The Fundamentals of Sound in Post Production

It's time to open up the digital audio workstation and look at the basics of common tools used in post production audio from Equalizers, Compressors, Noise Reduction, and Delay effects.

This lesson is proudly sponsored by RØDE Microphones:
http://www.rodemic.com/

The other videos in this Series:

The History of Sound at the Movies
http://youtu.be/Ot5IryUt9SM

The Science and Engineering of Sound
http://youtu.be/ZbUTyMC8_X8

The Basics of Recording Audio for Digital Video
http://youtu.be/S9cP1WHL0Zo

Introduction to Automated Dialogue Replacement
http://youtu.be/lP_673W270Y

Introduction to Foley and Sound Effects for Film
http://youtu.be/_Jznye0iqYE

---

Closed Caption:

Hi, John Hess from FilmmakerIQ.com and today
we'll get into the post production side of
audio - establishing fundamentals and looking
at digital audio workstation tools for mixing
and perfecting your film's soundtrack.
In this lesson we are going to be diving deep
into shaping sound so it's important that
you have a fair grasp of the different aspects
of sound - You may want to review, if you
haven't already, our video on the Science
and Engineering of sound as we will be using
many of the terms laid out in that lesson.
For this lesson I will be demonstrating with
the tools that are inside the Adobe Creative
Cloud including Premiere Pro and their Audition.
All the tools I mention here should be available
in other audio editing programs and many may
be included in your NLE of choice.
The first tool or weapon in the sound editor's
arsenal is the equalizer. But what exactly
is an equalizer?
In real basic terms an equalizer boosts or
cuts the amplitude of certain frequencies
which alters the harmonics or overtones resulting
in the change of the character of the sound.
Let's imagine the audio response of a wave
as a straight line on a graph where the x
axis represents the frequency going from low
to high and the y axis represents amplitude.
Now let's say we want to boost ONLY the high
frequencies - say everything above 5,000 Hz.
Our straight line is now broken into two levels
with a slope in between: this is called a
high shelf.
This type of equalization, called a first
order filter, is the simplest kind of equalization
to perform using electronic components. This
is found on your basic consumer hi fi systems.
To continue, let's say we want to cut the
sound of the low frequencies in our recording
below 100Hz - our line reflects that with
a low shelf cut.
Now if we go to the extreme and eliminate
all sounds from above or below a certain frequency,
this shelf is called a pass filter - a High-pass
filter essentially lets all the high frequencies
pass, eliminating all the low range, where
as a low-pass filter does the opposite - let
all the low range pass and killing off the
high frequencies.
But what if we want to target a more specific
range of frequencies? That's where second
order filters come in. This is often called
the peaking filter or parametric equalizer
and it has three settings: The frequency,
which is what frequency you wish to target,
the gain: how much you want to boost or cut
that frequency, and the Q or quality factor
which is how wide the parabola of the adjustment
will be. High Q values will have a steeper
slope. Sometimes Q is expressed in octaves
- the more octaves a Q has the more wider
and gentler the effect.
A really high Q filter are used to completely
eliminate a particular frequencies is sometimes
called a notch cut or a band-stop filter.
These are used to eliminate constant frequency
based noise like a electronic hum or to prevent
feedback in a live audio setting.
Another type of equalizer you may come across
are Graphic equalizers. These will be commonly
found on mix boards, they behave the same
way as parametric equalizers except instead
of selecting specific frequencies and changing
the q value, all the frequencies are presented
as sliders with a predetermined interval and
q value.
So how and why do we use equalizers? There
are essentially three main uses- first is
to fix inadequacies in the recording: Microphones
aren't perfect and some have a specific frequency
response and you may want to use the equalizer
to compensate and create a flatter response.
You can also target specific hums with a notch
filter and eliminate them or use a high pass
filter to cut low range rumble caused by wind
noise.
The second use is when you're mixing audio
sources that are competing in a similar frequency
space - a common occurrence when mixing voice
over with a background music track, if you
cut the background music in the 1200 HZ range,
