Digital Recording, Mixing and Mastering.

Volume 2
..--+=({ethix-wing})=+--..

What do I do first and in what order do I work?

The first thing I do is to listen to the music carefully listening for noise, hum and other deficiencies that may have
been overlooked by the recording engineer. This is not a put down of the mix engineer skills but an acknowledgment
that a second opinion will often pick up things that may not have been heard previously. In most cases the problems
lie with noise, equalization and arrangement. The next thing I do after listening to a track carefully is to apply noise
reduction. In many cases noise reduction alone will make the track sound much better bringing out frequencies that
were hidden or masked by the noise adding sharpness and clarity. Next I make EQ and compression adjustments
bearing in mind that both these processes will boost the signal somewhat. Low and high frequencies often need
clarification (isolating then boosting or reducing) either being too excited or to dull. Next I add compression which
adds further definition to the highs, mids and lows while adding punch and a smooth texture to the overall sound. If
you're going to use compression now is the right time in that compression will squelch all other processes you may
have already done (spatial enhancements, reverb, etc.). Next I add spatially enhancements and for that I use
plug-ins that widened the stereo field adding full rich sound to the track. Stereo fields can be widened or reduced
shifted left or right to suit your need. Bear in mind that when the stereo field is widened so is the reverb so use
caution when adding reverb of or wait and add it after this process is done. Next I add effects like reverb or delay if
needed. Finally I use a read ahead limiter to increase the overall volume and impact of the track. This is just a
general outline of how I work. In many cases some of these steps are skipped, not being necessary for the
particular track I'm working on.
Q: WHy is it that I can burn a cd at a relatively hot level(hitting 0 db all the way through), yet when I listen to
pro-recordings such as Tori Amos, Lenny Kravitz, etc...the apparent volume seems almost twice as loud?

A:If your sound card is 16 bit, you probably are only getting 12-13 bit, whereas pro's can maintain 16 bit because
they are working in 20 or 24 or 32 bit. The rule of thumb I learned is there are 6dB to one bit, so do the math
(16-12)*6 = 24dB There is the volume you are missing.

The increase in loudness is created primarily by compresssion and limiting. Long story short, if you "rip" a loud
commercially produced CD and examine the waveform you'll see a big "block" of sound. What the mastering
engineer has done is to limit peaks on the song- spikes in the waveform that hit zero dB and are much higher than
the rest of the waveform; and also he likely compressed the mix as well- bringing lower parts to the same or similar
volume as higher parts. All this is done for the reason you mentioned- to increase the average volume of the
waveform and by association the apparent volume.

Doing the stats on album music was something else- music that I listen to over and over seems to hover around the
-14dB average RMS with peaks around -1 to -1.5 db rather than the -9db/lotsa OdB peaks which seem to be
common in singles.

Ballads and classical music are really interesting- a piano and voice piece at -19 dB av.RMS can sound as "loud" as
a pop piece at -14dB average and limiting it to "match" on paper can make it sound NASTY, like being slapped
upside the head with a giant fiberglass hotdog.
If you like the mix and you don't think anything needs
EQing like above, compression can help but the wrong
kind can dull the mix. If your mixes aren't normalized
to 99%or so you'll need to set your own threshold, but
-6 to -10dB is where you want the comp to start
working. A ratio of 2:1 is usually not desired unless
the mix isn't very tight, but it sounds like your mix
is tight if the instruments are in your face, a ratio
of 1.1:1 to 1.5:1 is where I like to start but each
situation is different.. If the compression sucks some highs out or
even mids, you may want to use the above compression
but with in a certain range up to 8k, or lower
depending on the mids-
Or maybe only 20-200 needs the comp.

I like to comp before final EQ unless you can hear

something you can fix with parametric. If you don't
para EQ first, but comp first, you may need to use
EQ to adjust any mismatches after comp.
When you're satisfied with the mix, in CEP, I use the
amplitude effect to see how much I need to boost to
achieve 100% normalization(maximum volume before
digital distortion) view this in decibels not in
percentage and it will give you an amount to increase
to maximum volume, but don't do it. Instead close it
out and goto the Hard limiting effect and where it
says boost input by:_ put in a number 2 to 6 dBs above
the amount the amplitude effect gave you depending on
how much dynamic range you want. Increasing by 6 will
have less dynamic range than 2.
These techniques work for me and they might for you.
Mastering

by Bud Brenner

1. COMPRESSION

Program compression is meant to be a subtle effect, with an average gain reduction of

only 2-3 dB, and usually with a gentle ratio/slope of 1.5:1 or 2.1. My Manley tube
compressor has only one ratio for compression and it is 1.5:1. The transient parts of a signal can be reduced
smoothly but it is the underlying constant level signals that you might hear being turned down and back again by the
attack and release settings chosen.
Correct attack and release times will differ depending on the tempo and style of the
material being compressed. If you have a very busy signal with lots of things going on in the mix, there are many
dynamic events that can trigger the threshold and allow the
compressor to operate with faster settings, specifically the release settings. Remember,
it’s the constant level signals you have to worry about because their note values are longer than the transients, and
under a quick attack and release times with more than 3 dB of gain reduction, they can be heard to decrease and
increase in level. This is the good old ‘Pumping & Breathing’. A good program compression setting will have no such
artifacts.

