You’ve just finished recording what you think is a pretty good song in your project studio. The
playing is good, the recording is clean and the mix is decent. So you burn it to a CD and
proudly pop it in your CD player. But when you hear it played after a “commercial” CD, you
realize that something is wrong.
What’s Wrong With My Song?
•It’s not loud enough. It sounds wimpy next to other CDs. Turning it up or mixing down
at a higher level doesn’t solve the problem. It sounds louder, but not, well LOUDER.
•It sounds dull. Other CDs have a sparkle that cuts through with excitement. You try
boosting the EQ at high frequencies, but now your song just sounds harsh and noisy.
•The instruments and vocals sound thin. Commercial songs have a fullness that you
know comes from some sort of compression. So you patch in a compressor and turn
some controls. Now the whole mix sounds squashed. The vocal might sound fuller, but
the cymbals have no dynamics. It’s full…and lifeless.
•The bass doesn’t have punch. You boost it with some low end EQ, but that just sounds
louder and muddier. Not punchier.
•You can hear all the instruments in your mix, and they all seem to have their own
“place” in the stereo image, but the overall image sounds wrong. Your other CDs have
width and image that you just can’t seem to get from panning the individual tracks.
•You had reverb on the individual tracks, but it just sounds like a bunch of instruments
in a bunch of different spaces. Your other CDs have a sort of cohesive space that
brings all the parts together. Not like rooms within a room, but a “sheen” that works
across the entire mix.
Don’t worry. It’s not that you’re doing anything wrong. There are just some things you still
need to do to get that “sound”. You just need the right tools and an understanding of how to
use them. You won’t become Bob Ludwig
dramatic improvements in your master recordings with a little work.
We put this document together to help others in their quest for better sounding masters. We
don’t claim to be mastering masters. If we could master the next Christina Aguilera hit would
we be writing code and manuals or sitting in a mastering studio with Christina Aguilera?
What we can give you is professional quality mastering software (iZotope Ozone™) and
guidance on how to use it. But in the end there are no right answers, no wrong answers, and
no rules. At least if there are, we still haven’t found them. So in the end just experiment and
have fun.
Intended Audience For This Guide
•If you don’t know anything about mastering and don’t have Ozone, we still
hope this guide will help you. Sure, we think you should use Ozone. But we
learned a lot about mastering from “the online audio community” and we
1
overnight (or probably ever) but you can make
1
http://www.gatewaymastering.com/ Bob Ludwig has won the TEC award for mastering every year he’s
want to give something back in return (in addition to iZotope Vinyl2). This
guide can be freely copied or distributed for noncommercial purposes for that
reason.
•If you don’t understand mastering but do have Ozone, you’re in luck. Ozone
gives you the tool to get “that sound” and this guide shows you how to do it.
•If you have Ozone and know the basics of mastering, this guide will stil l show
you tricks or techniques that are possible in Ozone. Just say “yeah, I knew
that” when appropriate for the other parts.
2
http://www.izotope.com/products/vinyl/vinyldx.html Analog modeli ng plug-in for lo-fi destruction. Tha t
Although there are many definitions of what “mastering” is, for the purpose of this guide we
refer to “mastering” as the process of taking a mix and preparing it for manufacturing. In
general, this involves the following steps and goals.
The “Commercial Sound”
The goal of this step is to take a good mix (usually in the form of a stereo file) and put the
final touches on it. This can involve adjusting levels and in general “sweetening” the mix.
Think of it as the final coat of polish, or the difference between a good sounding mix and a
professional sounding master. This process can involve adding broad equalization, multiband
compression, harmonic excitation, loudness maximization, etc. This process is often actually
referred to as “premastering” but we’re going to refer to it as mastering for simplicity. Ozone
was created to specifically address this step of the process: to put that final professional or
“commercial” sound on a project that’s been mixed down to a stereo file.
Consistency across the CD
Consideration has to be made for how the individual tracks of a CD work together when played
one after another. Is there a consistent sound? Are the levels matched? Does the CD have a
common “character”? This process is generally the same as the previous step, with the
additional consideration of how individual tracks sound in sequence. This doesn’t mean that
you can make one preset in Ozone and just use it on all the tracks so that they all have a
consistent sound. Instead, the goal is to minimize the differences between tracks, which will
most likely mean different settings for different tracks.
