1000
IT'S IN THE MIX: OPTIMIZING AUDIO TRACKS
Have you ever visited the control room of an audio studio?
Gadgets everywhere. Patch bays, wires, racks populated with cold,
Equalizer -
This box is a fancy tone control. With it you can reduce one or
more sound frequencies or boost one or more frequencies. Since the
world is filled with sounds you want to hear and others that get in
the way, by using an equalizer you can filter out the unwanted
squawking crows, blowing wind, or humming fans while boosting the
sounds of a narrator's voice or a bass fiddle.
Equalizers come in two types: the graphic equalizer which has a
number of sliders, each slider representing a different frequency to
be boosted or cut, and the parametric equalizer with a knob that dials
the frequency you wish to adjust and another knob that determines
whether that frequency is reduced or increased. Audio mixers often
have several parametric equalizers allowing you to dial one frequency
for boosting, another for reduction, and a third for whatever.
Equalizers aren't exact: you can't select 60 Hz on the
equalizer and excise audio hum from a bad recording, without touching
the 59 Hz and the 61 Hz frequencies around it. Equalizers generally
effect a range of frequencies and for this reason are used delicately
in the music recording business. Slicing one particular frequency
range can leave a musical instrument sounding a bit odd.
To the average videographer, equalizers are helpful for reducing
the burble of wind when a microphone is used outdoors. A low cut
filter (also called a high pass filter) reduces the low frequencies
where you find the sounds of wind rumble. Although using such a
filter will reduce the normal bassiness of a man's voice, the effect
is minor and leaves plenty of intelligibility to the narrator's words.
Since speaking voices generally range between 150 Hz and 2000 Hz,
there are a lot of highs and lows that can be cut without losing the
words. Don't cut them unless you have to, however, because you'll be
removing fidelity and ambiance from the sound. You don't want your
narrator's voice to sound like it came over a telephone (telephones
also have a narrow audio bandwidth). In short, unless you have a
nasty noise in the background that you simply must remove so that it
doesn't distract from your desired audio, leave the sound spectrum
alone and deactivate your equalizer.
Equalizers are sometimes handy for "tuning" a room's public
address system. If you have set up your mikes and speakers for a
speech or a play and you hear a ringing or whoop or wail as feedback
envelops the room, it may be that certain frequencies are bouncing
back to your microphone stronger than others. If you can detect which
frequencies those are and reduce them a little, you can reduce the
feedback without lowering the volume and effectiveness of your
loudspeaker system.
I particularly enjoy the chest shaking depth and power that
comes from heavy duty bass, but there are times when low frequencies
are our enemies. Low frequencies involve huge amounts of power. They
saturate (over magnetize) audio tape leaving no room for the other
frequencies. Bass frequencies sometimes suck the power from your
amplifier, also leaving little room for the other frequencies, either
drowning or distorting them. They pin our VU meters (push the needles
off scale) sometimes without adding much useful to the mix. You don't
want to waste precious audio resources, be it wattage, magnetism, or
digits, on noise or thumping or rumble or hum of equipment, wind, foot
stomping, and other non-melodic sounds. Cardioid and other
directional microphones tend to exaggerate bass sounds of people
speaking into them, and often turn percussive consonants like p's and
b's into mini explosions. Fine mesh screens in front of the mikes
(called pop filters) may be our primary defense against such phonetic
fireworks, but our second line of defense is the equalizer or low cut
filter. In short, cut down on the bass frequencies unless there's
something useful happening there.
Next consider the musical bass frequencies you are recording;
these are the legitimate low frequency sounds of various instruments.
You may wish to record these frequencies, but not at their full volume
to avoid saturating the tape or exceeding the ceiling on your digital
recorders.
If you are recording or retransmitting a mix of sounds, you
again may wish to reduce the low frequencies. Low frequencies tend to
"muddy up" the sound mix, reducing the clarity of the other
instruments. If you have recorded an instrument with full fidelity
and are now mixing the tracks together, you may at this time wish to
reduce some of the low frequencies to keep the final sound from
becoming too weighty, ponderous, or muddled.
Low frequencies are a bit unfriendly to reverbs (described
shortly). Again, they confuse the sound space, trampling on the
higher frequencies of other instruments.
At the other end of the spectrum are troublesome sizzles,
whistles, and hisses, often associated with the letter "s". Some
sibilant speakers and some sensitive mikes conspire to emphasize "s"
sounds. Screeching s's are about as appealing as fingernails on a
chalkboard, so a little high frequency trimming may help. There are
also specialized devices called "de-esssers" that, like audio
antibodies, seek out and vanquish just that one consonant alone.
