Making Digital Music Recordings |
|
These tiny devices can
hold 300 hours of music. This article tells you how to
create the music files that such units need. |
You now have more choices for
getting the best compromise between sound quality and music file
size when recording your CDs and other music sources to a
digital music player such as an Apple iPod or Microsoft Zune.
Clearly, some formats are
better than others. How do you know which to choose?
This article tells you what you
need to know to make this choice.
>
The Main Differences between the
Formats
Digital music formats fall
into three broad categories.
Uncompressed
An uncompressed music file
has all the digital information possible. The most common
example of this type of file is a .WAV file. Another
common file type is the .AIFF file type.
Depending on your sampling
rate settings, these types of files can eat up 10MB of disk
space per minute of music time. Even an 80GB music player
doesn't last for long at such data rates - you'd get only 133
hours of music on an 80GB player if storing uncompressed .WAV
files.
The flip side to the very
large file size is that it also takes a very long time to
download uncompressed music files from websites. A 5
minute song on a 100kb dataline could take over eight minutes to
download - clearly there's no way you could listen to such music
in a real-time streaming situation.
For these reasons, it is
very rare to see digital music being distributed in uncompressed
formats.
Lossless Compression
Lossless compression uses
some clever methods to use less file space to store the same
data as in the uncompressed file. Typically, it does this
by using short 'code' groups of data to represent longer groups
of real data.
For example, perhaps the
music data consists of sixteen values that we'll represent by
the letters A - P. A typical piece of music might contain
this set of values
BCAEFNCAJKOCAMB
This might look like a
meaningless jumble of letters, but the compression program
notices that there are three sequences of CA in this file, and
so it assigns a special value to represent the two letter
sequence CA - let's say this special value is the number 1.
So the file can now be represented as
B1EFN1JKO1MB
And we are now using 12
instead of 15 characters to represent the same information (a
20% reduction).
Lossless compression can
sometimes halve the size of the music file (ie about 5MB of data
per minute of music), without causing any noticeable
deterioration in sound quality. This is of course a good
thing, but because the next form of compression can make file
sizes so very much smaller, there has been very little interest
in lossless compression. It is very rare to see digital
music being distributed in lossless compression formats,
although there is an Apple Lossless file format (supported by
the iPod).
Lossy Compression
This is the most common form
of music file compression. Typically, it makes a file much
smaller by taking similar groups of values and making them the
same. Here are two examples - one audio and one visual.
The visual example shows you
how lossy compression works with images. Here is a jpg
image with six levels of compression. The images start at
35kB in the first image (with least compression) and then become
successively smaller as more and more compression is added -
18kB, 12kb, 6kB, 4kB and 3kB.
Can you spot any difference
between the first two images? Probably not, even though
the second image is only half the size of the first. Even
the third image - one third the size - still looks almost the
same.
For a casual picture, even
the fourth picture (six times smaller) is okay. By the
fifth picture though, some obvious fuzziness and lack of detail
can be noticed, and the sixth picture starts to be obviously
lacking in quality.
  
  
Musical Samples of Lossy Compression
And now here is a 1 minute
43 second
music sample (a waltz from Act II of Tchaikovsky's Swan Lake), recorded with three different levels of
compression in MP3 format.
The first compression level is at a rate of
192kb/sec (this is about one seventh of the uncompressed data
rate). The second compression level is at a rate of
96kb/sec (one fourteenth the uncompressed data rate) and the
third sample is at a rate of 48kb/sec (one twenty-eighth the
uncompressed rate).
Depending on the audio
equipment you're using to listen to these six music samples and
how well trained your ear is, you'll probably notice no
difference in quality between the first and second samples, and
will have to concentrate very hard to notice a difference in
quality between the first and third samples.
But you will probably notice
that the fourth sample isn't as clear, and the fifth sample
starts to sound like an old recording through a cheap radio, and
the last sample is obviously bad.
Audio File Size
(click to play) |
Sample Rate |
Compression |
4026 kB |
320 kbps |
4 x smaller |
2014 kB |
160 kbps |
8 x smaller |
1008 kB |
80 kbps |
17 times smaller |
504 kB |
40 kbps |
33 times smaller |
254 kB |
20 kbps |
67 times smaller |
102 kB |
8 kbps |
167 times
smaller |
In both the visual and sound files, you have now seen and heard
how it is possible to introduce a great deal of compression
without any obvious visible or audible loss of quality.
So, for most of the time and most applications, lossy
compression - if not too extreme - is considered perfectly
satisfactory.