the sweet spot of human voice, you can make
some more room for dialogue or voice over
tracks:
The final reason and arguably most important
use of EQ is for creative reasons making the
track sound better - or just different.
For instance boosting the bass frequencies
on a dialogue track, say around 160hz will
add power to human voices -but too much can
make the track muddy and unintelligible. You
can add a bit of presence by boosting the
5kHz range but again too much will cause ear
fatigue. The sibelence or �ess� sounds
can be found between 4 and 10 kHz, you can
boost this for more of a clear sound or cut
it to get rid of harsh ess sounds. Playing
with these different EQ settings will get
you closer to your desired sound. If you're
mixing instruments - there are many charts
available online that give you a general guideline
for which frequencies to target depending
on the instrument.
You can even go further and push EQ to create
brand new sounds. For instance - EQ can also
be used to simulate the sound coming from
a radio or walkie talkie:
Houston I think we have an EQ problem.
In music, dynamics refer to the general loudness
of a passage from piano which is soft to fortissimo
which is loud and forceful. Dynamics in sound
engineering is same concept - the dynamic
range is the difference from the very soft
to the very loud. Now sometimes we need to
compress that range - to make the difference
between soft and loud passages smaller.
This is the work of a tool called the compressor.
To visualize what the compressor does, let's
use, what else, a graph. On the x axis we'll
put our input level in decibels and on our
y axis will be our output level.
If we don't apply any compression at all we
will have a straight line curve going 45 degrees
up the the chart with a slope of 1. For any
given input, the output will be exactly thesame.
A compressor works by essentially squashing
down sound that goes above a certain threshold
- let's say we we want to dampen everything
that goes above-12dB. A compressor essentially
draws a new line starting at -12dB this time
with say a slope of one half - or 2:1 compression.
This means for ever 2 dB increase in volume
above -12dB from the input, there will only
be a 1 dB increase in output volume. A more
drastic compression would be 4:1, for each
4dB increase of input there would only be
one dB increase in the output.
Compressors have settings for attack and release
to determine how quickly or slowly they kick
in. Too fast and you can get a pumping sound,
too slow and spikes in the audio can slip
through.
Once we have compressed the dynamic range,
we can safely boost the entire track to make
everything generally louder if desired.
If you push the slope flatter to say 20:1
or 100:1 you get what is called a limiter.
A limiter essentially prevents peaks from
going over a specific target generally used
for broadcast and they have very short attack
and release times.
The opposite of a compressor is called, as
you might imagine, an expander. Going back
to our curve - an expander is a part of the
curve that has a slope of greater than 1.
Let's say we want the audio that is below
-20 dB to get quieter faster - our curve reflects
that with steeper slope.
Expanders are generally only used for the
quieter parts of the dynamic range. A noise
gate is one kind of expander - a noise gate
is essentially like a high pass filter except
for amplitude. Anything louder than the threshold
will get through, anything lower than the
threshold will be expanded down into nothing.
Attack and release settings are available
for expanders as well and need to be tinkered
with to find the best settings.
So why would we need to use a compressor?
Compressors help smooth out sudden increases
in volume caused by momentary changes of distance
from the mic or just natural changes in volume.
Smaller dynamic ranges may be necessary for
your venue - if you're mixing audio for a
video that will be shown on the floor of a
subway station there's going to be a lot of
ambient noise and you'll need to boost the
soft parts in order to compete with that noise.
Which gets us to one of the main use of compression
- to make the audio sound more powerful and
louder than it really is. Over the years the
recording industry has move towards making
their albums sound as loud as possible - comparing
the waveforms from a recording from the 70s
and a modern song show just how much compression
is used these days - that's not always a bad
thing as people are often listening to music
in their cars or on earbuds where it's important
to keep a consistent level while having that
feeling of loudness.
A really handy tool for bringing out more
