of Microprocessors

MUSIC SYNTHESIS SOFTWARE 697
However, if a period of silence is needed, a statement having a lone rest
specification can be used.
Figure 18-23 shows some common musical situations and how they
might be coded in NOTRAN. This sequencing method is quite general and
should be able to handle any situation likely to be encountered in reasonably
conventional scores.
Level 3 Routines
The primary function of Level 3 software is to decode the NOTRAN
statements that were just described and extract their information content.
Conversely, if a typing or specification error is found, the condition should
be reported to the user. The routines for statement scanning and syntax
checking, however, are in the realm of compilers and other language analysis
programs and are therefore beyond the scope of this text. Basic NOTRAN as
specified here is very simple in comparison with a typical programming
language; thus, a statement interpreter should not be very difficult to write
even by one experienced in general programming but inexperienced in compilers.
In fact, a successful interpreter can be written with no knowledge of
classic compiler theory at all, just common sense. The main danger, however,
is a program that might misinterpret an erroneous statement rather
than flag it as an errot.
Once a statement is decoded, its information content is systematically
stored in tables. In the case of a voice statement, for example, the various
envelope parameters are stored in a table that describes the characteristics of
that voice. The waveform parameters are then used to compute a waveform
table (or tables) that is stored away, and a pointer to the table is stored along
with the envelope parameters. The voice 10 is also part of these data. Before
allocating additional table space, a scan is performed to determine if the same
voice 10 had been defined previously. If so, the new information replaces the
old; otherwise, more table space is allocated. Percussive voice definitions are
handled in the same way except that the parameters are different and no
waveform table is needed. TEMPO statements simply update a single tempo
variable, which then influences all succeeding time calculations.
When an actual event is encountered in a note statement, several
operations must be performed. First, the voice 10 is extracted and used to
locate the table of information created earlier when the corresponding voice
statement was processed. Next, a new event control block (ECB) is created
(by scanning the control block area until an inactive one is found), and
pertinent information such as waveform table addresses and envelope
parameters are copied over to it. The frequency parameter in the ECB is set
by further scanning the specification for pitch information. Next, the duration
and articulation specifications are analyzed and the information used in
conjunction with the tempo variable to set the sustain duration. Finally, the
duration is compared with the current "shortest duration," which is updated