Equalizers go upstream of most other audio processors. In other
words, once your mike signal is preamplified to mixer level or line
level, you pass it through the equalizer before sending the signal to
reverbs, delays, and other gadgets. This is because those other
devices are more likely to respond well to "tailored" sound, sound
they can manipulate easily without dealing with troublesome chaff such
as booming bass and sibilant highs. The entire signal is fed through
the equalizer, treated, and output to the next device, unlike the
signal path applied to reverbs and delays. For reverbs and delays,
part of the sound passes untouched to its destination while another
part of the sound is sidetracked and passed through the reverb or
delay device, and recombined with the original sound. We'll hear more
about this shortly.
Variable gain amplifier -
Variable gain amplifiers (VGAs) modify the dynamic range, the
ratio between loudness and softness of sounds. Examples of VGAs are
noise gates, expanders, limiters, and compressors.
A variable gain amplifer is like a genie with his hand on your
volume control. As the genie hears the sound coming into the VGA, he
raises or lowers the volume of the output sound nearly
instantaneously, and only for a moment. Depending upon the settings,
the genie could lower the volume for a millisecond to take the edge
off a certain percussive sound. Working just the opposite, the genie
could raise the volume for just a millisecond, increasing the punch of
the percussive sound. In the first case, the VGA can tame the
hardness of a drumbeat. Adjusted to an extreme, the VGA can turn the
tap-tap of a snare drum into the chug-chug of a steam engine. In the
second case, the VGA accentuates the drumbeat bringing it to the
forefront of your audio making it more pronounced.
Adjusting the VGA another way, the genie would ride audio like
any audio technician, lowering the volume slightly whenever it peaked
above a certain level, and raising it slowly when incoming audio
volumes seemed a bit low. This is what we have grown to know as
automatic volume control. And just like the real audio person, the
circuit can be "fooled". If you were to speak very softly for a
minute or so, the automatic volume controls would rise bringing your
voice to the right volume. The increased amplification would also
increase the strength of ambiant sounds in the room such as echoes,
hum, hiss, shuffling of paper, whatever. Then if you suddenly
sneezed, the loud sound would knock the volume control down instantly
(but not soon enough to save the listeners from a blast from their
speakers), and now the volume would be so low that your voice would be
barely audible. Slowly the automatic control would allow the volume
to creep back up to normal.
Automatic gain controls are handy when you don't have enough
hands. You simply switch them on and then go ahead to do other
things. They don't do too bad a job; most home VCRs have automatic
gain controls in them to adjust sound volume. A little more subtle,
however, is the limiter. The limiter is a VGA that effects only high
volumes. It's only job is to keep the surprise sneezes from blowing
everyone's ears out. It also keeps an occasional verbal outburst from
distorting badly. It doesn't crank down the entire volume control
leaving you with semi-silence after an outburst; it just cuts off the
peaks leaving the valleys and mid-mountains untouched and natural.
Noise gates work at the other end of the spectrum. When things
are quiet, they turn the volume down more. This is great for reducing
hiss and unwanted background sounds. When music is playing and people
are speaking continuously, you can't hear these weak noises. When the
music or narrator is silent, however, these sounds become audible to
your ears. By turning down the volume during the silent moments, you
don't miss any of your wanted sound, but the noise is reduced during
the silent passages so that it cannot be heard during that special
time when your ears are most sensitive to noise. Noise gates have to
work quickly to turn themselves back up the instant there is
legitimate sound. Perhaps you have heard a noise gate that was over
used. Someone would be talking and between every word there would be
an uncanny silence, and as they spoke, you would hear the whoosh of
some other background sound mixed with their words.
In short, noise gates make quiet sounds disappear. They also
can accelerate the decay of reverberation changing how you perceive
the size of a room. Used in reverse, they can increase the attack of
sound making drums and musical instruments sound more percussive.
Compressors compress the dynamic range of sound. If a sound
were to double in loudness, a compressor could make it increase in
volume by only 50% Weak sounds are hardly touched at all. Loud
sounds and extremely loud sounds end up having nearly the same volume.
