 |
Description  |
|
|
BACKGROUND OF THE INVENTION
This invention relates to a tone waveshape generation device employed in an
electronic musical instrument and, more particularly, to a device capable
of reading out successive waveshapes of plural periods stored in a memory.
A device which prestores successive waveshapes of plural periods from the
start to the end of tone generation in a memory and generates tone
waveshape signals by reading out these prestored waveshapes is known,
e.g., in the specification of U.S. Pat. No. 4,305,319. The United States
patent discloses a drum generator in which snare drum sounds and other
drum sounds are prestored in the form of successive waveshapes of plural
periods in respectively corresponding memories and these waveshapes are
read out in response to a sound timing signal (play strobe signal). This
type of prior art tone waveshape generation device has heretofore been
used as a tone source of rhythm sounds and increase in the memory capacity
to some extent has not posed any serious problem since necessity for
changing the tone pitch is not involved in this type of device.
If, however, this type of tone waveshape generation device is applied to
generation of scale notes, it becomes necessary to prepare successive
waveshapes of plural periods for respective different tone pitches or tone
ranges with a result that the capacities of memories become extremely
large. If, for example, the duration of tone generation is 5 seconds, the
sampling period is 32 kHz and successive waveshapes of plural periods are
prepared over 4 octaves one for each tone range which has been determined
for 3 scale notes (keys) in 12 scale notes of one octave (totalling 16
tone ranges), a memory having a capacity of "32k.times.5.times.16=2560
kilo words" is required.
An electronic musical instrument of a type in which, in the above described
manner, a complete waveshape from the start to the end of generation of a
tone is prestored for each key (note) and then is read out is disclosed in
the specification of U.S. Pat. No. 4,383,462. In the waveshape memory WM31
shown in FIG. 3 of this United States patent, a complete waveshape is
stored and this complete waveshape is read out in response to a signal KD
which represents a key depression timing.
An improvement has been conceived for preventing the capacity of the memory
storing the complete waveshape from becoming too large. According to this
improvement, the attack portion of the tone is stored in its entirety but
only a part of the sustain portion is stored and the stored part of the
sustain portion is repeatedly read out to generate the entire sustain
portion. In the above U.S. Pat. No. 4,383,462, an example of such
improvement is shown in FIG. 6. A complete waveshape in the attack period
is stored in the waveshape memory WM61 and at least one fundamental period
of a tone waveshape is stored in the waveshape memory WM62. An attack
waveshape is read out from the memory WM61 in response to the key
depression (KD signal) and the tone waveform of the fundamental period is
repeatedly read out from the memory WM62 after completion of the read out
of the attack waveshape (IMF signal) until the end of tone generation (DF
signal). According to this improvement, the memory capacity can be reduced
to, e.g. about one-fifth. If in this case the memory length of the
memories corresponding to the respective tone pitches (tone ranges) is
made uniform, blank area will occur in the memories. The memory length is
determined by the lowest tone due to the fact that the rise time of the
tone increases as the tone becomes lower and hence blank area will occur
in a part of the zone of the memory storing the higher note waveshape
which is shorter in the rise time resulting in the waste of the memory
zone.
SUMMARY OF THE INVENTION
It is, therefore, an object of the invention to effectively utilize the
memory zone of the memory without waste in a tone waveshape generation
device using a memory storing successive waveshapes of plural periods.
According to the invention, in a waveshape memory, successive waveshapes of
plural periods are stored in memory zones corresponding to respective tone
pitches or tone ranges in memory capacities which differ from one another
depending upon the tone pitch or tone range. According to information
representing the tone pitch or tone range of a tone to be generated, the
memory zone of the waveshape memory is designated and waveshapes of plural
periods stored in the designated memory zone are read out. Since
capacities of the respective memory zones of the waveshape memory are made
different, the wavesahpe memory can be effectively utilized without waste.
For example, designation of the memory zone is performed by designating a
start address representing the first address of that memory zone and
memory length information concerning the memory capacity of that memory
zone (i.e., the number of address or words of the memory zone).