life in an audio track is the multiband compressor
- it essentially combines the best of EQ - the
control of harmonics and overtones with the
control over dynamic range that a compressor
has. A multiband compressor essentially breaks
the track into different bands of frequencies
which you can independently apply compression.
For example on voice tracks you can compress
and boost that 160hz range for adding power
while leaving everything else alone. Most
programs will have several presets to pick
from and I almost always find myself reaching
for the multiband compressor when finishing
my mixes.
Expanders can be used as noise gates which
can push our noise floor lower but there is
another technique in the digital realm for
eliminating noise which is called the Fast
Fourier Transform or FFT.
Inside Adobe Audtion, FFT is a stand alone
filter or part of their noise reduction suite
and it works by first taking a snapshot of
your audio waveform - creating a profile of
the unwanted sound. Then using various settings
you can subtract the offending noise from
the entire track.
Now the problem with FFT processing is too
much can result in something called chirping
which is squirrely weird digital bird sounds.
You can avoid chirping but not completely
removing background noise. To my ears a little
bit of ambient noise is not necessarily unwanted
as it can give a little warmth to a track.
But FFT isn't used only noise reduction as
you can use it to remove practically any sound
from car horns, to footsteps to instrument
hits. There are a lot of neat and amazing
things can be done with FFT.
Now we get into probably the most fun filters
- certain the first ones I tried out as a
kid first playing with a digital audio program.
Using a delay filter - we can create some
really interesting effects. By repeating the
audio with a delay of 15 milliseconds or less,
we get an effect called combing where interference
patterns created resemble that of a comb.
Now combing is generally avoided in the recording
stage, it's caused by quick slappy echo but
as an effect it may be able add something
unique and interesting to the mix.
With a delay of 15-35 millisounds we start
getting chorusing effects where the brain
is starting to perceive more than one voice
or instrument is being sounded. Chorusing
filters can also vary the pitch and timing
of the delays for more effects. This may be
useful for creating bizarre and other worldly
characters for your audio
Beyond 35 miliseconds and we will begin to
perceive an echo effect.
Along with echo are reverb filters. Instead
of being a direct delayed copy, reverb is
the mixture of a large number of random and
decaying echos. Advanced digital reverb generators
can even simulate the time and frequency response
of a specific rooms like concert halls. Echo
and reverb can give your audio track a sense
of space - whether that's a large cavern or
even a small hard room.
Now if we take a wave and we squeeze the time
we are by very essence adjusting the frequency.
Make the time shorter and the frequency will
go up. Stretch it out longer and the frequency
will go down. This is the most basic form
of pitch shifting and it's sometimes linked
to the Chipmunk Effect - where the original
songs were sung at half the speed and an octave
below and then played back at twice the speed.
But say you want to change the time of the
track without chanigng the pitch - or vice
versa. To do this, audio programs use either
Phase Vocoders or sinuosodal spectral modeling
to stretch and squish waveforms making things
like auto tune possible. These essentially
model the new desired sound frequency waves
using rather complicated math which we'll
just leave to the audio engineers and programmers
Wow, we've barely scratched the surface of
what goes into audio engineering and sound
design. But the whole purpose of this lesson
has been to provide some foundational ground
work from EQ, Dynamics, Noise Reduction, and
Time and Pitch effects, I hope this has cleared
up some of the mystery of working with post
audio. No filmmaker, no sound mixer, or artist
working in any medium, can simply watch a
video or take a class and suddenly become
the top of the field. It takes practice practice
practice. And in the case of audio mixing,
a lot of time just fiddling with those knobs
and buttons and experimenting with how it
sounds when you boost this range or cut that
frequency. Experiment, play and repeat - if
it sounds good, and you have a decent pair
of speakers - then it IS good. Don't be afraid
to try things and fail, because it's all on
the path to making something great. I'm John
Hess, I'll see you at FilmmakerIQ.com