Making SoundFonts
Creating a SoundFont is not an easy task. The people at
Utopia spent some 1 1/2 years tweaking the Utopia Live! SoundFont.
Why? Firstly the RAM on board a wavetable synthesizer,
in this case the EMU10K1, is limited. The EMU10K1 supports a maximum of
64MB but if you have less than this amount of RAM, system performance
will be affected if a huge SoundFont was loaded up. The first sound card
that supports SoundFonts is the Sound Blaster AWE32, and it had a small
512KB of RAM available for SoundFonts!
As a result, having a small SoundFont that contains 128
instruments and drum sets is not easy! Making the SoundFont small and
at the same time sound good is not easy too! In the interview,
Roel will be talking more in detail about the techniques used to make
the SoundFont smaller.
To give you a taste of what goes into making a SoundFont,
I'll be talking very briefly about the parameters and considertation in
creating an instrument sample to be included in a SoundFont.
Looping
Every instrument in a wavetable synthesizer is created from
waveforms. As the player can hold down a note for a variable amount
of time, and instruments like the flute can go on forever, it is not practical
to record an instrument continuously for a long period of time. If it
was done this way, the file size will be too large! One of the most important
techniques to makes sample sizes small and instruments play indefinitely
is looping.
This is a shot of a screen in Vienna SoundFont editor,
showing the waveform of an Alto Sax sample in Utopia Live!. When the note
is held down, the sample is looped from the green vertical bar to the
cyan bar. This gives the illusion of a continuous sounding saxophone sample.
The tough part comes in when the SoundFont author has to
find the correct looping points (ie. where to place the green and
cyan bars) to create a smooth, prolonged sound of an instrument without
any distortion. More often than not, the sound from the end of the looping
point (cyan) to the beginning of the loop (green) is too different to
provide a smooth transition.
You may want to try looping a sound in Vienna. I've tried.
Its not easy!
Multi-Sampling
Many of the instruments are multi-sampled as it is not
always possible to use a single waveform and transpose it to different
pitches, while retaining the same timbre. Multi-sampling makes use of
different waveforms across the different ranges of notes so that the entire
range of the instrument sounds more realistic.
For example, the Acoustic Piano in Utopia Live! makes
use of 14 waveforms to recreate the full range of the piano.
Layering
Other instruments used layering to thicken the sounds.
Utopia's Jazz Guitar also mixes in a harp waveform so when
the jazz guitar is triggered, the synthesizer actually plays 2 waveforms!
More Parameters
Just when you thought that looping was enough to create
instruments, think again! Many instruments don't stop abruptly when the
note is released, they actually go through a decay where the volume
gets softer until the instrument sound fades away. Think of the sound
of a piano or guitar.
This is just one of the things that have to be programmed
to make an instrument sound realistic, there are more! Here are some screenshots
of the parameters that can be altered in Vienna to create an instrument.
Confused now? :) Yeah, its that complex. What I've talked about just barely
scratches the surface of the skill involved in creating samples! If you
are interested, you can create a killer piano SoundFont and make big bucks!
There are people who are willing to pay for a good SoundFont, even if
its one instrument!
Now let's talk about the Utopia SoundFont...
|
Back To Top