling it by the selected 
amount.
Note: Similar to the Length control 
above, this control gives you the 
ability to radically change the throat width. 
Again, if you are looking for a “realistic” vocal 
characteristic, start with modest settings of 
this control. (And again, this range is indicated 
by color on the control scale.) More extreme 
settings can produce dramatic results, but 
probably not what anyone would call “realistic.”
Command (Mac)/Control (PC) click the control 
to reset it to its default value of 1.00.
Model Glottal Waveform
As was explained above in the Source Glottal 
Waveform section, the glottal waveform is 
the waveform produced by the vibration of 
the vocal chords. We used the Source Glottal 
Waveform control to help THROAT neutralize 
the effect of the original vocal’s glottal 
waveform. The Pulse Width and Voice Type 
controls let you define the glottal waveform 
you want to model.
Glottal Pulse Width
This control allows you to select the pulse 
width of the modeled glottal waveform. If you 
are at all familiar with analog synthesizers, you 
can think of this as being vaguely similar to the 
variable pulse width control on a square wave 
oscillator (and if you’re not familiar with analog 
synthesizers, don’t worry, just move the slider 
and listen to what happens).
You will find that the most dramatic timbral 
changes are usually found at the extreme ends 
of this control’s range. If you’re looking for 
realistic, stay in the middle 80% of the range. 
						

							
11

Important Note: Keep in mind that 
this control interacts with the Glottal 
Voice Type control below. Each Voice Type sets 
a default Pulse Width that is associated with 
it. So if you set this control and then select a 
new Voice Type below, the value of this control 
will change to the new voice type’s associated 
pulse width default. For that reason, it is usually 
best to select the Glottal Voice Type first and 
then make any desired adjustments to the 
pulse width.
Command (Mac)/Control (PC) click the control 
to reset it to its default value of 64.
Glottal Voice Type
This control is the converse of the Source 
Glottal Waveform control described above. 
We used that control to tell THROAT the 
characteristic of the original performance. We 
use this control to tell THROAT what kind of 
characteristic we would like to model.
The choices here are the same as the choices 
for the Source Glottal Waveform control (i.e., 
soft, medium, loud, and intense). If you want 
to preserve the stylistic character of the original 
vocal, set this control to the same voice type 
as you set the Source Glottal Waveform (e.g., if 
you set Source Glottal Waveform to “soft,” set 
this control to “soft”). 
As mentioned above, this control interacts 
with the Glottal Pulse Width control. Each of 
the Voice Types sets a default Pulse Width 
that is associated with it. Once a Voice Type 
is set, the pulse width may then be adjusted 
separately. Note:
 Despite the value names 
(soft, loud, etc.), this control does 
not necessarily change the actual level of the 
signal (although the “intense” setting does 
sometimes result in some level gain). Its 
purpose is to model the glottal waveform that 
would result from the various styles of singing. 
If you want to adjust the level, use the Output 
Gain control described below.
To select vocal type, click on the Vocal Type 
pop-up and then select the desired stylistic 
character from the pop-up list. 
Graphic Throat Display
Okay, here’s where the fun really starts. 
The Graphic Throat Display allows detailed 
tweaking of the model throat. For the purpose 
of this control, the model throat is divided into 
four sections defined by five boundary points. 
For reference, these sections and boundary 
points are illustrated on the head graphic to 
the left of the Throat Shaping display. Point 
(1) starts at the vocal chords and the sections 
progress through the throat and the mouth out 
to the lips, which are at point (5).
The Throat Shaping display itself consists of 
two elements: the original throat plot and the 
model throat plot. The original throat plot is 
colored blue and is not user-changeable. It 
represents the original length and width of the 
four throat sections and serves as a reference 
relative to which changes to the model throat 
are made.
The model throat plot is colored red and 
includes five control points that can be grabbed 
and moved, effectively adjusting the length and 
width of individual sections of the throat.
Additionally, the background of the display is 
divided into two sections by color. The lighter 
blue central area represents the range of 
adjustments that are consistent with typical  
						

							
12

human anatomy. As long as all control points 
and plot lines fall in this area, the results of 
your model will be more or less “realistic.”  
The darker blue area represents the range of 
adjustments that exceed the dimensions of 
typical human vocal tracts. When any control 
point or plot line falls in this area, the result 
may or may not sound like something that you 
might recognize as a human voice. The more 
points and/or plot lines fall in this area, the 
more extreme the effect.
When you open a new instance of THROAT, 
the default state of the Throat Shaping display 
is with the original and model throat plots 
exactly superimposed, indicating no difference. 
If you adjust the Throat Length and/or Throat 
Width sliders, you will see the entire red model 
plot move to reflect the changes. Length 
changes are self-evident. For width changes, 
any part of the model plot that lies above the 
original plot indicates a widening of the throat 
while any part of the model plot below the 
original plot indicates a narrowing of the throat.
Here are some things to consider when 
working with the Throat Shaping display:
•  Point (1), which represents the vocal chords, serves as the anchor of the vocal tract. It can 
be adjusted for width but not length. 
•  When you move points (2), (3), or (4)  horizontally, you are adjusting the relative 
lengths of the adjacent sections. For 
example, if you move point (3) to the right, 
you are lengthening the section between 
(2) and (3) while shortening the section 
between (3) and (4). The overall length of 
the throat model remains the same. 
•  Only by moving point (5) can you change the  overall length of the model (apart from using 
the Throat Length slider, of course). 
•  As we’ve mentioned, if you’re looking  for “realistic” vocal characteristics, you 
would do well to start with relatively small 
adjustments that result in all control points 
and plot lines remaining in the central light 
blue area. 
•  In addition to the plot point positions, watch  the contour of the plot lines connecting 
them. It’s possible to place the points in 
positions relative to each other that cause 
the plot lines to bow out towards the edges 
of the display (or even pin against an edge of 
the display). This will almost always result in 
artifacts of one sort or another. (Of course, if 
it’s artifacts you’re looking for, they may be 
just what you want.)
•  When you have created a custom model  contour, the Model Throat Length and Width 
controls will adjust the overall throat length 
and width while retaining (and scaling) your 
custom contour. 
•  Extremely striking effects can be created  by moving plot points in realtime. You can 
do this manually (for one point at a time) or, 
much more powerfully, you can use your 
host’s automation capabilities to program 
movements of all five points simultaneously. 
For the purposes of automation, each point 
is represented by two parameters, one 
for horizontal position (length) and one for 
vertical position (width).
In addition to the original and model plots, 
when THROAT is processing audio the display 
will also contain real-time representations of 
the original and modeled throat contours. As 
with the plots, the blue contour is the original 
throat and the red contour is the model throat.
As you begin to get familiar with the Throat 
Shaping display, trial-and-error will no doubt be 
the first order of the day. However, with a little 
experience, you will soon be able to predict 
what effect a particular plot adjustment will  
						