With slower tempo tunes such as ballads, we can assume that there may be less dynamic
events per second with even longer note values, and your compression settings,
specifically your release setting could/should be somewhat slower for this type of material. This results in a
smoother compression. Your attack times in all cases should be set to taste as you are trying to smooth out less
than perfect dynamic performances by the musicians. Just remember that the faster the attack times, the more
compression you’ll be applying and more compression means that you’ll have to look closely how these faster
attack times affect the other settings on your compressor.
You may find that you have to raise the threshold setting or slow down the release time if you use a faster attack
time.
‘For every action there is an equal and opposite reaction’. This first law of physics is alive and well in your own
compressor so experiment with different settings using these guidelines to help you along.
Sampling Rate

The number of times a sound waveform is checked for position each second is the sampling rate. The sampling rate
is similar to the frame rate in movies. As you can imagine, with higher sampling rates you store more information
about the sound’s changing amplitude. This gives you more fidelity.
As a matter of fact, it is impossible to accurately record frequencies above one-half of the sampling rate. This
threshold frequency is called the Nyquist frequency, and should be considered when selecting a sampling rate.
Frequencies higher than the Nyquist frequency show up as alias noise.

The downside to very high sampling rates is that since each sample takes up space in memory (1 byte for 8-bit
samples, 2 bytes for 16-bit samples), higher sampling rates will fill up your hard drive faster than lower sampling
rates. For instance, a stereo digitized sound of 44,100 Hz 16-bit data (approximately what your CD player uses)
lasting 10 seconds takes up almost 2 megabytes of space! This means if you have a 40 megabyte hard drive you
couldn't even store 4 minutes of sound data, and that's without having any programs or other data on your system.

Expanding the Model

Instead of using just the two-position model with the in and out positions, let’s suppose we had a system that
allowed us to have 100 positions or even more. If this were the case then we would be able to make much more
complex sounds. We could move it just a little bit or we could move it all the way out. This would allow us to have
more precise control of the amplitude of the waveform. The more positions we have, the more flexibility we have in
producing sound. For example, if we were to represent amplitude as a number from one to four, any values that fell
between would be rounded to the closest value. This rounding error is called quantization noise. When more
positions are available, rounding errors become smaller.

You will often see a sound card referred to as 8-bit, or 16-bit. We can directly relate this to the number of positions
in which we can place the speaker. With an 8-bit card we can place it in 256 different positions and with a 16-bit
card we can place it in any of 65,536 positions. Although you might think that a 16-bit card should have twice as
many positions as an 8-bit card, this is not the case. It actually has 256 times as many positions. Even though
16-bit samples take up twice as much space as 8-bit samples, it is recommended that when at all possible you use
16-bit samples to minimize quantization noise.

At what level, then, should a signal be recorded digitally? The standard method for digital metering is to use the
maximum possible sample amplitude as a reference point. This value (32768) is referred to as 0 decibels, or 0 dB.
Decibels are used to represent fractions logarithmically. In this case, the fraction is: sample amplitude divided by the
maximum possible amplitude. The actual equation used to convert to decibels is: dB = 20 log (amplitude/32768)

Now, let’s get back to the real question: at what level should audio be digitized? If you know what the very loudest
section of the audio is in advance, you can set your record levels so that the peak is as close to 0 dB as possible
and you’ll have maximized the dynamic range of the digital medium. However, in most cases you don’t know in
advance what the loudest level will be, so you should give yourself at least 3 to 6 dB of headroom for unexpected
peaks (more when recording your easily over-excited drummer friend).

DRUM LEVEL
When starting a mix, set the drum levels to approximately -6B. That way, when everthing else goes on top, the
meters will not pin."
Steve Negus - musician/producer/engineer.
I can see no reason why all the pre-master tracks should not be peaking at 0db (or slightly under) as long as
compression hasn't been used to take it there.

If it peaks at less than -3dB or higher, chances are good the mix engineer did that purposely to avoid a clip. Not
necessarily bad but with diligence it IS possible to get the peaks closer to 0dB- I often get up to -.5dB or -1dB when
i'm trying. Remember you loose resolution when the CD volume is low. I often mix down a couple dB when i'm trying
to make a rough mix or save time- when I wouldn't want to do another pass of the mix because the levels clipped.

