[0001] The present invention relates to a circuit for an electronic musical instrument,
in particular to a circuit having the features set forth in the preamble of patent
claim 1. Such a circuit is disclosed in German published patent application 27 15
510 based on Japanese priority application P 51-38466.
[0002] Hereinafter, individual sections of a wave form will be designated with the internationally
used words "attack", "sustain", "release".
[0003] In accordance with the teaching of the publication referred to above, a complete
wave form is memorized including attack, sustain and release if a percussion sound
is to be simulated as, e.g., piano notes. If by striking a key on a manual of the
instrument, such a note is called upon, the wave form memory is read-out completely
even if the player will prematurely cease to strike the key. However, in this latter
case, the wave form is modified in that data read-out after release of the key are
multiplied by an attenuation function.
[0004] If it is necessary to permit production of a sound of arbitrary sustain duration,
as e.g. organ notes, the publication referred to above suggests to use, as the release
portion of the note, quickly or slowly attenuated data of the sustain period.
[0005] It is an object of the present invention to improve the circuit as defined in the
preamble of patent claim 1 such that always the complete wave form including the release
section is read-out wherein the transition from sustain to release can be provided
at any point of the sustain section, without the necessity to complete reading-out
of the entire sustain wave form before release can commence. In fact, this permits
to memorize relatively long periods of a natural sound and to thereby improve the
simulation while nevertheless the reading of the sustain may be interrupted with only
negligible delay in order to commence release data reading upon cessation of a key
stroke.
[0006] The characterizing portion of patent claim 1 defines the means which, in accordance
with the present invention, are provided to achieve the object defined above.
[0007] Accordingly, during the memorization of scanning data, certain memory elements -
preferably regularly spaced - are preserved for the command words. Upon their reading,
a decoder detects whether in the time period since the reading of the previous command
word a signal "sound off" has occurred (which would e.g. mean in case of an electronic
organ that the respective key had been released by the player). Only under these conditions,
the command words will cause a "jump" to the memorized release section which will
then be read-out.
[0008] Preferred additional features are defined in the subclaims, and the results of their
features will become apparent upon reading of the following detailed description of
embodiments.
[0009] Reference is made to the accompanying drawings.
[0010]
Fig. 1 shows the envelope of a sound, for example, a trumpet note.
Fig. 2 illustrates schematically how such a sound is reproduced by means of the circuit
of the invention.
Fig. 3 illustrates with a temporal scale considerably increased over that of Fig.
2 the interspersing of the command words into the scanning values.
Fig. 4 illustrates the envelope of a percussion sound (e.g. piano), in one case with
the key held depressed and in the other case with the key briefly struck.
Fig. 5 shows the reproduction of such sound.
Fig. 6 shows in analogy to Fig. 3 the interspersing of the command words.
Fig. 7 is a considerably simplified block circuit diagram according to the invention.
[0011] Fig. 1 shows the temporal events of a sound generated by means of an instrument permitting
to "hold" a note, as e.g. a trumpet note. The amplitudes show three characteristic
sections: Attack A, Sustain S, and Release R.
[0012] Fig. 2 shows the envelope plot as reproduced for such a note by means of the circuit
of the invention. The sections A and R are memorized in their entirety while of the
section S, only a portion S' is memorized and will be repeated in case of a duration
of a "sound on"-signal with a duration exceeding that of the memorized portion S'.
It will be appreciated that the reproduction will sound the more natural the longer
subsection S' is chosen, however, there is the inherent limit of memory capacity and
access time. A duration of subsection S' should have a minimum reading duration of
about 100 milliseconds, and 500 milliseconds would be preferable.
[0013] On the other hand, when already a "sound off"-signal is present, occurring somewhere
within S', it is undesirable to complete the actual reading of S' before the sound
is continued with the R section. For this reason, under predetermined addresses of
the wave form memory holding the amplitude scanning values, command words GOREL are
written in. Upon presence of a "sound off" instruction ( i.e. in case of an organ,
release of the respective key), at first the reading of subsection S' will continue
but only until an address holding a GOREL is met. From then on, a jump is made to
continue with section R. The GOREL commands may be addressed with a temporal spacing
of about 10 or, as a maximum, 20 milliseconds - depending on the clock frequency -
while their minimum spacing is one period of the sound oscillation. This has the consequence
that a certain wave form memory is suitable for a limited range of clock frequencies
only; higher or lower clock frequencies are handled by other memories which are loaded
in a commensurate manner. For example, one memory may be provided for each octave
to be playable by the instrument.
[0014] As illustrated in Fig. 3, it is preferred to select the addresses of the GOREL such
that they are at amplitude zero crossings of the memorized sound because under this
condition, noise is reduced to a minimum.
[0015] Fig. 4 is analog to Fig. 1 but shows percussion sounds, e.g. of a piano.
[0016] Such sounds have an A section which is immediately followed by a R section. The envelope
is shown in Fig. 4 for the case of a steadily depressed key causing a steady decrease
of the amplitude, and also for the case of a sudden key strike.
