Specifications

32
Basic Sampling Concepts
Samplers manipulate audio signals in digital form.
The sampling process can be described as 1)
segmenting a continuously changing analog signal into
extremely short time intervals, 2) converting the
amplitude level of the signal in each segment into digital
data, and 3) storing the data in memory. Sampling
occurs when a sampler records sounds. Analog audio
signals, which are continuously flowing waves of sound,
are converted into digital “steps,” and if these steps are
fine enough, the human ear will not be able to
distinguish the digital sample from the original analog
signal. “Sampling rate” and “bit resolution” are key
words indicating how fine these steps are.
The “sampling rate” (sampling frequency) indicates how
finely the time intervals of a sample are divided. Sound
is typically digitized at frequencies up to a maximum of
one-half the sampling rate. If the sampling rate is 44.1
kHz, sampling takes place 44,100 times per second, and
harmonics up to 22.05 kHz can be recorded. The higher
the sampling rate, the greater the range of sounds that
can be digitized, and the more faithfully sharp changes
in amplitude within a short period of time can be
recorded (Figure 1).
The bit resolution (number of quantization bits)
represents the number of digits used to convert the
amplitude into the binary system of 0’s and 1’s, and the
dynamic range is the bit count x 6 (dB). If the bit
resolution is 16, the sound is digitized using 65,536
(2
16
) steps, and has a maximum dynamic range of 96
dB. The higher the bit resolution, the more faithfully
changes in amplitude can be digitized (Figure 2).
Audio data that has been recorded by a sampler is called
a sample (also called wave data), and it is stored in the
sampler’s wave memory. The sampler converts samples
of various sampling rates and bit resolutions into
pitches that correspond to the notes of a standard
musical scale. The sampler then outputs the data as an
analog signal, and in some cases converts it into a
digital audio signal. How accurately the signals are
reproduced also has a significant impact on sound
quality similar to the accuracy when sampling (during
recording). The key factors in this regard are the bit
resolution of the D/A converter, the bit rate of the digital
audio signal output by the sampler, and the accuracy of
the digital processing within the sampler.
The samples that current samplers handle are
primarily 44.1 kHz, 16-bit stereo samples, the same
audio quality as CDs, but new formats are emerging in
the world of audio, and will be adopted by samplers in
the future. Samplers have become widely popular, and
you might find one almost anywhere. This part will be
discussed later.
Part 1 Introduction to Sampling
◊ Basic Sampling Concepts ◊ Basic Sampling Techniques ◊ Sampler Parameters
Basic Sampling Techniques
Creators of sample sound libraries use sound editing
techniques unique to samplers to produce high-quality
voices that closely resemble the actual instruments.
Here, we will introduce three basic techniques. Not only
are these techniques used in creating sample sound
libraries available on the market, but also in creating
sampled waveforms for PCM synthesizers.
■ Looping
Looping is a familiar technique. For sounds that
simulate real instruments, it refers not to looping a
phrase, but to repeating a portion of a sample over and
over again to obtain a long sustaining sound.
For example, let’s consider sampling the sound of a
piano. Of course, reproduction quality would be best if
we could sample the entire duration—from the moment
the note is struck until it completely fades away—but
this would quickly use up the sampler’s wave memory.
Instead, the sound is sustained by repeating a portion of
the sample and using an amplitude EG (envelope
generator) to make the sound decay like a piano before
saving it into wave memory (Figure 3).
■ Multi-Sampling
When the pitch of a sample is varied over a wide range
to create a scale, the farther away the sample is from the
original pitch, the more the characteristics of the sound
change. Even with a simple piano sample, if the pitch
were changed to a large degree, you would never
recognize it as a piano. For this reason, “multi-
sampling” is often used. Multi-sampling is done by
dividing the instrument into multiple key ranges (ideally,
one sample per key), thereby minimizing the pitch range
represented by one sample (Figure 4).
■ Velocity Switching, Velocity Cross-Fade
With real musical instruments, dynamic variations in
tone can be obtained according to how the instrument is
played—in the case of a piano, how hard the keys are
struck. By using multiple samples of different dynamics
and switching or cross-fading among them depending
on velocity (how hard the key is struck), accurate sound
playback can be achieved (Figure 5).
Sampler Parameters
Virtually all samplers have the same basic sample data
parameters. In order to understand how samples work,
you should familiarize yourself with the parameters
described below, even if you’re just using a MIDI
keyboard to play a sampler. Samplers have numerous
other parameters beyond the basic ones required to make
a tone, and their number and type depend to a great
extent on the model.
The Yamaha A Series in particular has a vast number of
parameters and parameter types. For this reason, the
A Series can be considered “samplers that synthesize.”
■ Start Point (Start Address)/End Point
(End Address)
Used to specify playback of only a portion of a sample.
Even if the sample is long, you can set these parameters
to play back only a short interval. It’s a good idea to
sample with a margin before and after the part of the
sound you want, because you can conveniently adjust the
start and end points as you like afterwards (Figure 6).