Preparation for Duplication
The final step usually involves preparing the song or sequence of songs for manufacturing and
duplication. This step varies depending on the intended delivery format. In the case of a CD it
can mean converting to 16 bit/44.1 kHz audio through resampling and dithering, and setting
track indexes, track gaps, PQ codes, and other CD specific markings. Ozone is not designed to
address these functions by itself, but instead meant to work within dedicated products such as
Steinberg Wavelab, Sonic Foundry Sound Forge, Cakewalk SONAR, Adobe Audition (Cool Edit)
and others.
Technically, Ozone is a plug-in, although it really encompasses several modules to
provide a complete system for mastering (or technically “pre-mastering” as it addresses the
processing but not the CD layout, file conversion, etc.) In addition to providing audio
processing, it provides meters, tools for taking snapshots of mixes, comparing settings, and
rearranging the order of the mastering modules within the system.
64-bit Audio Processing
When processing audio, Ozone can perform hundreds of calculations on a single sample of
audio. In a digital system, each of these calculations has a finite accuracy, limited by the
number of bits used in the calculation. To avoid rounding errors from interfering with the
audible portion of the audio, Ozone performs each calculation using 64-bits. Can you hear 64
bits? No. But that’s the point. The rounding errors (inherent not just in Ozone but in any
digital system) are pushed down into the inaudible range with Ozone.
Analog Modeling
Ozone is the result of extensive research in analog modeling, i.e. creating digital processing
algorithms that mimic the character of analog equipment. While it’s technically impossible to
model analog equipment exactly with digital 1s and 0s, Ozone provides compression,
equalization, and harmonic excitation that recreates the behavior exhibited by analog
equipment.
So what is this “character” of analog? There have been volumes written on this topic, and
we’re not sure if anyone really can explain it completely. But in the most general sense,
analog processing has certain nonlinear aspects that a mathematician would consider "wrong"
but many people believe sounds better musically. Any analog equalizer, for example, applies a
small phase delay to the sound. These types of “imperfections” provide the analog
characteristics of warmth, bass, sparkle, depth and just an overall pleasing sound.
Digital Precision
While analog modeling can provide a pleasant character or “colorization” of the sound, in some
situations precise or “transparent” signal processing is desired. For example, you may wish to
equalize or notch out a frequency without introducing the phase delay inherent in analog
filters as which was mentioned above. For these applications, Ozone also provides digital or
“linear phase” equalizer modes and multiband crossovers. Which should you use? It’s entirely
subjective, and with Ozone you have the choice of processing modes.
Meters and DSP
Some mastering engineers don’t need meters. They only need to listen. They can hear a
sound and know its frequency, or hear a level and know when it’s compressing. For the rest of
us, though, each module within Ozone combines audio processing controls with visual
feedback through appropriate meters. When equalizing, you can see a spectrum. When
compressing, you can see a histogram of levels. When widening, you can see phase meters.
There is no substitute for using your ears, but think of it like driving a car. When you first start
driving, you spend a lot of time looking at the speedometer. Over time, you develop an
instinct and need the meters less. But from time to time, we’ve all looked down and thought
“hmmm, I had no idea I was driving that fast”. Whether using Ozone or not, whether you’re
just starting with mastering or have been doing it for years, you can always benefit from the
second opinion that a good set of visual displays can provide.
UI Efficiency
A mastering session can be long and tiring. The last thing you need to be stressed about is
how to turn a knob with a mouse. There are no knobs in Ozone. It’s pure software, not
software stuck in some hardware paradigm of yesteryear
3
. Instead of spending time thinking
about how to make Ozone look like a 1960s compressor, we spent countless hours using it
and refining it to make it as usable as possible. It’s flat and simple with support for keyboard
shortcuts and wheel mice.
3
We’re not religiously against the hardware look. iZotope Vinyl has knobs and screws and brushed steel.
In a simple plug-in that can be fun, but Ozone had far too much depth to continue that “hardware”
paradigm.