Alternatively, information representing the final address of the memory
zone may be used instead of the memory length information.
BRIEF DESCRIPTION OF THE DRAWINGS
In the accompanying drawings
FIG. 1 is an electrical block diagram showing an embodiment of the
invention applied to a monophonic electronic musical instrument;
FIG. 2a is an example of division of the respective memory zones in a
waveshape memory in FIG. 1;
FIG. 2b is an example of a waveshape stored in a memory zone in a waveshape
memory in FIG. 1;
FIG. 3 is an electrical block diagram showing an example of the address
generator in FIG. 1;
FIG. 4 is an electrical block diagram showing an embodiment of the
invention applied to a polyphonic electronic musical instrument;
FIG. 5 is an electrical block diagram showing another embodiment of the
invention applied to a polyphonic electronic musical instrument;
FIG. 6 is a diagram showing an example of the clock pulse used in FIG. 5;
and
FIG. 7 is an electrical block diagram showing an example of the address
generator in FIG. 5.
DESCRIPTION OF PREFERRED EMBODIMENTS
FIG. 1 shows an embodiment of the invention applied to a monophonic
electronic musical instrument. A key switch circuit 11 detects a key
depressed in keyboard 10 in accordance with the predetermined order of
priority and thereupon produces a key code representing the depressed key
(consisting of a 3-bit octave code B3-B1 and a 4-bit note code N4-N1) and
a key-on signal KON. A waveshape memory 12 stores, at memory zones
corresponding to respective tone ranges, successive waveshapes of plural
periods. Memory capacity of respective memory zones differs from tone
range to tone range each consisting of three keys. Three keys of one tone
range are keys of semitone intervals (e.g., C2, C.music-sharp.2 and D2).
Assuming that the key range in the keyboard 10 includes C2-C7 (5 octaves
plus 1 key), the number of memory zones (the number of the tone range of
three keys) is 21.
In each memory zone in the waveshape memory 12, there are stored successive
waveshapes of plural periods ranging from the rising portion to a part
(repetitive portion) of the sustain portion of the waveshape as shown in
FIG. 2b. The attack portion is read out only once whereas the repetitive
portion is repeatedly read out whereby a full waveshape of the sustain
portion is generated. A decay waveshape is generated by repeatedly reading
out the sustain portion while a decay envelope is imparted to the read out
waveshape.
A memory zone to be read out in the entire waveshape memory 12 can be
designated by a start address representing the first address in the rise
portion and memory length information representing the number of all
addresses in that memory zone. The repetitive portion of that memory zone
can be designated by employing a repetitive address representing the first
address in the repetitive portion.
FIG. 2a shows an example of division of the respective memory zones in the
waveform memory 12. For instance, the zone of 44 kilo words from address 0
to address 43999 stores successive waveshapes of the lowest tone range
consisting of keys of C2, C.music-sharp.2 and D2. The zone of 38 kilo
words from address 44000 to address 81999 stores successive waveshapes of
a tone range consisting of keys of D.music-sharp.2, E2 and F2. There are
start address, memory length and repetitive address peculiar to each
memory zone, as was previously described with reference to FIG. 2b. Each
memory zone stores successive waveshapes peculiar thereto.
The frequency of the successive waveshape stored in each memory zone is set
such that it becomes equal to the frequency of a tone of a middle key in
the tone range (e.g., C.music-sharp.2) in the lowest tone range when it is
read out with a predetermined reference sampling clock. In a case where a
tone waveshape for the lowest key (e.g., C2 in the lowest tone range) in
each tone range is to be generated, the successive waveshapes are read out
by using a sampling clock which is produced by lowering the rate of the
reference sampling clock by 100 cents. The read out of the tone waveshape
corresponding to the highest key (e.g., D2 in the lowest tone range) in
each tone range is made by using a sampling clock which is produced by
raising the rate of reference sampling clock by 100 cents. As a result,
although tone waveshapes corresponding to three keys in the same tone
range are read out from the same memory zone in the waveshape memory 12,
frequencies of the read out tone wavehsapes are different from each
adjacent tone waveshape by 100 cents due to the difference of the sampling
clock rate by 100 cents. In the above described manner, reading is
performed in a pitch synchronizing way so that the sampling clock is
synchronized with the pitch of the tone to be generated and one sample
point corresponds to one address.