							
13

have. Checking out the factory presets, with 
a particular eye towards model plot shapes, 
should help you on your way.
Reset
Clicking the Reset button cancels any custom 
contour you have programmed, but retains any 
global Stretch and Width settings set by the 
Model Throat Length and Width controls. To 
reset those controls, Command (Mac)/Control 
(PC) click them to set them to their default 
values.
Output Gain
This control lets you adjust the output level of 
the modeled vocal over a range of +/ - 24 dB. 
As you will discover, some models result in 
substantial level changes. This control is used 
to bring them back up or down to the desired 
level.
Command (Mac)/Control (PC) click the control 
to reset it to its default value of 0 dB.
Level Matching
As mentioned above, some model settings 
result in substantial level changes. The Level 
Matching function attempts to compensate 
for level differences between the original and 
processed versions by automatically applying 
gain adjustments to the modeled version. Very Important Note:  This 
function is provided for ease of A/B 
comparisons with the original vocal. It inserts 
some processing into the modeled vocal path 
that can have a very small but nonetheless real 
effect on the overall audio quality. Once you 
have a model you like, turn the Level Matching 
function Off and use the Output Gain to adjust 
the proper level. 
Click the Level Matching button to toggle its 
state. The button will display On or Off as 
appropriate.
Bypass
This control is used to (you guessed it) bypass 
the plug-in. It has been designed to provide 
artifact-free bypass switching so that you can 
use it to seamlessly enable THROAT only 
where desired on a track.
Click the Bypass button to toggle its state. The 
button will display On or Off as appropriate. 
Keep in mind that “On” means that the 
Bypass function is on, i.e., the plug-in is 
bypassed.
Usage Tips
•  THROAT is a monophonic processor.
•  For best modeling performance, THROAT 
needs to be able to detect the pitch of the 
original performance. To do that, THROAT 
requires a clean, pitched, monophonic 
signal. If THROAT can not reliably detect the 
pitch of the input, either because of a noisy 
or effected signal or because the input is not 
a single monophonic voice, it will not fail, but 
will fall back to an alternative (but not quite 
as accurate) modeling method. 
•  As has been mentioned several times  already, for natural, “realistic” results, always 
start with relatively small adjustments to the 
various throat modeling controls, whether 
the Length and Width sliders, or the Graphic 
Display control points.
  The actual range of variation in human  anatomy is quite small relative to the overall 
range offered by THROAT. When you’re 
first getting started, try limiting control 
changes to around +/ - 15%. As you get a 
feeling for the effect of various settings, 
slowly increase the ranges until things start 
sounding weird (not that that’s necessarily a 
bad thing). 
						

							
14

•  If you just want to start playing, try the following brief  tutorial.
  Start by assigning THROAT to a vocal part  and set the above contour on the Throat 
Shaping display:
  Be sure you have set the proper Vocal  Range and Source Glottal settings and leave 
all the other controls at their default values.
  Now, while listening to your vocal, slowly  move the Model Throat Length slider from 
1.00 to 1.25. Note the effect. Depending, 
of course, on the source vocal, you might 
typically hear the vocal quality become 
subtly darker, more male, older, etc. 
  Then try moving it back to 0.75. Again note  the effect. In this case the result might be 
subtly more bright, female, young, etc.
  Next, return the Model Throat Length slider  to 1.00 and repeat the above with the Model 
Throat Width.
  Finally, experiment with various  combinations of the two controls. As you 
become familiar with their effects, try more 
extreme settings of the controls.
•  Speaking of weird, it may be useful to  understand just why extreme models are 
perceived as “unrealistic” or strange.   As mentioned above, the range of human 
vocal anatomy is actually relatively small. In 
order that we can differentiate between the 
voices of the many people we encounter in 
our lives, the auditory processing function 
in our brains is extremely sensitive to very 
small differences in vocal timbre within the 
typical range of variation. But as a result 
of this, we have no frame of reference for 
vocal timbres that fall outside of the range 
of common human anatomy. We can’t 
picture the person who would sound like 
that. So we think more in terms of “filtered,” 
“tubular,” or whatever. Trust us, if there 
were people with vocal tracts of the more 
extreme dimensions THROAT is capable of, 
this is what they’d sound like.
•  While THROAT has been designed for voice,  experiments here have shown that it can 
produce quite striking effects when used on 
other instrumental tracks. Try it with drums, 
guitars, bass, in fact pretty much anything. 
(In particular, applying it to a drum loop and 
then following that up with PUNCH has 
resulted in some truly twisted beats. Give it 
a try.)
•  Play, play, play.