Reverberation Tone Generating Apparatus

This invention relates to reverberation tone generating apparatus, and more particularly, reverberation tone generating apparatus wherein a reverberation tone is formed by synthesizing a reverberation tone having irregular delay time and amplitude level and a reverberation tone having regular delay time and amplitude level.

In prior art apparatus for generating a reverberation tone with an electronic circuit, such analog delay element has been used as a BBD (bucket brigade device) or a CCD (charge coupled device). As illustrated in FIG. 1, such delay element includes a feedback loop FBL which feeds back a delayed output signal to the input of a delay circuit DC constituted by a BBD or the like. The delay circuit DC delays by a time DT a signal obtained by adding an input musical signal and the fed back signal and outputs the delayed signal.

With this delay circuit a reverberation tone can be obtained that manifests a regular impulse response determined by the delay time DT of the delay circuit DC and the gain g of the feedback loop FBL, as shown in FIG. 2.

The result of analysis of an actual reverberation tone produced in a concert hall shows that an initial reflected tone (echo) ECH reflected from wall surfaces or a floor surface and having irregular amplitude level and delay time appears firstly and thereafter a reverberation tone RV having regular amplitude level and reverberation time appears as shown by the inpulse response characteristic shown in FIG. 3. Consequently, with the delay circuit shown in FIG. 1 it is impossible to produce a natural reverberation tone obtained in a concert hall.

FIG. 4 shows another prior art circuit in which a plurality of delay circuits DC.sub.1 through DC.sub.n, whose delay times are randomly set, are connected in series, the levels of the outputs of respective delay circuits being independently controlled by level control circuits GC.sub.1 through GC.sub.n and then synthesized by a synthesizing circuit CC to obtain an initial reflected tone ECH and a reverberation tone RV havng impulse response characteristics as shown in FIG. 3.

However, in order to produce a reverberation tone rich in naturality with the circuit shown in FIG. 4, it is necessary to provide a large number of the delay circuits which not only increases the size of the apparatus but also is uneconomical.

Accordingly, it is an object of this invention to provide a novel reverberation tone generating apparatus capable of producing a reverberation tone rich in naturality with a small and economical circuit construction.

Another object of this invention is to obtain a reverberation tone generating apparatus capable of producing a high quality reverberation tone and in which S/N ratio does not lower even when the reverberation tone time is elongated.

Briefly stated according to this invention apparatus for generating a reverberation tone whose delay time and level vary irregularly is combined with apparatus having a feedback loop of an output delayed signal for producing a reverberation tone regularly.

According to this invention, there is provided a reverberation tone generating apparatus comprising a first delay circuit which delays an input musical tone signal for different delay times to produce a plurality of delayed musical tone signals, a level control circuit for independently controlling levels of the plurality of delayed musical tone signals, a second delay circuit having a feedback loop for feeding back an output signal of the second delay circuit to an input side thereof, the second delay circuit synthesizing an input musical tone signal and a delayed output signal thereof at a predetermined ratio and delaying a signal thus synthesized for a predetermined time, and means for synthesizing an output signal of the level control circuit and an output signal of the second delay circuit for outputting a synthesized signal as a reverberation tone signal of the input musical tone signal.

In the accompanying drawings:

FIG. 1 is a connection diagram showing a typical prior art reverberation tone generating apparatus;

FIG. 2 shows the impulse response characteristic of the apparatus shown in FIG. 1;

FIG. 3 shows an impulse response characteristic of a reverberation tone produced in a concert hall or the like;

FIG. 4 is a connection diagram showing another example of prior art reverberation tone generating apparatus;

FIG. 5 is a block diagram showing one embodiment of the reverberation tone generating apparatus embodying the invention;

FIG. 6 is a functional block diagram showing the performances of the embodiment shown in FIG. 5;

FIGS. 7 and 8 are block diagrams showing basic constructions of two types of delay circuits;

FIG. 9 is a timing chart useful to explain the operation of the delay circuit shown in FIG. 7;

FIG. 10 is a graph showing the initial reflected tone (echo) generated in the embodiment shown in FIG. 5;

FIG. 11 is a graph showing the frequency characteristic of a delay circuit having a comb type filter construction;

FIGS. 12 and 13 are graphs showing the characteristics of reverberation tones generated in the embodiment shown in FIG. 5;

FIG. 14 is a block diagram showing the construction of a data memory device utilized in the embodiment shown in FIG. 5;

FIG. 15 is a block diagram showing the construction of a delay length data memory device utilized in the embodiment shown in FIG. 5;

FIG. 16 is a block diagram showing the construction of the address counter utilized in the embodiment shown in FIG. 5; and

FIG. 17 is a functional block diagram showing another embodiment of the reverberation tone adding apparatus according to this invention.