To master on a PC you need some type of editing software and a sound card. There are plenty
of reviews and articles on software and sound cards, so we defer to other sources for you to
make your choice.
One important point is that when mastering you’re really just focused on improving a mixed
down stereo file. Applications such as Wavelab, Sound Forge, and Adobe Audition (Cool Edit)
are designed specifically for working with stereo files. However, you can bring a stereo file into
a multitrack program (i.e. SONAR, SAW, Samplitude, Vegas, Cubase, Nuendo, Logic, etc.) as a
single stereo track and master it that way. We caution you against doing mixing and
mastering in one step, though. That is, trying to master while also mixing the multitrack
project. While you could put Ozone as a master effect on a multitrack project, the first
practical problem is that this requires more CPU than necessary as the software is both trying
to mix your tracks as well as run Ozone (which does require more CPU than a typical plug-in).
The second problem is that you’re tempted to try to mix, master, arrange, and maybe even
rerecord in the same session. When we’re working we like the separation of recording/mixing
and mastering. You focus on the overall sound of the mix and improving that instead of
thinking “I wonder how that synth part would sound with a different patch?” Get the mix you
want, mix down to a stereo file, and then master as a separate last step
Mastering Effects
When mastering, you’re typically working with a limited set of specific effects.
• Compressors, limiters, expanders and gates are used to adjust the dynamics of a mix. For
adjusting the dynamics of specific frequencies or instruments (such as adding punch to
bass or warmth to vocals) a multiband dynamic effect is required, as opposed to a
single band compressor that applies to the entire range of frequencies in the mix.
• Equalizers are used to shape the tonal balance.
• Reverb can add an overall sheen to the mix, in addition to the reverb that may have
been applied to individual tracks.
• Stereo Imaging effects can adjust the perceived width and image of the sound field.
• Harmonic Exciters can add a presence or “sparkle” to the mix.
• Loudness Maximizers can increase the loudness of the mix while simultaneously
limiting the peaks to prevent clipping.
• Dither provides the ability to convert higher word length recordings (e.g. 24 or 32 bit) to
lower bit depths for CD (e.g. 16 bit) while maintaining dynamic range and minimizing
quantization distortion.
4
4
Like everything in this guide, this is just our suggestion based on the way we work (when we’re working
on music and not coding DSP). Work the way you work best.
We don’t think there’s any single “correct” order for effects when mastering. In Ozone, the
default order of the mastering modules (the path the signal follows through Ozone) is:
1) Paragraphic Equalizer
2) Mastering Reverb
3) Multiband Dynamics
4) Multiband Harmonic Exciter
5) Multiband Stereo Imaging
6) Loudness Maximizer
7) Dither
This order can be changed. In fact, you should experiment with different orders. The only
exception in all cases that we can imagine is that if you’re using the Loudness Maximizer and
Dither they should be placed last in the chain.
Bonus Tip: For a complete guide on dither, we invite you to check out our dithering guide at
To change the order in Ozone, click the “Graph” button.
This brings up a display of
the modules. You can
reorder the modules by
simply dragging them
around.
Note that the location of
the meters in the signal
chain can also be changed.
This allows you to set
whether the spectrum is
based on the signal going
into or coming out of the
EQ, for example.
Monitors
It’s important that you monitor on decent equipment when mastering. If your playback system
is coloring the sound, you can’t possibly know what’s in the mix and what’s caused by your
playback system.
That doesn’t mean that you can’t get decent results with relatively inexpensive equipment.
The key is knowing the limitations of what you’re monitoring on and learning to adjust for it in
your listening.
For studio monitors, the most common problem is lack of bass, specifically below 40 Hz or so.
These monitors just don’t have the size or mass to move that much air at that low a
frequency. One solution is to complement a pair of studio monitors with a subwoofer. If so,
make sure you adjust the subwoofer so that it doesn’t exaggerate the bass.
How do you do this? If you have a mic that’s flat down to 20 Hz, here’s a quick and dirty way
to do it.
