Antares Infinity user manual

							51
Crossfade Taper: This control adjusts the taper shape of the crossfade that
is used in the attack/loop merge. It is adjustable from 0 (linear taper) to 100
(equal power taper). It should be adjusted to provide the smoothest
transition between the attack and loop segments of the sample.
100 
EQUAL POWER TAPER
0 
LINEAR TAPER
Linear taper should be used for editing simple periodic waveforms like solo
instruments. Equal power taper should be used when editing complex non-
periodic waveforms like string sections or noise.
THE SPR LOOPER
The SPR Looper uses a patented process which randomizes the phase of
each spectral component in the loop. SPR means Spectral Phase Randomiza-
tion. It is excellent for looping sounds which have very non-periodic
waveforms. It is especially good at looping complex sounds which require
exact preservation of the original tone color like string sections, choirs, and
so on. 
						

							52
HOW IT WORKS
The SPR Looper is similar to the Rotated Sums Looper in that it rotates loop
data a random amount through the loop. The difference with the SPR
Looper is that it treats every constituent frequency of the loop separately.
This approach is more thorough than the Rotated Sums Looper. Because the
SPR Looper rotates each spectral component randomly, different trials using
the same input sound will yield different results.
If the computations cause the loop data to exceed 90% of maximum, the
entire file will be scaled so that the loop data are 90% of maximum and
thereby do not clip. Thus, data are scaled so no overflow occurs. When
scaling occurs, the user is notified with an alert which indicates how much
the data was scaled. The scaling information will allow you to maintain
volume control of your samples. If you are getting clipping in the fade from
the attack segment, check out that the crossfade taper may be set more
towards linear taper.
Performance Note: Sometimes the Rotated Sums Looper and the SPR
Looper have an undesirable effect on the output loop. The energy of
any transient sound in the loop will be distributed evenly over the
length of the loop. Clicks and pops will be output as relatively steady
state noise.
For example, when looping a close miked ‘cello, the bow rasp, which
normally occurs in phase with the rest of the waveform, is distributed
randomly across the loop. The result sounds like the ‘cello mixed with
tape hiss. We recommend that these two tools be used only with
instrumental groups or other random non-periodic waveforms.
UNDERSTANDING SPR CONTROLS
The SPR Looper divides the looping process into two distinct phases;
processing the loop data, and merging the attack and loop segments. These
can be controlled separately and executed at different times.
With both phases of the process enabled, the loop data and the attack data
are altered every time the SPR process is run. It is sometimes desirable to
disable the attack/loop merge phase of the process, and do several takes of
processing the loop data without merging the loop with the attack. This
saves time and allows for selecting the best sounding loop before the
attack segment is altered by merging. After the best sounding loop is
found, the merge can be done separately by disabling the loop processing
phase.
Soundfile Menu: This menu has the same function as in the other loopers.
Process The Loop Segment: Check this item if you wish to apply the SPR
process to the data inside loop. When this item is not checked, the slider
labeled Pre-condition Data (%) is inactive and no change will be made to
the loop data. 
						

							53
Pre-condition Data (%): The Freeze Looper and the SPR Looper both use the
Pre-condition Data feature before processing the loop data. The purpose of
this feature is to taper the data at the loop start and loop end so that all
the energy components in the loop begin and end on zero. The control has
a range from 0% to 100%. Any setting larger than a few percent leaves
little or no energy at the loop points to cause a click which might inject
unwanted noise into the process.
Increasing the Pre-conditioning percentage favors the timbre of the sound
near the middle of the loop. Using different values gives different aural
qualities to the resulting sound. The diagram below shows the effect of
changing the amount of the loop tapered by the Pre-conditioning process.
(The taper does not show up in the finished loop.)
LOOP DATA AMPLITUDE AS A FUNCTION
OF PRE-CONDITION DATA PERCENTAGE
0%
25%
50%
100%
Merge The Attack And Loop Segments: Checking this item enables the
second phase of the looping process; merging the loop segment with the
attack segment of the sample. Enable this part of the process only when
you want to crossfade the loop data with the attack segment of the sound.
When this item is not checked, the sliders labeled Loop Start Test Points,
Crossfade Length, and Crossfade Taper are inactive.
When Merge the Attack and Loop Segments is checked in the dialog,
Infinity will automatically search a number of points in the loop segment to
find the point which most closely correlates to the end of the attack
segment. The number of test points that are tried during the search can be
controlled using the Loop Start Test Points control. Processing time in-
creases as more points are tried.
When the best merge point is found, Infinity rotates the data in the loop
until the merge point is located at the loop start. The loop is then merged
with the attack segment. The type of crossfade that is used only modifies
the attack segment. The length and taper of the crossfade that is used to
merge the two segments can be controlled. 
						