For the sake of description, the basic construction and operation of the delay circuits shown in FIGS. 7 an 8 will be firstly described. Then the process of forming a reverberation tone will be described with reference to the performance block diagram shown in FIG. 6 and finally the preferred embodiment shown in FIG. 5 will be described in detail.

Where amplitude data SPD(t) of an input musical tone signal sequentially sampled at a predetermined sampling period T.sub.o are to be sequentially stored in a digital memory device and an amplitude data SPD (t-i) stored at a time (t-i) is to be read out at a time later by an interval i, an address interval .DELTA.ADR representing a change during the inteval i is added to or subtracted from an address information ADR(t) at a sampling time t according to the following equation (1) or (2) to determine an address information ADR(t-i) at time (t-i), and then the address information ADR(t-i) is applied to the address input of the digital memory device.

Thus, the amplitude data SPD(t-i) stored at time (t-i) can be read out at a time later by i expressed by

In other words, where an address interval .DELTA.ADR corresponding to the desired delay time i is applied as a delay time information it is possible to read out the amplitude data SPD(t-i) stored at the time (t-i) at a time later by the interval i. The equation (1) that determines the address information ADR(t-i) at time (t-i) is applicable to a case where the amplitude data SPD(t) is sequentially stored from a higher order address toward the lower order address as the time elapses. The equation (2) is applicable where the amplitude data SPD(t) is sequentially stored from the lower order address toward the higher order address.

Accordingly, the delay circuit according to this invention comprises, as the fundamental elements, a digital memory device DM sequentially storing the amplitude data SPD(t), an address information generator AG that forms the read address information shown in equation (1) or (2), and a delay length data memory device DDM which generates the address interval .DELTA.ADR as a delay time information DLD.

FIG. 7 shows one example of the delay circuit based on this concept and constituted by the digital memory device DM, the address information generator AG, a delay length data memory device DDM and a multiplier M.

As shown by the timing chart shown in FIG. 9, the digital memory device DM sequentially stores in its memory areas of addresses 0 through 9 the amplitude data SPD(t) sampled at a predetermined period T.sub.o according to a clock pulse .phi. starting from the higher order address 9 toward the lower order address, and is constituted by a random access memory device (RAM) or a shift register.

The designation of the write and read addresses of the amplitude data SPD(t) in the digital memory device DM is effected by the address information generator AG which comprises an address counter AC and an adder AD and forms write address informations ADR(t), ADR(t+1), ADR (t+2), . . . ADR(t+i) whose values are renewed with the sampling time and an read address information ADR(t-i) shown by equation (1) and these write and read address informations are outputted as an address information DM.multidot.ADR for the digital memory device DM. More particularly the address counter AC counts the number of clock pulses having the period T.sub.o to output its count as the write address information ADR(t) of the amplitude data SPD(t) at the present sampling time, and the information ADR(t) is applied to one input of the adder AD. The delay length data memory device DDM supplies a time information DLD (.DELTA.ADR=i/T.sub.o) corresponding to a desired delay time i to the other input of the adder AD. Then the adder performs an arithmetic operation represented by equation (1) at a given sampling time to output the result of addition as a read address information ADR(t-i) of the amplitude data SPD(t-i) before interval i, and then outputs the output information ADR(t) of the address counter AC as the write address information ADR(t) of the amplitude data SPD (t) at the present time, as it is.

In this manner, at time t, the amplitude data SPD(t-i) which was stored at time (t-i) before an interval i is read out from the digital memory device DM, while the amplitude data SPD(t) at the present time t is stored in the area of the address designated by the address information ADR (t).