1) Take a song with a good range of frequencies in it. We just randomly chose Vasoline
(Stone Temple Pilots)
say this was the quick and dirty method)
2) Put Ozone’s spectrum in average mode and loop a section of the song. Save it as a
snapshot (click the Snapshot button, click Snapshot button A and you’ll see a frozen
blue line)
3) Place the mic in the spot where you would be listening from, and play the loop through
the monitor/subwoofer combination. We used Cakewalk SONAR with effects on input
enabled, so that we could see the result in real time.
4) Adjust the subwoofer level until the sound picked up by the microphone (the green
line) is close to the spectrum of the source (the blue snapshot).
5
. As long as there’s a broad spectrum, it doesn’t matter (we did
It’s not exact and there are several variables here (the response and location of the
microphone being the most significant) but it can get you close.
You’ll never get a perfect listening environment, and you can never predict how what you’re
listening to will translate to the systems others will use to playback your song. With that in
mind, here are some tips we’ve picked up over the years for learning to master on studio
monitors:
1) Listen to music that you know well and have listened to on many systems. Spend
some time “getting to know” your monitors. Play your favorite CDs through them. You
probably know how these CDs sound on a home system, a car radio, etc. and this will
help you learn to adjust your listening for your monitors.
5
Not entirely randomly, as we lik e STP and the CD was nearby. But there’s no scien tific reason.
2) The bass will typically be under-represented on small studio monitors.
3) Monitors are very focused in terms of their soundfield, and the imaging is typically
more pronounced than on other systems.
Headphones
Heaphones are another option for monitoring. There are entire sites and forums dedicated to
headphones (such as http://headroom.headphone.com) so again we’ll leave our hardware
recommendations out of it and just advise you to ask around on forums.
When working with headphones, here are a few things to keep in mind.
1) Bass is sometimes under-represented on headphones, since bass on loudspeakers is
often perceived from physical vibrations (what you feel) as well as from the acoustics
(what you hear)
2) Imaging on headphones is very different than imaging on speakers.
3) Equalization can be very different on headphones compared to loudspeakers. The
listening room, your head and even your outer ear have filtering properties that alter
the frequency response of the music. This “natural equalization” is bypassed when you
listen on headphones. If you’re interested in learning more about this phenomenon,
look into “diffuse field” headphones.
Before you jump into a marathon mastering session, here are seven things that are good to
remind yourself of periodically.
1) Have someone else master your mixes for you. OK, in most project studios we realize
that the same person is often the performer, producer, mixer, and mastering engineer. At
least get someone else to listen with you. Or find someone who will master your mixes if you
master theirs. You’re too close to your own music. You’ll hear things other listeners won’t
hear, and you’ll miss things that everyone else does hear.
2) Take breaks and listen to other CDs in between. Refresh your ears in terms of what other
stuff sounds like. OK, the pros just instinctively know what sound they’re working towards, but
for the rest of us being reminded from time to time during the process isn’t such a bad idea.
3) Move your listening position. Studio reference monitors are very focused and directional.
The sound can change significantly depending on your listening position. Shift around a bit.
Stand across the room for a moment.
4) Listen on other speakers and systems. Burn a CD with a few different variations and play it
on your home stereo system, or drive around and listen to it in your car. Don’t obsess over
the specific differences, but just remind yourself what other systems sound like.
5) Check how it sounds in mono. Check how it sounds with the polarity
inverted on one speaker. People will listen to it this way (although
maybe not intentionally) and while your master probably won’t sound
great this way hopefully it won’t completely fall apart either. Ozone
provides a quick check for this by clicking on the Channel Ops button.
You can quickly switch to mono, switch left and right speakers, and flip
the polarity of speakers.
6) Monitor at normal volumes, but periodically check it at a higher
volume. When you listen at low to medium volumes, you tend to hear
more midrange (where the ear is most sensitive) and less of the lows
and highs. This is related to something called the Fletcher-Munson
effect, which involves how different frequencies are heard differently
depending on the playback volume. So check from time to time how it
sounds at different volume levels.
7) When you think you’re done, go to bed, and listen again the next morning.
A reasonable starting point when mastering is equalization. While most people understand
how equalizers work and what they can do, it’s not always easy to balance a mix with one.
What’s the Goal of EQ when Mastering?
