Each additional algorithm for audio effects or music synthesis creates exciting
new opportunities for the content developer, but it creates new challenges
for the hardware developer because the algorithm is usually required to run
concurrently with a bunch of other algorithms. Each algorithm consumes
computational resources, so higher concurrency requires faster hardware.
The challenge lies not in devising an algorithm that provides the desired
effect, but finding a way to make it work concurrently with all the other
algorithms required to provide the full range of desired effects.
For 1998, the concurrency requirement in PCs is largely defined by the APIs
that we discussed previously:
8 streams (more than 8 will result in a muddle; 4 is about the limit of
perceptibility, but market requirements will dictate the higher number)
8 streams (speech, telephony, system events, and unlocalized sounds)
64 voices (downloadable sounds will be used in part for sound effects, so
a 64-voice capability will preserve the ability of the wavetable engine to
provide 32-voice musical accompaniment)
This list requires significant horsepower, so
hardware developers face a serious challenge.
An industry group is forming to take MIDI synthesis to the next level. One
mandate for the group is to enhance the DLS specification. The current version
of the standard was designed for compatibility with as broad a range of existing
products as possible. The next version will call for refinements that give
content developers more precise control over sounds. Another standardization
effort that may play into this effort is MPEG-4, which will define not only
a technique for compressing digital audio, but also a protocol for specifying
music synthesis functionality.
Better audio quality through USB and 1394
Just as AC97 recognized the difficulty of accurately converting a digital
representation of a waveform to an analog one with chips that combine analog
and digital circuitry, many manufacturers recognize that leaving the codec
inside the PC imposes a limitation on audio performance because of the same
problem leakage of all the digital signals into the analog signal.
USB and 1394 make it possible to move the codecs out of the PC altogether.
The digital signal gets shipped to a suitably equipped loudspeaker or amplifier
where it is converted to analog within a relatively controlled, electrically
quiet environment. Such a system architecture will make it possible for computers
to deliver audio performance comparable to what we expect from our hi-fi
systems at little additional cost.
Better music synthesis techniques
While all of these advances to the technology for music synthesis and audio
effects will improve sound quality dramatically, new technology on the horizon
will improve it even more. A new music synthesis technique known as waveguide
synthesis is capable of synthesizing musical sounds that are even more realistic
and expressive than wavetable. Waveguide synthesis is based on physical models
of instruments equations that describe mathematically the way an
instrument behaves. By programming these equations, it is possible to simulate
the instrument electronically. Some parameters of the equations correspond
to musically useful characteristics, such as the pressure of the bow against
the string or the bite on a reed. Changing these parameters can alter the
sound quality in ways comparable to what happens in the real instrument.
The use of such advanced synthesis techniques will result in sounds that
are not only more realistic, but also vary in natural ways.
On the downside, waveguide synthesis is more compute-intensive than wavetable
synthesis, but faster processors will overcome this limitation. Also, it
can be exceedingly difficult to create the physical models. The physics of
instruments are very complicated, which is why it is still largely a mystery
as to what makes a Stradivarius violin sound so good. Furthermore, it is
almost impossible to know where to start in creating a physical model for
an instrument called Bright, which is one of the instruments
required by General MIDI. Expect to see synthesizers with greater realism,
ease of use, and expressiveness in PCs because people care about sound quality.