I should make a point about normalization - it is NOT the way to get higher levels. When you normalize something,
you're also bringing up the level of the noise floor - a very bad thing. Your mixes should be as hot as possible,
without peaking and without too much compression to keep your noise floor down to a minimum. Look at it this way
- if your peak level is at 70% and your noise floor is at 20%, when you normalize you're adding 30% to BOTH those
numbers (assuming you're normalizing to 100%). I would also NOT normalize to 100% - go to 99% or 99.5%. This
gives room for additional gain, should it be needed. Normalizing shouldn't even be used, except in the mastering
stage, if at all avoidable. It is a common misconception that normalizing is a 'magic fix' and it's completely untrue. I
see as much abuse of normalizing as I do of compression.
Shielded cable strangles your sound. If I had a few hours to go into it I'ld give a lecture on the virtues of various cable
types, construction and wire type. NS-10M's are not the worlds best sounding speaker, bright, no true low end
extenstion, lack of coherence, I could go on for days about what's wrong with them, but let's discuss how to make
them sound their best. First set them right below ear level on their side with the tweeters on the outside. Use a solid
mounting surface and secure the speakers with blu-tac or some other form of removable strong adhesive. If they are
going to be mounted on stands, fill the hollows of the stands witha sand lead shot mixture (50/50). try to place them
about 3-6 feet (about 2m for all you UK folks) from your head and from each other, angled toward you. Now for your
original question, speaker cable. To help smooth the harshness of the high end and to tame the bloated low mid
response, I would say use a high quality solid core, or better yet completly solid, twisted pair or quad, solid copper
cable. Kimber, or if you can swing the big bucks, Vector (if you can find it) are my personal favorites. I've been using
Vector CV-4 cable for the last 10 years and it's simply stunning(1 meter cost more than your NS-10M's). On my
computer I use Kimber Quad (older style), the price on this cable is very reasonable, and it will tame the most unruly
speaker.
Our guide to buying, setting up and using monitors...

With one speaker on top of the wardrobe and the other behind the sofa, it's no surprise your mixes sound odd. Don't
panic, let pro engineer Mick Williams guide you through buying, setting up and using monitors...

Some people say the only valid excuse for a bad mix is bad monitoring. While the level of expertise of the person
mixing is an obvious contributing factor, even experienced engineers would be hard-pressed to do their optimum
work when faced with a monitoring system that doesn't accurately reproduce the frequency spectrum. In a nutshell,
if you can't hear it properly you can't mix it properly, so accurate and effective monitoring is essential for any studio.

But the complexities don't stop there. Having decided to buy some studio monitors, which do you choose? The point
of having good quality monitors is so you can hear your music accurately enough to mix it so it will sound good on
whatever system it's played back on. Since music can be played back over various different speaker systems, from
large club sound systems to domestic hi-fi equipment or pocket-sized transistor radios, what you need are monitors
that can accurately represent these myriad playback systems.

However, there is no such thing as a standardised monitor, as all monitors sound different. Even two sets of the
same type of monitor can sound different from each other in different rooms or when being driven by different
amplifiers, so it's a case of buying a decent set of monitors and getting to know and trust them.

The pros of a pro

Professional studios usually tackle mixing by having several sets of monitors to switch between. There will be large
monitors, usually soffit mounted (that's fixed in the wall to you and me) which represent the full sound spectrum
including the bass end. And there will be smaller nearfield monitors placed closer to the mixing position which, due
to the limitations of cabinet size, have a more limited bass response. Nearfield monitors simulate playback
conditions comparable with your home listening environment, reproducing the quality of sound played back on, say,
a standard hi-fi system.

As few of us have the budget or space for huge monitors, it's nearfields that must be the speakers of choice. So, the
first question must be, if nearfields are meant to sound like domestic hi-fi speakers, why not use your existing hi-fi
speakers and save some money? The truth is, hi-fi speakers are often deliberately designed with 'colouration' that
flatters the music, rather than reproducing it, warts 'n' all. Nearfield monitors reproduce the entire frequency range as
accurately as possible with a minimum of distortion and colouration, so they're much truer to the real sound.

If you mix music on hi-fi speakers tuned to make the bass sound louder, you might not add enough bass to your
mix, so it'll sound lightweight and lacking in bass when played on other systems. Studio monitors are also built
more robustly to take higher sound levels; useful when you want to solo a particularly raucous sound at high volume.

Passive and active

When looking for monitors, you have a choice between passive and active systems. Passive speakers need a
separate power amplifier whereas active speakers have the amp (or amps in the case of bi-amped systems where
the tweeter and bass drivers have separate amps) built into the speaker enclosure. Active speakers mean you can't
choose your own power amp, but their built-in amps are specifically designed to work with their speakers, creating
an efficient, matched system.

