374
Telephone Bell
divide to a common mode, but not quite. Following a spectrogram in real time
you will often see bands dancing around a certain point, sometimes exhanging
energy between two or more peaks in a regular pattern before finally settling
on one dominant mode. In this case it is around the range 1:2.5 to 1:3.5 (which
is more obvious from listening than looking at the graphs).
Summary of Model
So, the shape, material and structure, and excitation determine how waves
propagate around the object and how its resonances create certain modes of
vibration. In this system, energy movements are heard as sound, or die away
through dissipation. Objects of regular shape and material makeup focus sound
into more definite frequencies than irregular and inconsistent ones. Let’s take
stock of some things we know:
•
The shape, size, and material of a bell remain fixed.
•
The excitation is very short compared to the length of the sound.
•
A bell is usually made from a dense, elastic, and regular material.
•
There are a number of pure sinusoidal components in a bell sound.
•
Their relationship is initially complex, but simplifies with time.
•
They each evolve in different ways through the duration of a bell note.
•
Eventually all die to nothing. Energy is radiated as sound or lost to
entropy.
•
Overall, the spectrum of a bell is quite sparse.
Specification
And let’s recap on specific parameters that define our target sound.
•
The fundamental frequency of the bell is about 450Hz to 500Hz.
•
It dies away in about 3 seconds, give or take a fudge factor.
Method
Remember what we are focusing on: a specific kind of bell, an old-style tele-
phone bell, made of brass, a couple of inches across, struck repeatedly with a
small hammer connected to a buzzer. That gives us at least three components
to consider: the hammer, the bell, and the buzzer. Additionally, we need to be
thinking about any housing or case that the bell operates in, sources of energy,
and objects that act as sinks of acoustic energy or radiators. So, turning to the
method, how can declarative knowledge, stuff about what a bell is and how it
makes a sound, be moved to imperative or procedural knowledge, stuff about
how to implement the bell and select tools from our DSP box of tricks? Now is
the time to consider why we might choose one synthesis method over another,
and so we need a little computational wisdom to help know which approach
might be the most efficient. For real-time implementation in games we also
care which is easiest to code, most flexible at run time, and which presents the
most useful interface to application components above. Then we can fill out
Do'stlaringiz bilan baham: |