Compressors make it possible for sounds that vary excessively in
volume, like the voice of someone shouting, to remain within the range
of what the mixer, recording media, radio transmitters, and
loudspeakers will bear. A typical example of highly compressed sound
is one of those annoying pitchmen on AM radio; he seems to be shouting
in your ear, but the VU meters would show that his actual volume level
rarely exceed 1 dB.
Compressors are good for keeping control of exuberant narrators
and holding the lid on singers and musical instruments that range from
a whisper to a blast. Compressors are also used on wireless
microphones to assure that loud outbursts don't overdrive the
transmitter circuits.
Back to narrators for a moment --- amateur narrators tend to
speak in monotones --- very dull to listen to. Listen to some
professional announcers and you'll notice how they constantly vary the
pitch and volume of their speech to keep the message exciting. This
kind of speech pattern pushes VU meters all over the scale, requiring
constant attention from the audio person. Enter the compressor,
squeezing the whispers and the screams into a 4 dB range. Now the
announcer sounds excited and animated, but the audio levels are tame.
Incidentally, some announcers can mimic animated conversation and keep
within the 4 dB volume range without a compressor. But boy do they
sound weird when they do this at the dinner table!
If compressors are like bondage for sound, expanders are the
opposite, letting sounds leap to life. Unlike compressors, expanders
will do nothing to a weak sound, but will take a loud sound and make
it louder, adding more emphasis to it. Expanders often work in
cooperation with compressors; compressed sound is a little unnatural,
but is the price we pay for squeezing widely varying volumes through
recording devices and transmitters that have limited range. Once we
are past the recording devices and transmitters, and the sound is
ready to be amplified and fed to a loudspeaker, an expander can
decompress a sound returning it to its full dynamic range.
Describing this another way, say you had some sounds that ranged
between 0 and 100 dB in signal strength. Say that your recorder
distorts when it hears anything over 60 dB. You don't want to just
turn down the volume of your recording, that would hurt the weak
sounds too much. Instead, you use a compressor which hardly touches
the low volume sounds between 0 and 40 dB. The incoming sounds
between 40 and 80 dB get squashed down to fit in the range between 40
and 50 dB at the compressors output. The incoming sounds between 80
and 100 dB get squashed into the range between 50 and 55 dB. Thus,
sounds that ranged from 0 to 100 dB going into the compressor came out
between 0 and 55 dB, well within the range of a digitizer, recorder,
wireless microphone transmitter, or whatever. We could stop there or
we could attempt to restore the sound to its original dynamic range
once we have passed our equipment bottlenecks. With an expander, the
incoming barely-compressed sounds between 0 and 40 dB would remain
untouched. The louder sounds between 40 and 50 dB would be amplified
and come out in the range between 40 and 80 dB. The heavily
compressed sounds between 50 and 55 dB would be expanded to the range
80 to 100 dB, thus reconstituting the full dynamic range of the
original sounds.
Like equalizers, varible gain amplifiers are usually connected
early in the sound chain, before signals are sent to reverbs, echo,
delay, and other processors. Also like equalizers, VGAs have all of
the signal pass through them; they don't sample a bit of the signal,
change it, and then add it back to the sound stream.
Delay-
A delay is a repeat of the original sound just milliseconds
after the original sound. It generally adds thickness and richness to
a sound, and can also be used to artificially construct a mental image
of the room where the sound was generated. Imagine for a moment that
a person were speaking to you while standing on a mile tall pedestal.
Presume it's a calm day and you are up there with him. Ninety percent
of the vibrations from his voice will go other places than towards
your ear. The small amount of sound that strikes your ear will seem
thin and strange. Anechoic (soundproof) rooms and "dead" (soundproof)
audio studios also sound this way. Music and voice both sound
unnatural because, in the real world, we always have a floor and walls
around us reflecting delayed strains of the original sound back to our
ears.
Now imagine someone with a tall crane hoisting a wall and
positioning the wall behind the person speaking. If the wall is
merely a foot behind the speaker, you will hear his original words
plus a weaker delayed repeat of those words about 1 millisecond later.
His voice will sound stronger and more realistic, but you won't quite
know why. One millisecond is such a quick delay that your ear cannot
tell that it's hearing the sound twice. If a wall is moved slowly
away from the speaker, the delay becomes greater. Between 5 and 15
feet, the wall will continue to reinforce the person's voice making
him sound natural and normal, but the sound will change character as
the wall increases its distance. When the wall is 15 feet away from
the speaker, the sound is delayed about 30 milliseconds and your ear
begins to perceive it as a discernible delay or a repeat. Moving the
wall further eventually turns the delay into your classic
echo...classic echo.