For music and MP3 files, a
typical sample rate is 128kbps or higher. I discuss the
optimum MP3 compression rate and
sound quality in my article on making MP3 recordings.
Different Types of Lossy Music
Formats
There are perhaps three main
types of lossy music formats - in alphabetical order, they are
AAC, MP3 and WMA.
AAC is the newest of the
three formats and MP3 is the oldest. Perhaps
unsurprisingly, AAC is in many respects the 'best' format and
MP3 the 'worst' in terms of giving the best music quality for
the greatest compression.
, and WM
How is One Format Better
than Another
space the song uses depends
on the song and the import settings you choose in iTunes preferences. The import settings also
affect the audio quality. In general, the larger the file the
better the sound.
This chart provides an overview of the formats you can play and
save files in with iTunes.
Encoding format
Can be played using
File size (approx)
Comments
AAC (MPEG-4)
iPod, applications that support QuickTime
> 1 MB/min (High Quality setting)
Default for Windows and Macs with QuickTime 6.2 or later
MP3
iPod, computers, most digital music players
1 MB/min (High Quality setting)
Default for Macs with QuickTime 6.1 or earlier
AIFF
Many applications
10 MB/min
Use for burning high-quality audio CDs from imported songs
This is basically the
regular normal CD format
WAV
Windows computers without iTunes and computers without MP3
software, and Macs
10 MB/min
Apple Lossless
Some iPod models, applications that support QuickTime
5 MB/min
Use for burning high-quality audio CDs from imported songs
also .wmv but not supported
Lossy
transcoding IS BAD. Don't do that. Every time you transcode
from
lossy to lossy you loose more. Even if your end format is
"better" than mp3, you CANNOT increase the quality, only
decrease it. If you have a lot of mp3s for which you don't have
the cd source, leave them as mp3.
For one thing, both iTunes and WMP are perfectly fine with a
library full of mp3 files. They don't require to use AAC or WMA.
You can have WMP11 put album art with your old mp3s.
Modular encoding
AAC takes a modular approach to encoding. Depending on the
complexity of the bitstream to be encoded, the desired
performance and the acceptable output, implementers may create
profiles to define which of a specific set of tools they want
use for a particular application. The standard offers four
default profiles:
Low Complexity (LC) - the simplest and most widely used and
supported;
Main Profile (MAIN) - like the LC profile, with the addition of
backwards prediction;
Sample-Rate Scalable (SRS), a.k.a. Scalable Sample Rate (MPEG-4
AAC-SSR);
Long Term Prediction (LTP); added in the MPEG-4 standard - an
improvement of the MAIN profile using a forward predictor with
lower computational complexity.
Depending on the AAC profile and the MP3 encoder, 96 kbit/s AAC
can give nearly the same or better perceptional quality as 128
kbit/s MP3.[1]
The firMP3 is a way of compressing
sound files into smaller sizes. Just like jpg compression makes
picture files smaller, so too can MP3 compression make sound
files smaller (than, for example, they are on a regular CD).
MP3 files contain music. MP3
files can also hold considerable information about the piece of
music such as the name of the artist, the song title, and
various other information as well.
MP3 files have become an
industry standard format that most computer music playing
programs can play.
How to Make an MP3 File from a
CD (or other source)
You need a program to
convert the music from the CD or tape or whatever into MP3
format. There are plenty to choose from, and most are quite
inexpensive. Perhaps the best known, and the one I use myself,
is
MusicMatch. Another popular program is
Easy CD-DA.
Microsoft's Windows Media Player can also do this if you buy a
small add-on to the program.
Simply put a CD in your
CDrom, start up your program (sometimes it will even start
automatically, change a few settings, then hit the record
button. Easy!
Compression - Something for
Nothing?
The most distinctive thing
about an MP3 file is that it is very much smaller than a regular
CD file - typically only one tenth the size of a CD file. In
other words, the music data has been compressed.
You have a choice of many
different levels of compression. But you don't get something for
nothing - the more compressed the file, the poorer the sound
quality. Choosing the right amount of compression involves a
trade-off between sound quality and file size.
The type of sound quality
that is 'best' for you depends on how discerning a listener you
are, and how you will be listening to the music. If you are
listening to it through a car stereo system, then you don't need
as good a quality as if you are listening to the music through a
set of high end headphones. If you are playing the music back
through a small set of cheap speakers, you again don't need the
same quality as if you are playing it through a $10,000 hi-fi
system.