The amplitude data SPD(t-i) thus read out from the digital memory device DM later by the interval i is multiplied with a coefficient K for controlling the amplitude level in the multiplier M so that the level of the amplitude data is controlled, and the level controlled amplitude data K.multidot.SPD(t-i) is converted into an analog signal by a digital to analog (D/A) converter not shown. Such operation is performed at each sampling time. As a consequence, a reverberation tone i time later than the input musical tone can be produced. In this case, when a plurality of delay time informations DLD which are different from each other at a sampling time are given sequentially, on the time division basis, a plurality of informations regarding reverberation tones having different delay times at the same sampling time can be produced. Accordingly, in this embodiment, the delay circuit shown in FIG. 7 is utilized to form initial reflected tones having complicated reverberation characteristics whose amplitude level and delay time differ depending upon the difference in the distances to the reflecting members such as surrounding walls.

FIG. 8 shows another example of the delay circuit, in which the address counter AC of the address information generator AG is constituted by a preset type down counter. Thus, a delay time information DLD corresponding to a desired delay time i is preset in the address counter AC and the preset value is counted down so as to match the repetition period of the address informations ADR(t), ADR(t+1) . . . ADR(t+i) outputted from the address counter AC with a delay time designated by the delay time information DLD, whereby an amplitude data SPD (t-i) stored before the interval i is read out from an area of an address in which the amplitude data SPD(t) at the present time t is to be stored.

In other words, where the digital memory device DM has 10 words as shown in FIG. 8, the maximum value of the address interval becomes 10 so that it is possible to read out an amplitude data SPD(t-10) delayed a maximum of 10.multidot.T.sub.o. However, where the desired delay time is made to be 6.multidot.T.sub.o, for example, an address in which the address data SPD(t) sampled at the present time t is to be written is matched with an address in which an amplitude data SPD (t-i) before the interval i was stored by making the output information DM.multidot.ADR outputted from the address counter AC to be a repetition of 5, 4, 3, 2, 1; 5 . . . 0 so as to reduce the range of the address utilized in the digital memory device DM, thereby reading out the amplitude data SPD(t-i) written the interval i before from an address in which the amplitude data SPD(t) at the present time is to be written. To this end, in the delay circuit shown in FIG. 8, a maximum value detector MXD is provided for detecting the fact that the output information DM.multidot.ADR from the address counter AC has changed from 0 to 9 and for presetting the delay time information DLD outputted from the delay length data memory device DDM in the address counter AC.

The delay circuit shown in FIG. 8 is constructed such that instead of storing the amplitude data SPD(t) sampled at the present time t in the digital memory device DM as it is, the amplitude data SPD(t-i) before the interval i is fed back at a predetermined ratio so as to write the sum of the fed back value K.multidot.SPD (t-i) and the amplitude data SPD(t) sampled at the present time t. To this end, there are provided a multiplier M which multiplies the amplitude data SPD(t-i) read out from the digital memory device DM before the interval i with a coefficient K and feeds back the multiplied amplitude data to the data input of the digital memory device DM, and an adder AD which adds together the output data K.multidot.SPD(t-i) from the multiplier M and the amplitude data SPD (t) at the present time t and supplies the sum [SPD(t)+K.multidot.SPD(t-i)] to the data input of the digital memory device DM.

Accordingly, with the delay circuit shown in FIG. 8, where the desired delay time i is equal to 6.multidot.T.sub.o, the address counter AC is preset with a delay time information DLD represented by DLD=6-1=5 at a time when the output information DM.multidot.ADR of the address counter AC changes from 0 to the maximum value, in this example 9, whereby the address counter AC repeatedly outputs an address information DM.multidot.ADR which varies as 5, 4, 3, 2, 1, 0, 5, . . . 0 as the sampling time proceeds in each sampling period T.sub.o. At each sampling time, the amplitude data SPD(t-i) stored before the interval i in the area of the address designated by the address information DM.multidot.ADR is firstly read out and then data [SPD(t)+K.multidot.SPD(t-j)] formed by adding together at a predetermined ratio the amplitude data SPD(t-i) and the amplitude data SPD(t) sampled at the present time t is written in the area of the address from which the amplitude value SPD (t-i) has been read out.