When a delay time is less than 5 ms, some of the delay's
electrical vibrations cancel some of the original vibrations. Certain
frequencies will be nullified while others remain (a condition called
comb filtering). The result is a hollow, through-a-pipe sound, a bit
like the voice of Darth Vader. Somewhere between the zone of
cancellation and the zone of distinct delays is the desired
reinforcement zone where the delays are constructive and thicken the
sound. These are the delay amounts you will find most useful for
normal audio sweetening.
In the real world, we generally have more than one wall, so
there are several sound reflections picked up by the microphone and
recorded. By combining several delays together, one can recreate the
perception that a room of nearly any size exists. If the delays are
all short, a voice can sound as though it were coming from within a
vehicle, a hallway, or a small room. If the short delays are strong
(ie. nearly as loud as the original sound), the sound will appear to
be coming from a bathroom. Increase the delay time and the room
becomes larger. Strengthen the delay volume and the walls become
"harder" like in a gym or parking garage.
How can you use this in the video world? Say you made a
recording of a person speaking in a room. If while editing you need
to dub in new lines, you could bring that person back to that room and
record them, thus maintaining the "room sound". If the room isn't
available and you bring that person back to the studio or have someone
else record the person in another town, naturally the character of the
sound will be different. You would try to use the same kind of
microphone at the same distance from the person in order to match as
many variables as possible, but often the subtle difference in room
reverberation will make your inserted audio sound glaringly different
from the rest of the recording. Here is where you can run the signal
through a delay (or several) and attempt to sonically recreate the
missing walls.
For a more creative example, we are always trying to make sets
seem like the real thing on the TV screen. Sound carries some of the
subconscious message to the listener, so if you can make your tiny set
sound like a big room, or the metal interior of a submarine, you can
transfer the audience to a place that never existed. Some inexpensive
cardboard and paint and a few delayed sounds can become a convincing
spaceship or foxhole without the cost or cramped quarters.
When working with music, delays are useful for creating the
impression that there are more than one instrument. This trick is a
flimsy one, however. In the real world no two instruments play
exactly the same note at the same frequency in the same phase all the
time. One voice cannot be simply duplicated into a chorus of singers.
In the real world, some voices will vary above and below pitch
slightly. To more closely approximate the real world, there's a
button found on delays and music synthesizers called chorus. The
chorus is a swept delay which is a combination of one short fixed
delay and one changing delay. The changing delay is like having a
wall coming toward you, and then moving away from you. Electronically
the swept delay is being adjusted from a small amount of delay to a
larger amount and then back to the smaller amount. The frequency
varies a little above and below the normal frequency. Other
adjustments on your processor allow you to vary the modulation (rate
of sweep) of the swept delay creating a slow waaahhhhaa sound or maybe
a quick fluttering wowowowo sound. If you listened to it, it would
sound a little like vibrato in a singer's voice, or the sound of a
train whistle as it goes by you. To make the chorus sound even more
natural, the sweep rate is varied, perhaps making a woowowoooowowwo
sound. The random or fake-random sweep is similar to what you hear
when two voices are singing the same note.
Reverberation -
Reverberation is the reflection of sound in a defined space. It
is the sound you hear when singing in the bathroom, in a stairwell, or
in a cavern. The word is used interchangeably with echo which is
technically something different. Echo means decaying repeats of a
sound....sound...sound. Echo is what you hear when you scream "hello"
to a canyon wall. Echo makes an interesting sound effect, but will
turn a voice into cacophony. Guitarists can use echo to turn one
pluck of the string into many, adding complexity to the music. For
the most part, echo is simply used as an effect, and reverberation is
the preferred flavoring for sound. Incidentally, many mixers have
controls on them called echo send. This is a mixer circuit that takes
the microphone's preamplified signal, and sends it out of the mixer
for further processing (by an echo, delay, reverb, or whatever), then
takes the result back into the mixer and recombines it with the
original sound. You would vary the amount of the two signals with a
knob on the "echo send" part of the mixer.
Reverberation used to be made with mechanical devices such as
metal plates and steel springs. Audio signals would be transduced
into physical vibrations causing the spring to bounce around or the
plate to vibrate. At the other end of the spring or on a parallel
plate would be another transducer that changed the mechanical
vibrations into an audio signal again. Today most reverbs are digital
audio devices which sample a sound, convert it into numbers,
manipulate (perhaps repeating) the numbers, then convert the data back
to sound. Reverbs are often souped-up delays that feed their own
delayed sounds back into themselves in complex manners. Where a delay
represents one repeat of a sound (like from that wall behind the guy
on the pedestal), reverb is many repeats blended together (as you'd
expect from a real multi-walled room).
