Web Mastering: The Art of Optimizing Sound Files for
Napster and the Internet
by Josh Beggs
12/18/2000
Remember the true Internet rallying cry: no spectators--only participants! If
you're ready to contribute content to the Web by posting your original music on
Napster or on your own Web site, this article will show you the audio
production techniques that professionals use to make their audio sizzle.
We'll have you posting high-quality sound files in no time. Before we begin, I
want to say that this article assumes you've already recorded your music with
good input levels as close to 0dB (zero decibels) as possible; that the file
is in digital form on your computer; and that you know what Web file format
you are converting it to (MP3,
RealAudio, Windows Media, etc.).
Audio files need to be custom-tailored for their respective output
mediums: radio, film, television, home stereos, and, of course, the
Web. There are several key steps in between digitizing an
audio file and encoding it for the Web that determine the quality of
the end-user's listening experience. These steps are referred to as
Web mastering.
The general term "mastering" refers to the process of preparing a
group of sound files (i.e. music tracks of a CD) for final delivery
in one uniform manner, medium, and level of quality. This means that
each sound file has little unwanted noise (artifacts) and shares
the same relative decibel (volume) levels and equalization
characteristics as those in the rest of the audio files.
Why Mastering?
Mastering is similar to doing Q & A (Quality Assurance) on your Web site before
it goes live. It's not the most fun part of the job, but it's one of the most
important since a misspell or an HTML browser display error makes a site look
unprofessional. You may spend 100 hours designing a great site, but if you
don't spend any time optimizing graphics for quick download, or checking your
site on different browser platforms, all that hard work may be for naught.
People will only notice the error or how long it took to download the page.
Similarly, you'll want to perform the final audio production step and master
your audio files to eliminate errors and noise artifacts, create consistent
volume levels across all sound clips, and optimize equalization levels to
maximize the playback quality of your soundtrack.
The practical reason for mastering groups of sound files is to prevent
people from having to continually adjust the volume and equalization
settings of their sound systems or computers each time a new song or
sound passage is played. For example, it is generally undesirable
to have one song where the bass guitar frequencies stand out in the
mix and another song where the bass frequencies are relatively
quiet. Nor would you want to have one voice-over track significantly
louder than the next voice-over track. Mastering helps to balance each
song and sound file so the mix or Web soundtrack is consistent.
Mastering also disguises the sonic variation between songs that may
have been recorded in different environments or recording studios.
The term "Web mastering" refers to the final optimization process before
converting sound files to their respective Web formats. The main Web mastering
tools you'll work with are: equalization (EQ), normalization, and
compression. Equalization ensures the tonal consistency and quality between all the sound files that will be incorporated into your Web site. Achieved with a
sound editor, normalization helps guarantee that your files share the same
relative loudness and optimal dynamic range for Web conversion. Compression of
the sound file with a compressor or peak-limiter effect ensures the highest
fidelity when converting to low bandwidth formats with reduced bit-depths and
sampling rates.
For best results you should have a solid understanding of how audio
information is digitized. In particular, you want to be familiar with sampling
rates and bit-depth, terms that many find confusing and difficult to grasp.
Web Mastering Tutorial
So now that you have the basic concept, let's jump right into the Web mastering
process. First, let's assume you've already recorded an audio file onto your
computer with a sound-editing application. If you don't have a sound editor,
go to
CoolEdit.com and download
Cool Edit 2000, for a small fee. Bang for buck, Cool
Edit1 is probably the best deal
going and will work great until you're ready to try higher-end,
professional-grade audio applications, such as
Pro
Tools or Sonic Solutions.
That said, let's take a look at our real-life scenario. You've heard about
Napster's 20 million users and may have thought, "Hey it's time I put
my music on the service and maybe even post some music files on my own Web
site." Perhaps you want to post2 a
great live recording of your band or put up a copy of some of the hot mixes
you were spinning over the weekend. So, what do you do to transform your
original digital-audio source file into Web-ready, professional-sounding MP3
files?
Let's begin. As part of the Web mastering process you will want to
do some basic clean-up editing on your sound files before moving to
the final equalization, normalization, and compression steps.
Basic Sound Editing
If you have a digitized file of a live performance an hour or two in length,
separate the file into shorter audio clips by song. Once you have five or ten
separate audio clips, sort through each clip and perform some basic clean-up
editing to remove unwanted artifacts, such as coughs, sneezes, loud inhales
on a vocal mic before a song starts, a clank against the mic stand,
and so on. After some basic clean up you can add fade-ins and fade-outs
for each clip. In general, there are no set rules about how long
fade-ins and fade-outs should last at the beginning and end of a
tune. Try several options and use your ear to determine the timing
of the fade-ins and fade-outs that works best with each song.