Accordingly, with the delay circuit shown in FIG. 8, the address in which the amplitude data SPD(t) at the present time t is written and the address from which the amplitude data SPD(t-i) before an interval i is read out are the same, and the amplitude data SPD(t-i) before the interval i is fed back so that it is possible to take out data regarding a reverberation tone whose amplitude value and the delay time vary regularly. Thus, in this embodiment, the delay circuit shown in FIG. 8 is utilized to generate a reverberation tone following an initial reflected tone (echo) and having a regular reverberation characteristic.

When the amplitude data SPD is multiplied with the coefficient K, the data regarding the finally obtained reverberation tone would have a level larger than that of the original amplitude data. Accordingly, in an actual circuit, the data regarding the reverberation tone is applied to the output side terminal through an attenuator. Where the coefficient K is selected such that -1<K<0, such attenuator is not necessary.

The process of forming the reverberation tone will now be described with reference to the functional block diagram shown in FIG. 6.

The process of forming the reverberation tone in the embodiment shown in FIG. 6 comprises the step of forming an intial reflected tone whose amplitude level and delay time vary randomly and the step of forming a reverberation tone whose amplitude level and the delay time vary regularly. In FIG. 6, the initial reflected tone and the reverberation tone are formed by independent delay circuit systems.

In FIG. 6, the amplitude data SPD(t) obtained by sampling an input musical tone signal at a predetermined period T.sub.o is supplied to a first delay circuit system, that is an initial reflected tone forming unit 1, which utilizes the delay circuit shown in FIG. 7 and made up of a memory device D0 having memory addresses for 2048 words, multipliers M1 through M10 respectively multiplying ten types of the amplitude data SPD(t-i), SPD(t-i.sub.2) . . . SPD(t-i.sub.10) before intervals in (n=1 to 10) which are read out from the memory device D0 at the present sampling time and having different delay times with any amplitude level control coefficient K.sub.n (n=1 to 10), and an adder which adds together the outputs K.sub.1 .multidot.SPD(t-i.sub.1), K.sub.2 SPD (t-i.sub.2) . . . K.sub.10 .multidot.SPD(t-i.sub.10) for producing a total sum ##EQU1## as an initial value ECH (t) of the initial reflected tone at the present time t. The adder SUM contains a register R0 which temporarily stores the sum ##EQU2## until the next sampling time (t+1).

In the initial reflected tone forming unit 1 described above, the amplitude data SPD(t) of the input musical tone at the present time t is written in the area of the address corresponding to the present time t among the memory addresses of the memory device D0 for 2048 words. Since the total sum ##EQU3## at the previous sampling time (t-1) is stored in the register R0 in the adder SUM, the content of this register R0 would be reset. Then, for the purpose of reading out an amplitude data having a delay time of i.sub.1 from the memory device D0 among ten types of the amplitude data SPD(t-i.sub.1) through SPD(t-i.sub.10) before interval in, an address of the memory device D0 corresponding to the delay time i.sub.1 is designated so as to read out from that address the amplitude data SPD(t-i.sub.1) sampled i.sub.1 interval before. The address of the area where the amplitude data SPD(t-i.sub.1) i.sub.1 interval before is read out is calculated by equation (1).

The amplitude data SPD(t-i.sub.1) thus read out and having a delay time i.sub.1 is inputted to the multiplier M1 to be multiplied with an amplitude level control coefficient K.sub.1 corresponding to the first reflected tone ECH.sub.1 having a delay time i.sub.1. The output K.sub.1 .multidot.SPD(t-i.sub.1) of the multiplier M1 is supplied to the adder SUM to be added with the present value of the register R0, and the sum is stored again in the register R0. At this time, since the content of the register R0 has been reset immediately after the writing of the amplitude data SPD(t) at the present time t, the data written into the register R0 at this time is the data K.sub.1 .multidot.SPD(t-i.sub.1).

As above described, when the processings of reading out the amplitude data SPD(t-i.sub.1) having a delay time of i.sub.1 and of the level control are completed, in other words, when the processing regarding the first reflected tone ECH.sub.1 is completed, processing of reading out the amplitude data SPD(t-i.sub.2) regarding the second reflected tone ECH.sub.2 having a delay time of i.sub.2 and of the level control are performed in the same manner as the processing of forming the first reflected tone ECH.sub.1. As a consequence, the sum of the data K.sub.1 .multidot.SPD(t-i.sub.1) regarding the first reflected tone ECH.sub.1 and the data K.sub.2 .multidot.SPD(t-i.sub.2) regarding the second reflected tone ECH.sub.2, that is [K.sub.1 .multidot.SPD(t-i.sub.1)+K.sub.2 .multidot.SPD(t-i.sub.2)] is stored in the register R0 in the adder SUM.