When we’re trying to get our mixes to sound good, what we’re shooting for is a “tonal
balance”. Any instrument specific equalization has hopefully been done during arranging and
mixdown, so we’re just trying to shape the overall sound into something that sounds natural.
Sometimes that’s easier said than done, but there are some general techniques you can use to
get a decent tonal balance.
EQ Principles
Here’s a basic review of the principles of equalizers before jumping into the process.
There are many different types of equalizers, but they are all meant to boost or cut specific
frequencies or ranges of frequencies. Our focus here is on parametric equalizers, which
provide the greatest level of control for each band.
Parametric EQs are typically made up of several bands. A band of EQ is a single filter. You can
use each band to boost or cut frequencies within the range of the band. By combining bands,
you can create a practically infinite number of equalization shapes.
The picture below shows the equalizer screen in Ozone, but the principles are the same for
most parametric EQs. There are 8 sets of arrows, which represent 8 bands of equalization.
One band is selected, and has been dragged down to cut the frequencies in the range of 3753
Hz by –3.5 dB.
The bright red curve shows the composite or overall effect of all the bands combined. The
darker red curve shows the effect of the single band that’s selected.
Each band of parametric equalization typically has three controls:
Frequency
The center frequency dictates where the center of the band is placed.
Q and/or Bandwidth
Q represents the width of the band, or what range of frequencies will be affected by the band.
A band with a high Q will affect a narrow band of frequencies, where a band with a low Q will
affect a broad range of frequencies.
A Narrow Filter (Q=12 )
A Broad Filter (Q=0.3)
Q and bandwidth are related by the formula Q=(filter center frequency)/(filter bandwidth). So
as Q gets higher, the bandwidth of the filter gets narrower.
Gain
This determines how much each band boosts (turns up) or cuts (turns down) the sound at its
center frequency.
Using the Ozone Paragraphic Equalizer
Ozone includes a parametric equalizer presented in a graphical way, which is often referred to
as a paragraphic equalizer.
The paragraphic equalizer has 8 adjustable filter bands which can be used to boost or cut
frequencies. To adjust the gain of a band, you grab the center and move up or down. To
adjust the frequency, you drag left or right.
To adjust the Q or width of a band, you can grab the side handles of
the band and drag them apart.
Any of the eight filters in Ozone can be configured to be a bell (also referred to as a peak
filter), lowpass, highpass, lowshelf or highshelf. You can specify the shape of a filter by
clicking on the “Show Info” button and selecting a different shape for the filter from the table.
Bell Filter
As shown below, a bell filter has a width (Q) as well as a gain. The gain can be positive or
negative – to boost or cut the specified range of frequencies within the bell.
Lowpass and Highpass Filters
Unlike a Bell Filter, Lowpass and Highpass filters only have one “side” to them. You specify the
point that you want to start attenuating frequencies, and any frequencies below that point (for
a highpass filter) or above that point (for a lowpass filter) are filtered. The Q control for
lowpass/highpass filters specifies the slope of the filter, with lower Q values providing more
gradual rolloff.
Lowshelf and Highshelf filters
Like lowpass/highpass filters, these filters also are “one sided”. Shelf filters, however, don’t
drop off indefinitely. Instead, they resemble, well, a shelf, as you can see below. In this case,
the horizontal handles provide a slope control which specifies how tall the shelf should be – or
how much cut should be applied before leveling off to a constant (horizontal) line.
Controls for Adjusting EQ Bands
In addition to basic mouse support, Ozone supports the following controls for adjusting EQ
bands:
1) You can use the arrow keys to adjust a band up/down or left/right. If you hold down
the Shift key when using the arrow keys the adjustment is accelerated.
2) You can adjust the Q of a band by using the wheel of a wheel mouse or the
PgUp/PgDn keys.
3) You can select multiple bands by holding down the Ctrl
key and clicking multiple bands. To adjust them as a
group, drag the first band selected and the rest will
move with appropriate relative motion (or use arrow
keys to move the entire group). This is useful if you
have an overall shape that you like but want to raise
or lower the gain of the entire curve.