							54
Loop Start Test Points: This controls the number of points that are tested
during the search phase of the merging process. It works as in the Rotated
Sums Looper.
Crossfade Length: This control sets the amount of the loop that is blended
with the attack segment. It is variable from 0% to 100% of the attack
duration. A large crossfade length will cause the attack to slowly blend into
the loop sound. A smaller crossfade length will create a quick but blended
transition between the two segments of the sample. A setting of zero
creates no crossfade.
Linear taper should be used for editing simple periodic waveforms like solo
instruments. Equal power taper should be used when editing complex non-
periodic waveforms like string sections or noise.
AT TACK
SEGMENTLOOP
SEGMENT
50%
Crossfade Taper: This control adjusts the taper shape of the crossfade that
is used in the attack/loop merge. It is adjustable from 0 (linear taper) to 100
(equal power taper). It should be adjusted to provide the smoothest
transition between the attack and loop segments of the sample.
100 
EQUAL POWER TAPER
0 
LINEAR TAPER 
						

							55
THE SYNTHESIS LOOPER
The Synthesis Looper tool can be used to selectively smooth out lumpy
sounding frequencies in chorused and ensemble sounds. The new sound is
more stable and loop repetitions are less noticeable. This control is
achieved by allowing you to specify the sidelobe widths of the harmonic
series as a function of frequency. You can control the sidelobe widths to
continuously range from the wide sidelobes of the original ensemble sound
to the zero width sidelobes of the Freeze Looper.
This tool is appropriately used on chorused and ensemble sounds where the
instruments are playing the same notes or notes in octaves. This tool can
also be used as a sophisticated chorusing effect, to generate an ensemble
sound from a solo instrument sound.
This tool is designed for use on monaural sounds. Stereo sounds can be
processed, however, the phase relationships that determine stereo position-
ing will be lost. The resulting stereo sound will have a random phase
relationship, yielding a distributed spatial quality.
COMMENTARY
There are two classifications of sound. The first classification is solo instru-
ments (sans vibrato). These generate uniformly repeated waveforms. A
generic crossfade loop can be used to loop these sounds, however the best
results will be had using the Freeze Looper. The second classification is all
other sounds. These are sounds that change over time. Crossfade loop
quality becomes worse as the sounds become more complex. The SPR
Looper and Rotated Sums are most often used to loop these sounds.
However, a basic problem remains: No matter how perfectly the end of the
loop is connected to the beginning of the loop, the sound in between is
still of a quality that changes over time. Hence, you will always “hear the
loop” because the varying sound repeats itself, over and over again.
This quality of hearing the loop because the varying sound repeats itself, is
sometimes judged as a poor quality loop. It this case, the loop-maker is
really wanting for a tool to change the sound to one which does not have
as much temporal variation.
It is for this reason we have created the Synthesis Looper. This tool gives
you control over that aspect of the sound which creates temporal variation:
the harmonic sidelobe widths. 
						