[0017] As shown in Fig. 5, both these envelopes are memorized without, of course, provision
of a sustain section.
[0018] Fig. 6 shows that at the GOREL commands, there is always the possibility to jump
to the address of the same amplitude but in the R section of the memory which results
in an excellent simulation of natural sounds. This jump is performed under the condition
that the depressed key is released.
[0019] Fig. 7 shows in block form and reduced to the components which are of importance
for the implementation of the invention a circuit according to the principles of the
present invention.
[0020] Scanning values of the desired wave form are memorized in wave form memories of which
only one is shown and designated with 10. The data may be absolute values, differential
values or other digital representations of the amplitude variations. As a memory,
either ROM or RAM type memories may be used, in the latter case external loading means
are provided, preferably a microprocessor.
[0021] At certain memory addresses, command words GOREL and GOTO, respectively, are memorized.
[0022] If it is e.g. desired that a maximum 20 milliseconds delay after release of a key
be permitted before the A section or S' section (as the case may be) is quitted and
the sound is continued with the R section, for a basic frequency of the sound of,
say, 250 cps one GOREL should be memorized after each fifth sound period. If the reading
frequency is doubled, the temporal spacing between two succeeding GOREL commands will
be reduced to 10 milliseconds only. In order to maintain this 10..20 milliseconds
spacing, thus one envelope memory is to be provided for each octave.
[0023] GOTO commands are provided only if the memory contains a S' section, too, said subsection
S' being simply repeated if the respective key is held depressed long enough. The
reading duration of S' is at least 100 milliseconds. This means that for the example
given above, for a sound frequency of 500 cps at least fifty periods of a natural
sound are to be memorized to form S'. (With 250 cps, of course, the S' duration will
be doubled to 200 milliseconds).
[0024] The reading clock is supplied by clock generator 12 which addresses an address counter
14. The latter receives, via multiplexer 16, the start address (upon depression of
a key), and data retrieved under the read-out addresses are passed to a digital-to-analog
converter 18 which generates a respective analog signal to be further processed by
components as filters, amplifiers and the like, as is common practice in electronic
musical instruments. In addition, the data are fed to a decoder 22 which detects whether
a GOREL or GOTO has been read. The following possibilities are to be considered:
- key still depressed: Ignore GOREL.
- key still depressed: GOTO will set address counter to the start address of subsection
S' (if any).
- key released (symbolized by STOP line): Ignore GOTO.
- key released: GOREL will set address counter 14 to the "homologue" address of the
R section of the memorized sound, i.e. that address where
(i) with a preceding S' subsection, the R section begins, or
(ii) with a preceding A section (or portion thereof) the same or most similar amplitude
value of the R section is memorized.
[0025] In any such event, the digital-to-analog converter is disabled, and for this purpose
a disabling input is supplied to its BLOCK input.
[0026] The processing described above is implemented by means of a control logic 20 which
may be a ROM or a gate array logic and which supplies the respective control signal
SET. Such a control logic may be easily implemented by an expert skilled in the art
based on the following truth table:

[0027] It will be understood that the circuit embodiment as described above may be modified
in order to produce a polyphonic instrument. In this connection, reference is made
to European Patent 36074 which discloses a circuit arrangement operating with variable
clock rate.
[0028] The circuit of Fig. 7 may be implemented by using integrated circuits available on
the market.

1. A circuit for an electronic musical instrument, comprising a clock generator producing
a variable clock rate commensurate with sound frequencies to be generated, comprising
further a wave form memory readable by means of clock generator output signals and
holding scanning values of several periods of a note including its release section,
said values being digitally memorized, and comprising further a digital to analogue
converter by means of which the read-out scanning values are converted into analogue
signals to be processed by succeeding circuits so to produce audio-signals to be sounded
in analogue form characterized in that predetermined memory element addresses of said
wave form memory hold command words controlling, when read-out, transmission of a
command signal to said clock generator so to set the latter to an address under which
in said wave form memory the same scanning value as prior to said command but in the
release section of the envelope is read, and that the digital analogue converter is
disabled for said command words.
2. A circuit as set forth in claim 1, characterized in that between succeedingly readable
command words at least one period of a wave form is memorized.
3. A circuit as set forth in claim 2, characterized in that the clock generator frequency
and the number of scanning values memorized between succeeding command words are adapted
to each other such that the temporal spacing between succeedingly readable command
words will be 20 milliseconds mostly.
4. A circuit as set forth in claim 3, characterized in that a plurality of wave form
memories are provided each allocated to a range of clock generator frequencies.
5. A circuit as set forth in claim 4, characterized in that each wave form memory
has a memory capacity for a reading duration of at least 100 milliseconds.
6. A circuit as set forth in one of claims 1 to 5, characterized in that under second
predetermined memory element addresses of said wave form memory and in the order of
reading behind a predetermined number of scanning values of a sustained note, second
command words are memorized which enable upon being read, transmission of a second
command signal to said clock generator so to set the latter onto an address under
which in said wave form memory the first scanning value of the sustained note just
sounded is read-out again, and that said digital to analogue converter is disabled
for said second command words, too.