The Best Compression Compromise
If you are short of file
space and do not care about the music quality too much, compress
at a sampling rate of 128 kb/sec. If you have more space and
want audibly better sounding music, then increase to 160 kb/sec.
If you want to have excellent quality music, almost as good as
on a CD, then use a 192 kb/sec rate.
It is possible to compress
at rates as high as 320kb/sec, but you're really starting to get
into diminishing returns when climbing over 192kb/sec.
This is what is known as
'constant bit rate' (CBR) compression. There is also a different
form of compression called 'variable bit rate' (VBR)
compression. This works on the basis that sometimes music can be
compressed a lot, and sometimes it needs to be much less
compressed to preserve its quality. If you choose VBR, start
with a setting around 50% - the higher the percentage, the
better the quality (and the larger the file size).
Other Recording Settings
There are several other
important settings that you need to optimize in your program.
The first of these is to set
your program to take digital audio from the CDrom drive, not
analog audio. Look for a setting somewhere in your recording
settings - it might say 'DAE' or 'Digital recording mode'. If
there is an 'error correcting' option, take this, too.
Maybe there is a bandwidth
option - to choose how high the maximum frequency you will
encode is. The human ear, in theory, can hear up to about 20kHz,
but this ability drops off with age. Most speakers do not play
sounds this high in any case. The lower the bandwidth, the
better the quality of the remaining music. I'd suggest setting
somewhere between 15kHz and 18kHz as optimum, and if you have
poor quality speakers, you could even go down as low as 12kHz.
Changing the bandwidth setting does not change the file size.
Maybe your program also has
an option to specify how carefully it compresses the music. This
might be referred to as 'encoding' or 'processing'. The higher
you set this, the better the sound quality will be - without
causing the file size to grow. For that reason, I always set the
processing to 'maximum'. The only trade-off here is that it
really slows down the speed at which a CD is converted to MP3.
How Long Does it Take
The time it takes to convert
a CD to an MP3 depends on your settings - the higher the
processing and the broader the bandwidth, the more computing
your computer has to do.
The time it takes also
depends on your computer speed. For example, I use two computers
to convert CDs - one is a P3-450MHz and the other is a
P4-1.7GHz. The faster computer is about five times faster than
the slower computer at converting files.
With moderately high quality
settings, my fast computer will record at up to eight times the
CD play back speed (in other words, 80 minutes of music take ten
minutes to record and convert). With very high quality settings,
it runs at about 1.9 times play back speed (the 80 minutes now
takes 57 minutes. And, with the same very high quality settings,
my slow computer runs at about 0.3 times playback speed (80
minutes takes 265 minutes to record!).
There might be a way to
speed up your recording time - by updating your CDrom driver.
Usually the CDrom drive uses a generic driver. I used the latest
free version of Real Audio's Real One player and within that
there is an option to download and install a better Adaptec ASPI
driver. Doing this increased my CDrom's speed by about 30%!
Naming Files - CDDB
It is very important that
your music files have helpful names that tell you what the music
actually is that you have recorded.
The good news is that this
process can be automated, most of the time, with most recording
programs. The recording program will recognise a unique
identifier for the CD and then will, over the internet, query a
massive database (usually the CDDB database) to find out details
of what the CD is, who the artists are, what the track names
are, etc. This information isn't always 100% accurate, but most
of the time it is nearly 99% accurate. The recording program can
then automatically add this information to the files you create
so that you don't need to type in anything yourself.
Playlists
Maybe you have two or three
CDs that you have recorded. One track is duplicated on a couple
of the CDs, and there are a couple of other tracks that you
don't like, and you would like to change the order of the music,
with a mix of tracks from first one CD then a different CD, and
so on.
This is all easily possible
by creating a playlist. The playlist simply tells the playback
program which files (tracks) to play, and in what order. A
playlist can combine tracks (files) from as many different CDs
as you wish, and in any order you like.
Creating playlists can give
you tremendous flexibility in how you finally play back your
music.
Special Situation - Voice Only
Recordings
If you are recording
'talking books' or other types of voice only conversation, you
can use much more compression. And if the sound source is mono,
not stereo, you can use even more compression. For voice, set
your bandwidth down to maybe 8kHz and use a compression rate of
96kb or even 64kb (if mono).
Related Articles, etc
|
If so, please donate to keep the website free and fund the addition of more articles like this. Any help is most appreciated - simply click below to securely send a contribution through a credit card and Paypal.
|
Originally published
3 May 2002, last update
21 Jul 2020
You may freely reproduce or distribute this article for noncommercial purposes as long as you give credit to me as original writer.
|