							56
SOME THEORY
Temporal variation occurs when two sinusoidal frequencies are not exactly
in tune. For example, if one sine wave generator is tuned to 1000 Hz (Hertz
or cycles per second) and another is tuned to 1001 Hz, a 1 Hz “beat note”
will be produced. A temporal variation will also occur when there is any
integer relation between the frequencies involved: if one sine wave
generator is tuned to 500 Hz and another is tuned to 1001 Hz, a 1 Hz beat
note will also be produced.
Many solo instruments (sans vibrato) generate uniformly repeated wave-
forms. These sounds contain energy at the fundamental pitch, and over-
tone pitches, which have frequencies that are EXACT integer multiples of
the fundamental frequency. These frequencies are called the harmonic
series. For example, an oboe play A-440 generates energy at frequencies
440 Hz, 880 Hz, 1320 Hz, 1760 Hz, etc. Zero energy is generated at any
other frequencies (except for perhaps white noise generated at all frequen-
cies by air rushing through the instruments).
The second chair oboe, who does not have as much precision in pitch,
might play an A-440 out of tune, yielding energy at frequencies 441 Hz, 882
Hz, 1323 Hz, 1764 Hz, etc. Notice how the differences in pitch increase
proportionately with the harmonic index (1, 2, 3, etc.). These two players,
playing together will generate a sound with a 1 Hz beat note (a 1 second
repeated variation in sound).
Now suppose there were 100 oboes playing together (with vibratos). What
would the energy as a function of frequency look like? Remember the bell-
shaped curve? Draw a line in the dirt, stand back and throw 100 pennies at
the line. Most of them will land close to the line and some of them will
land far from the line. The number of pennies per inch will be highest at
the line, and lower farther from the line. If you draw a graph of the
number of pennies per inch verses distance, you will plot a bell-shaped
curve. This is technically described as a Normal Probability Density Function
or Normal PDF. If there were 100 oboes players, each playing their version
of A- 440 together (with or without vibratos) , the energy distribution of
the fundamental pitch as a function of frequency would form a Normal PDF
around 440 Hz. The width of the distribution would depend on how good
the players were. If they were very good players playing in tune and no
vibratos, the width would narrow. If they were unskilled and/or added
vibratos, the width would be wider. Since this Normal PDF is energy as a
function of frequency, concentrated at a single frequency, it is described as
a frequency with Sidelobe energy.
These 100 players would also generate energy at the harmonic frequencies.
The energy distribution of the harmonics as a function of frequency would
also form a Normal PDF. As with two players in the example above, these
differences in pitch increase proportionately with the harmonic index. Add
to this some elementary arithmetic about playing in octaves and we get the
following rules: 
						

							57
Rule 01: An ensemble composed of instruments playing the same note
or notes in octaves will form an energy distribution as a function of
frequency having energy at the fundamental frequency and integer
multiples of the fundamental frequency.
Rule 02: The energy distribution at the fundamental frequency will
form a Normal PDF whose sidelobe width depends on the intonation
quality of the players: in-tune players will generate a narrower
sidelobes than players who are more out of tune.
Rule 03: Each harmonic will have an energy distribution at its respec-
tive harmonic frequency that forms a Normal PDF. The sidelobe width
of each harmonic will increase in proportion to frequency with the first
harmonic being two times wider than the fundamental, etc.
There is a final rule for which the authors do not take credit: Energy cannot
be created or destroyed. Nor can energy change its frequency. Acoustic
energy can change its form to heat. None the less, if you are listening to
two flutes playing slightly out of tune, then at the moment at which there
is silence at your ears, there is also at least one loud sound somewhere else
in the room. Be advised that by moving your head, you will change what
you hear.
HOW IT WORKS
The Synthesis Looper tool calculates the energy at each fundamental and
harmonic frequency and allows you to specify the width of the sidelobes
the resulting looped sound will have. In this way, you could take the lumpy
sound of a highly out-of-tune ensemble and change it to the smoother
sound of a higher quality ensemble, making the variations in the loop
repetition less noticeable.
There are other possibilities, too. For example, you can change the first few
harmonics to focus the energy in the sidelobes to zero. This will stabilize
the sound and still provide the warmth of the ensemble at the higher
frequencies. You could also, in the extremes, make a soloist into an en-
semble or an ensemble into a soloist.
You can view the Synthesis Looper window by selecting it from the Looping
Tools menu item: First, you can use the pop-up menu at the top of the
dialog to select any of the open soundfiles for processing or playback. You
can also press the space bar or the speaker icon in the palette window to
play back the selected sound.
This tool performs one, or both, of two functions chosen using the
checkboxes Process the Loop Segment and Merge the Attack and Loop
Segments. The latter function is the same as that in the SPR Looper and
Rotated Sums Looper. It allows you to combine a looped sound into the
attack portion of the sound. You can use both tools simultaneously or you
can first create a satisfactory loop with Process the Loop Segment and later,
merge it using Merge the Attack and Loop Segments. 
						