A zone designating ROM (abbreviation of read-only memory) 13 prestores the
start address, memory length information, repetitive address corresponding
to each tone range consisting of three keys, receives at its address input
the octave code B3-B1 and the higher 2 bits N4 and N3 of the note code
provided by the key switch 11 (i.e., information representing the tone
range of the tone to be generated) and provides the start address, memory
size information and repetitive address in accordance with the tone range
to which the depressed key belongs. Correspondence between the note code
N4-N1 and the respective notes C-B in one octave is as shown in the table
below in which an adjacent group of three keys can be distinguished by the
higher two bits N4 and N3.
______________________________________
deviation from
the reference
note code sampling rate
note N4 N3 N2 N1 (cents)
______________________________________
C 0 0 0 0 -100
C# 0 0 0 1 0
D 0 0 1 0 +100
D# 0 1 0 0 -100
E 0 1 0 1 0
F 0 1 1 0 +100
F# 1 0 0 0 -100
G 1 0 0 1 0
G# 1 0 1 0 +100
A 1 1 0 0 -100
A# 1 1 0 1 0
B 1 1 1 0 +100
______________________________________
The less significant 2 bits N2 and N1 in the note code are supplied to a
pitch synchronizing controller 14 and used therein for distinguishing
which key among three keys of memory zone is depressed. The controller 14
generates a sampling clock PSYNC by utilizing a master clock pulse .phi..
The controller 14 produces the sampling clock PSYNC at the reference rate
when, as was previously described, the tone to be generated corresponds to
the middle key in the group of three keys (i.e., N2 and N1 are "01"),
produces the sampling clock PSYNC at a rate which is lower than the
reference rate by 100 cents when the tone corresponds to the lowest key in
the group (i.e., N2 and N1 are "00") and produces the sampling clock PSYNC
at a rate which is higher than the reference rate by 100 cents when the
tone corresponds to the highest key in the group (i.e., N2 and N1 are
"10").
The address generator 15 generates address information identifing an
address of the waveshape memory 12, advancing the address information in
response to the sampling clock PSYNC provided by the pitch synchronizing
controller 14. This address information sequentially increases with the
start address given by the ROM 13 being used as an initial value until the
value amounts to a final address determined in accordance with the memory
length information given by the ROM 13 when the reading address returns to
the repetitive address given by the ROM 13. The variation from the
repetitive address to the final address thereafter is repeated. Thus, the
successive waveshapes in the attack portion are readout once from the
memory zones of the waveshape memory 12 and thereafter the successive
waveshapes of the repetitive portion in the sustain portion are repeatedly
read out.
The envelope generator 16 generates a constant level envelope when the
key-on signal KON is "1" (representing that a key is being depressed) and
a decay envelope signal when the key-on signal has fallen to "0"
(representing release of the key). A multiplier 17 imparts the successive
tone waveshape signals readout from the waveshape memory 12 with an
amplitude envelope in accordance with the envelope signal generated by the
envelope generator 16. In the attack and sustain portions of the tone, the
tone waveshape signals readout from the waveshape memory 12 are directly
provided from the multiplier 17 whereas in the decay portion, the tone
waveshape signal imparted with the decay envelope in accordance with the
decay envelope signal is provided from the multiplier 17. The output
signals from the multiplier 17 are converted to analog signals by a
digital-to-analog converter 18.
The address generator 15 can be constructed as shown in FIG. 3. When the
key-on signal KON has risen to "1", a counter 20 is reset through a
one-shot circuit 19 thereby starting counting of the sampling clock pulse
PSYNC from 0. An adder 21 adds the count output of the counter 20 and
start address data together and supplies its output to the waveshape
memory 12 as an address information. Accordingly, the address information
starts from a predetermined start address. In a comparator 22, the output
of the counter 20 is compared with the memory length information and, when
coincidence has occurred, the repetitive address data is preset in the
counter 20. It is when the output of the counter 20 indicates the final
address of the particular memory zone that the coincidence output is
produced from the comparator 22 and, by presetting the repetitive address
data in the counter 20, the output of the adder 21 is returned to the
repetitive address. The start address data is an absolute address of the
waveshape memory 12 while the repetitive address data is a relative
address to the start address data.