4) If you hold down the Alt key and click on the spectrum, you have an audio magnifying
glass that lets you hear only the frequencies that are under the mouse cursor, without
affecting your actual EQ settings. This is useful for pinpointing the location of a
frequency in the mix without messing up your actual EQ bands. Releasing the mouse
button returns the sound to the actual EQ. You can set the width of this filter in the
Options dialog.
5) If you hold down the Shift key and drag an EQ band, the EQ band will be "locked" in
the direction that you're dragging. So if you just want to change the gain without
affecting the frequency (or vice versa) just hold the Shift Key while you drag.
6) If you'd rather use numbers as
opposed to visual EQ bands,
clicking on the Show Info button
gives you a table view of the EQ
band settings. You can enter
values for the EQ bands directly in
this table, or simply position the
cursor over a value and change it by turning the wheel of a wheel mouse. You can also
disable bands with this table by clicking on the square box to the left of a band.
7) You can select the shape of a filter by holding down the Ctrl key and right clicking on
the EQ filter. This will cycle it through the lowpass, highpass, bell, lowshelf, highshelf
shape options.
So you’re ready to EQ. Now what?
Listen and try to identify any problems that you hear. Start with the midrange (vocals, guitar,
midrange keyboard, etc.) as this will typically represent the heart and soul of the song. Does it
sound too “muddy”? Too nasal? Too harsh? Compare it to another mix, perhaps a commercial
CD. Try to describe to yourself what the difference is between the two mixes around the
midrange.
Too muddy?
Try cutting between 100 to 300 Hz (Band 2 in Ozone is set at 180 Hz by default. Try cutting
the gain a few dB using this band)
Too nasal sounding?
Try cutting between 250 to 1000 Hz. (Band 3 in Ozone is set by default at 520 Hz for this
purpose)
Too harsh sounding?
This can be caused by frequencies in the range of 1000 to 3000 Hz. Try cutting this range a
few dB. (Band 4 in Ozone is set at 1820 Hz for this purpose)
Hopefully, using a band or two in these regions will give you a better sounding midrange.
Remember that you can use the Alt-click feature to focus just on specific ranges and highlight
what you’re hearing. Another common technique is to start by boosting a band to highlight a
region of the spectrum, and then cutting it once you’ve centered on the problem area.
You’ll get the most natural sound using wide bands (Q less than 1.0). If you find yourself using
too narrow a notch filter, or too much gain, you may be trying to fix something that EQ on a
stereo mix can’t fix. Go back to the individual tracks and try to isolate the problem that way.
Note also that the wider the band, in general the less gain you need to apply.
In addition, your ears quickly get used to EQ changes.
You may find yourself boosting more than necessary to
hear the difference. Use the History window (click on
the History button) to go back and audition settings
prior to making changes. Comparing the difference
before and after a series of subtle EQ changes can help
prevent you from overdoing boosts or cuts.
In comparing your mix to commercial mixes at this stage, you’re probably tempted to boost
the bass using the equalizer. Resist the temptation. Don’t worry, your mix will get that low
end punch, but we’ll do it using a multiband compressor.
A reasonable use of EQ in the low end is to apply a shelf or
highpass filter below 30-40 Hz. Purists might find this alarming, as
yes, we can hear down to 20 Hz and some musical information can
be lost. Typically what people consider “bass” though is in the 50100 Hz region, and the audio in the 20-40 Hz range can usually be
rolled off. The benefit is that you can remove some low frequency
rumble and noise that could otherwise overload your levels.
Keep in mind that for bass, or any EQ change for that matter, every action has an opposite
reaction. If you increase one frequency, you can mask another frequency. The flipside of this
is that cutting one frequency can be perceived as a boost to another frequency. Each change
that you make can affect the perception of the overall tonal balance of a whole.
Bass guitars and kick drums can span a wide frequency range. Where the “oomph” of the kick
drum can be centered around 100 Hz, the attack is usually found in the 1000-3000 Hz region.
Sometimes you can get a sharper sounding “bass” sound by focusing on the higher frequency
attack, as opposed to the 100 Hz region which can cause “mud”.
On the other hand, if you want to add
that hip-hop style “ring” to the bass, try
a peak at 50-60 Hz as shown to the
right.