							58
If the computations cause the loop data to exceed 90% of maximum, the
entire file will be scaled so that the loop data are 90% of maximum and
thereby do not clip. Thus, data are scaled so no overflow occurs. When
scaling occurs, the user is notified with an alert which indicates how much
the data was scaled. The scaling information will allow you to maintain
volume control of your samples. If you are getting clipping in the fade from
the attack segment, check out that the crossfade taper may be set more
towards linear taper.
UNDERSTANDING SYNTHESIS LOOPER CONTROLS
The Synthesis Looper divides the looping process into two distinct phases;
processing the loop data, and merging the attack and loop segments. These
can be controlled separately and executed at different times. 
						

							59
With both phases of the process enabled, the loop data and the attack data
are altered every time the Synthesis Looper process is run. It is sometimes
desirable to disable the attack/loop merge phase of the process, and do
several takes of processing the loop data without merging the loop with
the attack. This saves time and allows for selecting the best sounding loop
before the attack segment is altered by merging. After the best sounding
loop is found, the merge can be done separately by disabling the loop
processing phase.
Process the Loop Segment: In the Process the Loop Segment section, there
are three controls. The Pitch Control allows you to choose MIDI Note or
Pitch Detection. Selecting the MIDI Note radio button will cause the
harmonic series frequencies to be determined using the MIDI note. If the
loop data is not in tune, incorrect harmonic amplitudes will be computed.
Selecting the Pitch Detection radio button will cause the harmonic series
frequencies to be determined by the loop data using the MIDI note as a
guide. This is accurate unless the data is too short.
% Retain Noise: The second control is the % Retain Noise slider and
number box. Some sounds have a characteristic background noise, that
when removed, makes the sound less natural. An example of this would be
a cello section, where the action of the bow on the string adds noise at all
frequencies. Other sounds may have background noise that is undesirable.
If this control set to 100, the noise floor on the output will be the same as
on the input. If it is set to 0, the noise floor will be completely removed.
Intermediate values set the noise floor (energy) proportionately.
Sidelobe Width (%): The third control is the Sidelobe Width (%) graph. This
specifies the output sidelobe width as a percentage of the fundamental
frequency. This percentage is multiplied by the fundamental frequency to
get a frequency variation that is used as the standard deviation of the
Normal PDF energy distribution. You use this graph by dragging either of
the points to a desired location to specify the desired sidelobe widths. Note
that whenever this window is selected and your cursor is over this graph,
you can view the coordinates of the cursor location on the Status Display
Area palette window. The sound of the result is very sensitive to the
position of the left point. For this reason, the two text boxes are provided
so that you can type in numbers, giving you more accuracy than dragging
the point on the graph. Note that the sidelobe width is specified as 10
times the percentage, that is, typing a 7 means .7 %.
Merge The Attack And Loop Segments: Checking this item enables the
second phase of the looping process; merging the loop segment with the
attack segment of the sample. Enable this part of the process only when
you want to crossfade the loop data with the attack segment of the sound.
When this item is not checked, the sliders labeled Loop Start Test Points,
Crossfade Length, and Crossfade Taper are inactive. 
						

							60
When Merge the Attack and Loop Segments is checked in the dialog,
Infinity will automatically search a number of points in the loop segment to
find the point which most closely correlates to the end of the attack
segment. The number of test points that are tried during the search can be
controlled using the Loop Start Test Points control. Processing time in-
creases as more points are tried.
When the best merge point is found, Infinity rotates the data in the loop
until the merge point is located at the loop start. The loop is then merged
with the attack segment. The type of crossfade that is used only modifies
the attack segment. The length and taper of the crossfade that is used to
merge the two segments can be controlled.
Loop Start Test Points: This controls the number of points that are tested
during the search phase of the merging process. It works as in the Rotated
Sums Looper.
Crossfade Length: This control sets the amount of the loop that is blended
with the attack segment. It is variable from 0% to 100% of the attack
duration. A large crossfade length will cause the attack to slowly blend into
the loop sound. A smaller crossfade length will create a quick but blended
transition between the two segments of the sample. A setting of zero
creates no crossfade.
AT TACK
SEGMENTLOOP
SEGMENT
50%
Crossfade Taper: This control adjusts the taper shape of the crossfade that
is used in the attack/loop merge. It is adjustable from 0 (linear taper) to 100
(equal power taper). It should be adjusted to provide the smoothest
transition between the attack and loop segments of the sample.
100 
EQUAL POWER TAPER
0 
LINEAR TAPER