Determination of the start address, memory length and repetitive address is
advantageously made as follows. For effectively utilizing the entire
memory capacity of the waveshape memory 12, the entire memory capacity of
the waveshape memory 12 is first determined rather than the number of
sample points of the successive waveshapes corresponding to the respective
tone ranges, and this entire memory capacity is divided properly for the
memory zones corresponding to the respective tone ranges. In dividing the
memory capacity, more memory capacity is generally allotted to a low
frequency tone range in which the rise time is long, as was previously
described. Assuming, for instance, that the entire memory capacity of the
waveshape memory 12 is 512 kilo words, the capacity of each memory zone
should preferably be set on the basis of 2 kilo words (i.e., it will
become an integer multiple of 2 kilo words). Thus, the start address of
each memory zone is determined on the basis of 2 kilo words and the start
address data can be expressed by 8 bits, for example
"512.div.2=256=2.sup.8 ". In other words, it can be expressed omitting the
address value of less than 2000. The memory length can also be expressed
omitting the address value of less than 2000. If, for example, a maximum
memory capacity in one memory zone is 44 kilo words, any memory length can
be expressed by a binary number of 5 bits. By using a relatively large
unit of division in dividing the memory capacity for the respective memory
zones, the numbers of bits for expressing the start address and memory
length can be reduced and the constructions of the ROM 13 and the address
generator 15 can thereby be simplified.
For an effective utilization of one memory size, successive waveshapes are
stored in the entire memory zone so that the waveshape amplitude value (or
phase) of the final address will be randomly determined. For smooth
continuation of the repeatedly read out successive waveshapes, the
repetitive address is determined suitably so that the waveshape amplitude
value (or phase) of the repetitive address will correspond to that of the
final address. For this purpose, the repetitive address is set on a one
word basis. As was previously described, since the repetitive address is a
relative address within one memory size, if the maximum memory size is 44
kilo words, the repetitive address can be expressed by a binary number of
16 bits.
FIG. 4 shows an embodiment of the invention applied to a polyphonic
electronic musical instrument. In this embodiment, a tone waveshape is
read out on a time shared basis from plural tone generation channels. In a
depressed key detection and key assigner circuit 23, depression and
release of keys in the keyboard 10 are detected, a depressed key is
assigned to any one of the tone generation channels, and a key code B3-B1,
N4-N1 and a key-on signal KON are produced on a time shared basis. A zone
designating ROM 13 is of the same construction as the one shown in FIG. 1
and provides memory length information, start address data and repetitive
address data channel by channel on a time shared basis in response to an
octave code B3-B1 and more significant 2 bits N4 and N3 of a note code
supplied thereto on a time shared basis. A frequency number converter 24
produces a numerical value (frequency number Fc) corresponding to one of
note intervals -100 cent, 0 cent and +100 cent in accordance with less
significant 2 bits N2 and N1 of the note code. Alternatively stated, in
this example, a tone frequency for each key in one tone range group
consisting of three keys is established by changing the rate of increase
of the address signal for a certain sampling timing in accordance with the
frequency number Fc.
In an address generator 25, in the same manner as was previously described,
the reading address of waveshape memories 12A and 12B is advanced with the
start address being used as an initial value and repeatedly changed
between the repetitive address and the final address. This advancing of
the address is performed for each channel on a time shared basis. Timing
of the change in the address, i.e., the sampling rate, is constant in
correspondence to the channel time division timing. By repeatedly adding
(or subtracting) the frequency number Fc at each sampling timing, the
waveshape reading address for each channel advances at a rate
corresponding to the frequency number Fc.