Most nearfield monitors have little bass reproduction below, say, 80Hz, so you won't hear the real low end. Still, if
you need to hear these frequencies, the bass response can be extended with the addition of a sub bass unit, which
can sit out of the way under your mixing desk.

Monitor placement
When it comes to setting the position of your monitors, following a few basic rules will result in an accurate stereo
image and reproduction of the frequency spectrum from your mixing position. Firstly, both speakers should be at the
same level, and they should preferably be placed on a level with your head, when you're at your favoured mixing
position, with the tweeters around ear height. It's not always physically possible to place speakers at a height level
with your head so it's quite acceptable to mount them higher up, but in this case tilt them down so they're pointing at
your head.

Secondly, the speakers should be angled slightly towards your listening position so the sound focuses towards your
head. The recommended textbook starting position is usually to have the speakers positioned to subtend an angle of
60 degrees to the listener. Basically, you sit at the apex of an equilateral triangle formed by yourself and the
speakers; this is the 'sweet spot' where you'll find the most accurate representation of the sound (see diagram
above).

It's usually better to mount monitors vertically so the sound from the tweeter and the bass driver arrives at the ear at
the same time, although some monitors, such as the Yamaha NS10Ms, are designed to be placed horizontally.
Also, the distance between the speakers shouldn't be more than about two metres or the central stereo image may
suffer. You could also run into problems if the distance between the two speakers is greater than the distance
between the speakers and the listener.

Position in the room

Unless you monitor solely on headphones, it's a fact of life that the room you're in will affect the sound you hear. The
size and shape of the room, together with the materials on the walls, ceiling and floor, can all exert an influence on
the sound, as can any objects in the room. Sound from the speakers will be reflected from and absorbed by the
various surfaces and objects which can result in distinct echoes, reverb and certain frequencies being cancelled or
reinforced.

All these things, if they present a problem, can be tackled by acoustic treatment such as bass traps or heavy fabric
draped on the walls, but any room influences can also be minimised by using nearfield speakers positioned
correctly. Nearfield speakers tend to reduce any room effects, as they are closer to the listener, so the direct sound
from the speakers dominates rather than any reflected sound.

Whatever speakers you use though, it's always useful to minimise the effects of reflected sound as much as you
can. Symmetrical positioning of the speakers in relation to the room is important. If the distance between the
speakers and their adjacent walls is not identical on both left and right then any reflections from the walls will be
different and may disrupt the stereo image.

By the same token, any nearby racks of gear could cause reflections, so, if possible, try to arrange the two monitors
on both sides of your mixing position rather than on just one. Reflections can also come from the surface of your
mixing desk, but placing your speakers on stands behind the desk rather than sitting them over the meter bridge can
minimise this problem.

Monitoring tips
Now your monitors are nicely set up, here are a few practical tips to help your mixing. The first one is don't monitor
too loud for extended periods. Protracted listening at high volume can not only cause permanent ear damage but it
can also wear out your concentration more quickly and dull your perception of top-end frequencies.

There is always a temptation to turn things up because, let's face it, music usually sounds more exciting that way,
but you may soon get immune to the constant high level. It's far better to monitor at a reasonably low level and just
turn it up occasionally for a quick high volume check. Monitoring at different volume levels is good practice anyway,
as turning the level right down allows you to hear if things are jumping out of the mix. It's also a good way to check
if, for example, the vocal or snare is too loud. Also - and this may seem a bit strange - listening to the track while
standing outside the room gives you a different perspective that may prove useful. Try it and see.

As there is no such thing as a standard monitor speaker, each speaker design provides its own version of the truth,
so it stands to reason that, to get the best results, you need to know your own speakers inside out and to trust what
they're telling you. The easiest way to get familiar with them is to play your favourite CDs through your monitors,
both in isolation and while mixing your own tracks, and compare the sound. Presumably you'll have some music in
your collection mixed in a professional studio on a top-class monitoring system, so comparing this to your mixes in
progress, checking not only the overall sound but also specific areas, will do no harm. I'm not talking about making
slavish copies here, but it will help check things like if there is enough top end, if the bass is too boomy, if the mid
range sounds too harsh, if you've added enough reverb, if the vocal sits well with the music and if the drums are too
loud.

If you have access to several sets of speakers so much the better. Switch between them from time to time to see
how the music sounds on each set and occasionally check how things are sounding on headphones. If you have just
one set of studio monitors you can always run off mixes in progress every so often to play back on a ghetto blaster,
domestic hi-fi system, car stereo or personal stereo. If you can get your mix to sound good on really crap speakers
as well as decent ones, then you must be doing something right.

Mick Williams