Equalization: Optimizing Sound Frequencies
Equalization is one of the most powerful tools you can use to enhance
audio quality. Through the process of equalization (EQ) you can amplify
or reduce selected frequencies or groups of frequencies that comprise
a given sound file. Equalizing files destined for the Web helps
make up for lost or distorted frequencies.
During the mastering process, equalization can enhance your audio and
compensate for unbalanced sound levels in three different ways. First,
equalization can be used to adjust the overall bass, midrange, or
treble of a sound file. Second, it can make the overall bass, midrange, and
treble more uniform from song to song or from sound file to sound file. Third,
it can eliminate certain frequencies that cause distortion, or, conversely, to
boost frequencies that get reduced when your audio file is encoded for the
Web.
During the audio encoding process for analog modem delivery, your
sound files go through drastic changes as bits of data are thrown out
and compressed to reduce file sizes. This process eliminates certain
frequencies and accentuates background noises that you may not have
noticed in the original audio source. Generally, the two areas to focus
on are the highest ¼ of your audio frequencies and the lowest
¼ of your audio frequencies. The higher frequencies (above
10,000Hz) tend to carry a lot of background noise, such as hiss and
scratchiness in an audio file. Whereas the lower audio frequencies
(below 75Hz) tend to carry a lot of the rumble that can occur from
background noises, such as H/VAC systems or microphone rumble caused
by wind.
To correct this problem, eliminate the low-end frequencies between
20Hz and 75Hz and the highest frequencies above 10,000Hz with the equalization
faders in your sound editor. Then boost the low-end bass frequencies between
75Hz and 110Hz to compensate for the elimination of the bass sounds below
75Hz. It's also a good idea to boost the mid-range vocal frequencies between
2,000Hz and 7,000Hz to add brightness to a mix and make the vocals clearer
after Web conversion, as shown in Figure 1.
Figure 1. Equalization boost of the 75Hz to 200Hz range
and 2,000Hz to 7,000Hz range
Normalization: Setting Uniform Volume Levels
After you have adjusted the EQ of your audio you will want to normalize your
sound files before encoding for the Web. There are three reasons to normalize
your sound files. First, normalization maximizes the bit-depth information
available by giving a sound a louder overall presence. Second, it provides
volume consistency across multiple Web pages or sound files. There's nothing
quite as jarring as experiencing dramatic shifts in volume when listening to
one song after the next. Normalization corrects this problem by raising all
sounds to one uniform standard amplification level, such as 95 percent
or -3dB below 0.
When encoding for the Web, it's a good idea to set the amount of
normalization to a number below 100 percent, such as to 95 percent or -3dB
below 0 to allow for extra headroom. Flash and RealAudio, for example, have a
tendency to distort certain sound files that are normalized between 95 percent
and 100 percent.
The third reason for normalization is to prevent distortion when encoding a
stereo file into a single-channel mono file. If you are converting stereo
files into low-bandwidth mono files you should normalize to roughly 70 percent
the signal level of your input source. If you do not normalize a stereo file
during or prior to its conversion to a mono file you may experience digital
distortion in your output file when the signals are combined and shoot above
the maximum 0dB level.
Normalization raises or lowers the overall amplitude or loudness level of a
sound file to a selected point, generally up to where the loudest amplitude
peak in the file rests just below the 0dB clipping level, as shown in Figure
2. The clipping level is the cut-off point where the audio signal information
can be accurately digitized. Audio information beyond the OdB cut-off point
becomes severely distorted and results in unpleasant noise. Normalization
makes maximum use of the dynamic range within the bit-depth of a digital audio
signal, but does not effect the relative dynamic range of the source audio.
In other words, the dynamic range between softest and loudest material in the
sound file is unaffected after normalizing, but the sound file is louder
overall.
Figure 2. The top audio file represents the audio
waveform before normalization and the bottom waveform represents the audio file
after normalization (set to 95 percent or -3dB below 0)
Generally, all sound files have a few short volume peaks or spikes. They
usually result from a loud guitar solo section or a particular voice passage
that's louder than adjacent sounds (see Figure 4). Simple normalization will
not be effective if there is already a peak just under the maximum 0dB cut-off
point. A peak limiter or compressor is needed to reduce the highest volume
peaks and further raise the overall loudness of the file.
If you don't have a peak limiter feature or special plug-in, the loudness of
your sound file can be raised several dB beyond the regular normalization
point by selecting a region in the sound file, then using the Amplify/Gain
command in Cool Edit or SoundEdit 16 to manually reduce the spike. You can
find spikes by visually scanning the sound file.
We are aware that some editors feature a command that automatically identifies
the loudest amplitude peak in the file. Once you have reduced the spikes by
-3dB to -9dB, normalize the file again.