Reverberation creates a room ambiance, much like delay does.
You can think of delay as a couple walls (usually the closest and most
important ones) and think of reverb as the entire room. There would
be a direct reflection from the nearest wall and perhaps the floor,
followed by weaker more numerous reflections from the rear wall of the
room, ceiling, and other places. Reverberation forms mental space so
that you can tell whether something was recorded in a living room or a
gymnasium. Reverb adds natural depth and excitement to music and
plays an important role in gluing together independent sounds that
have been mixed together from separate recordings. You could feed
many microphones into a mixer or have many tracks in an audio
recording, and each of them may sound singular and independent even
though they are mixed together. By using reverberation, the sounds
blend more naturally.
Reverberation devices may have a control called diffusion which
determines whether the sound reflections are highly defined or mixed
in a fuzzy way. In the physical world, parallel walls would give low
diffusion and high definition to a sound because the delayed
reflections come back to your ears fairly intact. Non-parallel walls
give high diffusion; there are no audible repeats of the original
sound, just the amorphous ring of the sound in the background.
Depth is another knob you may find on a reverb and it controls
the perception of where you are sitting relative to the speaker or
musical instrument being played. Listening to a trumpet from the
front row of an auditorium sounds different than listening from the
back row. In the front row, you hear an immediate, strong, original
sound followed by almost instantaneous direct reflections from the
floor and backdrop, followed by a weak reverberation of the sound from
the back wall or ceiling. If you sit in the back row, the reflections
hit you at the same time as the original sound, and the sound is weak
relative to those reflections. In short, the music "sounds" far away.
By turning up the depth control on a reverb unit, and weakening the
original sound that is mixed back with the reverb once the two have
passed through your mixer, you can create the illusion of a distant
voice or musician.
Faking things to sound real -
In real life, we are surrounded by natural reverberations. The
high frequency reverberations decay first leaving just mid and low
tones. Although this is natural, it doesn't sound good. When strong,
low frequencies are fed into a reverb along with high frequencies, the
low sounds cloud the mixture, sounding muddy. For better results,
audio technicians generally run the sound through an equalizer first,
cutting the low frequencies, then send the result to the reverb. This
is called an equalized reverb and it sounds brighter than reverb with
the full sound spectrum. In fact reverb settings sometimes use the
words dark and bright to describe the tonal character of the reverb.
All reverbs allow you to set the decay time, the length of time
it takes for the reverb sound to trail off. Fancier reverbs have a
low and high frequency decay adjustment to again give the high
frequencies an edge over their low frequency brothers. You might
adjust the low frequency decay to be quick and the high frequency
decays to be slow. Again, this makes up for how low frequencies muddy
up the sound of a reverb.
Reverb, like salt, should be used in moderation when flavoring
music. If you have a voice and an instrument, for instance, one of
them should be recorded dry (without reverb). This will help one
sound stand out from the other. If you have a lot of production in
your sound track (ie. many instruments or many things going on), less
reverberation is better; too much muddies up the sound. If you have
sparcer production, such as a single voice or only a few instruments
(ie. a single saxophone, singer, or flute) more reverb may give the
desired dramatic effect. Slow songs allow you to use a long decay
time in your reverb while fast songs need a fast decay so that the
reverb doesn't get in the way of the next note. For the highest
impact, you may decide to have a small section of music with no reverb
at all. This builds excitement through comparison and avoids
monotony.
Equalizers, VGAs, delays and reverbs are just a few of the
little black boxes you will find in the hands of audio technicians.
Videographers can also use these gadgets to season their sound and
create a sonic space that transports the listener into the world that
you have created. Like lighting, camera angles, and the use of color,
the sonic space you contrive sends a subconscious message that draws
your viewers into the program and captures their minds.
NOTE:
First Light Video Publishing (800-777-1576) markets five
excellent videotapes, the "Shaping Your Sound Series" ($329); hosted
by engineer and producer Tom Lubin. The tapes contain graphic
animations, live music examples, and clear demonstrations of recording
techniques and equipment operation. Two of the tapes specialize in
reverb, delay, equalizers, and gates.