The waveshape memories 12A and 12B prestore, as in the waveshape memory 12
in FIG. 1, successive waveshapes corresponding to respective tone ranges
each consisting of three keys in respective memory zones. In these
memories, the memory capacity is reduced by introducing an interpolation
technique to set a relatively coarse waveshape sampling interval. The
waveshape memories 12A and 12B are of the same construction with each
other. The memory 12A reads out a sample point which corresponds to an
address designated by the address generator 25 whereas the memory 12B
reads out a sample point which corresponds to an address next to the
address designated by the address generator 25. Accordingly, the waveshape
memories 12A and 12B produce waveshape amplitude value data D1 and D2 of
sample points which are adjacent to each other. The address signal
generated by the address generator 25 consists of an integer section and a
decimal section and the integer section is supplied to the wavehsape
memories 12A and 12B.
An interpolator 26 interpolates a waveshape amplitude value between the
amplitude value data D1 and D2 of the two sample points read
simultaneously out from the waveshape memories 12A and 12B with a finer
sampling interval in accordance with decimal section data of the address
signal provided by the address generator 25. The interpolation function
for this interpolation can be set as desired. For instance, in the case of
the 1st order interpolation, the interpolation is performed according to
the function
##EQU1##
where D represents interpolated output, a value of the decimal section of
the address signal and N a bit number of the decimal section.
An envelope generator 27 generates, in the same manner as in the envelope
generator 16 in FIG. 1, an envelope signal having a decay characteristic
and, in this embodiment, generates an envelope signal for each channel on
a time shared basis in response to the key-on signal KON. A multiplier 28
imparts the output tone signal of the interpolator 26 with a decay
envelope in response to the output of the envelope generator 27. An
accumulator 29 sums up waveshape amplitude values of the respective
channels for one sample point produced by the multiplier 28 on a time
shared basis. A digital-to-analog converter 18 converts the sum to an
analog value.
In a case where this invention is applied to a polyphonic electronic
musical instrument, address information for the read out of the waveshape
may be generated by counting, in the same manner as in the embodiment of
FIG. 1, the sampling clock corresponding to the relative pitch for each
key in one tone range. An example of a polyphonic electronic musical
instrument is shown in FIG. 5. A clock generator 30 divides the master
clock .phi..sub.0 (e.g. 16 MHz) in frequency to generate a clock pulse
.phi. (e.g. 256 KHz) which establishes a channel time division time slot
and three kinds of clock pulses S.sub.A, S.sub.B and S.sub.C corresponding
to relative pitches of respective keys in one tone range consisting of
three keys. The respective clock pulses S.sub.A, S.sub.B and S.sub.C have
frequencies corresponding to the pitch difference of 100 cents among three
keys in one tone range. The clock pulse S.sub.A corresponds to the highest
tone key (e.g. 32 KHz) of the three keys, the clock pulse S.sub.B to the
middle tone key which is lower than the highest tone key by 100 cents and
the clock pulse S.sub.C to the lowest tone key which is lower than the
highest tone key by 200 cents. The pulse widths of the clock pulses
S.sub.A -S.sub.C correspond to one cycle of the channel time division
timing. If, for example, the number of channels is 8, the pulses S.sub.A,
S.sub.B and S.sub.C have a relationship as shown in FIG. 6 and the pulse
generation timing of the clock pulse S.sub.A always corresponds to the
width of one cycle from the first channel to the eighth channel.
A decoder 31 decodes less significant 2 bits N2 and N1 of the note code
provided from the key assigner circuit 23 and produces signals C1, C2 and
C3 for discriminating the respective keys in one tone range consisting of
three keys. C1 is for the lowest tone key, C2 for the middle tone key and
C3 for the highest tone key.