Dynamic Range Compression: Reducing Sound Peaks and Spikes
Now that you have brought your sound files up to a consistent volume, it's
time to further optimize the dynamic range of your soon-to-be-encoded audio
clips. The dynamic range of an audio track is the difference in volume between
the tracks loudest and quietest points. When encoding, audio that has either
too strong or too weak of a signal will have a negative impact on the output
quality of the particular file. Some content, especially film content, has
extreme dynamic ranges that are utilized in order to help accentuate the
emotion created by the content. Unfortunately, this dynamic causes negative
results when encoded, causing previously unheard noise in the quiet parts and
distortion in the loud points. I suggest using some mild compression in order
to help the encoder best deal with the source audio.
Also, using a compressor or "look-ahead" peak limiter, such as the L1
Maximizer plug-in from Waves,
while normalizing your sound files will produce even better results than simple
normalization alone, as shown in Figure 3.
Figure 3. A standard normalized waveform (top) and a
compressed and normalized waveform (bottom)
Compressors and peak limiters reduce the highest peaks or waveforms in your
sound files, thereby allowing you to further boost the volume of softer sounds
(see Figure 4). Increasing the volume of the waveforms within a sound file
allows you to maximize the dynamic range when converting to low bit-rate Web
codecs. Generally, 3dB to 9dB of compression adds clarity and brilliance to
the sound and produces an overall effect that is pleasing to the ear. However,
be careful not to overcompress your sound files. Too much compression
(more than 12dBs) may make your sound files sound unnatural and harsh to your
listeners' ears.
Figure 4. A sound file with a sudden amplitude spike
going to the 0dB cut-off level that should be reduced in amplitude
Now that your sound clips are nicely compressed and normalized, have the
optimum amount of bass and treble, and are free of noise artifacts, you're
ready to post your sound files on the Web. After following these Web
mastering steps, your music might just sound as good as or better than that
pirated Metallica song on Napster.
Batch Conversion Tools
Just in case the steps in this article seem too time-consuming, there is
another way: batch conversion. Let's say you've still got another 20 tracks
to convert from your band's live recording archives. Or maybe your Web
division boss comes to you and says you have to post all of the company's
film and audio archives from the last five years on the Web.
Batch conversion tools let you simultaneously import multiple audio files,
set equalization, normalization, and compression levels, and encode the entire
batch of audio files into whatever Web format you desire, such as RealAudio
or MP3, with one click of the button. Of course, you will need to spend some
time setting up and testing the parameters and preferences for the
batch-encoding process, but once you do, you can save your session parameters
in the batch-conversion application for the next group of files you need to
convert.
Batch conversion tools save time and prevent human errors by automating the
mastering and encoding process for multiple audio files. Two of the most
popular tools for mastering and batch conversion are Wave Convert Pro and
BarbaBatch. Both applications convert multiple audio files into various
sample rates, word lengths, channels (stereo/mono), and file types (MP3, AIFF,
SDII, .WAV, .RA, .SWA, and QuickTime), while retaining optimal sound quality.
Wave Convert Pro runs native on Macintosh and is available for Windows 98 at
the Waves Web site. BarbaBatch
also runs on Macintosh and can be purchased at
audioease.com.
Josh Beggs is cofounder and president of
Raspberry Media,
a Design Firm in the San Francisco Bay Area specializing in Web-smart
architecture, interface design, and brand development for Internet
start-ups. Josh began his career in the multimedia industry as a
recording engineer and sound designer. He produced the interactive soundtrack
for EMI Records flagship CD-ROM, Queensrÿche's
Promised Land. After receiving impressive reviews from Billboard
Magazine for the soundtrack, Josh went on to explore interactive media design
with Raspberry Media. In addition to designing some of the top
Web sites on the Internet, he also follows his musical passions as a pianist
and recording artist.
- Sound Editors: You can perform basic sound editing with any digital
audio application that has routine cut, paste, and fade features such
as Cool Edit 2000 (Windows),
Sound Forge by Sonic Foundry (Windows),
SoundEdit
16 by Macromedia, and
Pro
Tools by Avid (Mac and Windows NT).
- Professional Encoding Services: Have a few more files left from
your bands last live recording that you need to post? Wondering how you will
manage your company's entire archive of audio and video content? What do you do when you want to post hundreds or even thousands of audio and video files?
Don't worry, there are a few people in the industry you can turn to. Loudeye
is one of them. Loudeye,
formerly encoding.com, specializes in helping companies manage, encode,
and stream large quantities of audio and video content. Their experience
encoding hundreds of thousands of sound files is the reason I turned to Aaron
Olson, Systems Engineer for Loudeye, for many of the tips in
this article. So if you feel overwhelmed and need professional services
you have options.