In the address generator 32, in the same manner as was previously
described, the readout from the waveshape memory 12 is advanced with the
start address being used as an initial value and the address information
is repeatedly changed between the repetitive address and the final
address. This advancing of the address information is performed for each
channel on a time shaped basis employing the clock pulses S.sub.A -S.sub.C
corresponding to the relative pitches of the keys assigned to the
respective channels. More specifically, one of the clock pulses S.sub.A
-S.sub.C is selected in accordance with the signal (one of C1-C3)
representing which of the lowest, middle and highest keys in one tone
range the key assigned to each channel is, and the selected clock pulse is
used as the count clock for advancing the address information in the
corresponding channel. Referring to FIG. 7 for a speciific example, the
selection of the clock pulses S.sub.A -S.sub.C is made by a clock selector
in response to the signals C1-C3 and its output is supplied to an adder
34. The counter is composed of a shift register 35 having the number of
stages corresponding to the number of channels (e.g. 5), the adder 34
adding the output of this shift register 35 and the output of the selector
33, a selector 36 and a gate 37. In a differentiation circuit 38, a key-on
pulse KONP is produced in response to rising of the key-on signal KON of
each channel provided on a time shared basis thereby to close the gate 37
and clear the count value of that channel. Thus, the counting by channel
in the counters 34-37 is started from 0 and, each time "1" is provided by
the clock selector 33 at the timing of that channel, the count of the
channel is increased by 1. The adder 39 adds the count output of the shift
register 35 and the start address data of that channel together and
supplies its output to the waveshape memory 12 as the address imformation.
A comparator 40 compares the count output and the memory length
information of the key assigned to that channel together and, when
coincidence has occurred, B input is selected by the selector 36 and the
repetitive address data of the key assigned to the channel is set at the
shift register 35.
In FIG. 5, accumulators 41A-41C are provided for the respective keys (three
keys) in one tone range. Tone signals of the respective channels provided
by the multiplier 28 are distributed in gates 42A-42C by the highest,
middle and lowest tone keys by the output signals C3-C1 of the decoder 31
and the distributed tone signals are respectively accumulated by the
corresponding accumulators 41A-41C. The accumulation in the accumulators
41A-41C is effected, as was described previously, only in one sample
section (8 channels of first through eighth channels) and renewed each
time the channel cycle is changed. A register 43A receives the accumulated
output for one sample section of the accumulator 41A in synchronism with
the clock pulse S.sub.A and delivers it out in synchronism with the clock
pulse S.sub.A. A FIFO (first-in first-out) register 43B receives the
accumulated output for one sample section of the accumulator 41B in
synchronism with the clock pulse S.sub.A and delivers it out in
synchronism with the clock pulse S.sub.B. A FIFO register 43C receives the
accumulated output for one section of the accumulator 41C in synchronism
with the clock pulse S.sub.A and delivers it out in synchronism with the
clock pulse S.sub.C. Thus, the component of the time division clock .phi.
is removed and the tone signal composite sample point amplitude value
containing components synchronized with the pitch of the tone signal
(i.e., components of the clock pulses S.sub.A, S.sub.B and S.sub.C) is
provided by the registers 43A-43C. The outputs of these registers 43A-43C
are converted to analog signals by digital-to-analog converters 44A-44C
and thereafter are mixed by a mixing circuit 45 and supplied to a sound
system. Accordingly, by adopting the construction of FIG. 5, the influence
of the time division clock .phi. (this clock is generally not synchronized
with the tone pitch) is removed whereby a tone of a good quality can be
composed even in a higher tone range.
In the above described embodiments, the waveshape memories 12, 12A and 12B
store successive waveshapes of the attack portion and a part of the
sustain portion in their respective memory zones. The invention however is
not limited to this arrangement but these memories may store successive
waveshapes of the attack portion the entire sustain portion and the decay
portion with the reading of the waveshape from the start address to the
final address being made only once. Alternatively, successive waveshapes
of the attack portion, a part of the sustain portion and the decay portion
may be stored and the waveshapes of the sustain portion may be produced by
a repetitive reading.
The waveshape memories 12-12B may store successive waveshapes corresponding
to the tone pitch of one key. In this case, each memory zone is read out
in the same manner by the reference rate. The division of the tone range
is not limited to three keys but may be made as desired, e.g., two keys,
four keys, half octave etc.
* * * * *
|
|
 |
|
 |
Description |
|