Similar processings are also performed for the third reflected tone ECH.sub.3 through the tenth reflected tone ECH.sub.10. As a consequence, the total sum ##EQU4## of the amplitude data K.sub.1 .multidot.SPD(t-i.sub.1) through K.sub.10 .multidot.SPD(t-i.sub.10) regarding the first reflected tone ECH.sub.1 through the tenth reflected tone ECH.sub.10 would be stored in the register R0, and this total sum ##EQU5## is outputted through a switch circuit SW as the instantaneous value of the initial reflected tone consisting of the first to 10th reflected tones ECH.sub.1 through ECH.sub.10.

As shown in the following Table I, the switch circuit SW selects the output of the register R0 during an interval Ta in which the initial reflected tone is formed, whereas selects and outputs the output of the second delay circuit system at a time Tb following the forming of the initial reflected tone, the sum of Ta and Tb being sampling period To.

The information ECH(t) selected by the switch circuit SW is converted into an analog signal by a D/A converter, not shown, and then supplied to a loudspeaker to be produced as an initial reflected tone for the input musical tone.

Consequently, by making different the delay times in of the first to 10th reflected tones ECH.sub.1 through ECH.sub.10 and the amplitude level control coefficients K.sub.n, it is possible to produce an initial reflected tone whose amplitude level and the delay time vary randomly as shown in FIG. 10.

Where the sampling period T.sub.o of the input musical tone is 0.04 ms (25 KHz) and when an amplitude data SPD(t-1626) stored at an address spaced by 1626 words, for example, from the write address for the amplitude data SPD(t) at the present time t is read out, the delay time i becomes

whereby an initial reflected tone delayed about 65 ms from the input musicl tone can be produced.

The amplitude data obtained by sampling the input musical tone at a predetermined period T.sub.o is also supplied to the second delay circuit system for forming a reverberation tone after forming the initial reflected tone.

This second delay circuit system comprises a delay memory device D10 which supplies to a digital bandpass filter BPF the amplitude data SPD (t) after delaying the same by an interval j, a bandpass filter BPF including a low pass filter LPF and a high pass filter HPF which passes only a predetermined frequency band component of the amplitude data SPD(t-j) delayed by j, a first reverberation tone forming unit 2 of a comb filter and adapted to form a reverberation tone data RVD.sup.1 having a coarse delay time spacing based on the amplitude data SPD(t-j) passed through the band pass filter, and a second reverberation tone forming unit 3 having an all pass filter construction and adpted to form a reverberation tone data RVD.sup.2 having a short delay time spacing based on the reverberation tone data RVD.sup.1.

In the circuit shown in FIG. 6, the amplitude data SPD(t) sampled at the present time t is stored in the area of the address ADR(t) corresponding to the present time t among 2048 memory addresses of the memory device D10. For the purpose of reading out an amplitude data SPD(t-j) sampled before an interval j among a number of amplitude data SPD(t) stored in the memory device 10, an address of the memory device D10 corresponding to the delay time j is designated. The address of the area where the amplitude data SPD(t-j) sampled an interval j before is read out is determined by equation (1) in the case of forming the initial reflected tone. The delay time j at this time is selected to be slightly larger than the delay time i.sub.10 regarding the tenth reflected tone ECH.sub.10, that is j>i.sub.10.

The amplitude data SPD(t-j) having the delay time j thus read out of the memory device D10 is inputted to the multiplier M11 of the low pass filter LPF to be multiplied with a predetermined coefficient K.sub.11, and the product K.sub.11 .multidot.SPD(t-j) is temporarily stored in the register R1. Then, an amplitude data SPD(t-j-1) written one sampling time (1.multidot.T.sub.o) before is read out from the memory device SD0 having memory area of one word and then multiplied with a predetermined coefficient K.sub.12 in the multiplier M12. Then, the output K.sub.12 .multidot.SPD(t-j-1) of the multiplier M12 and the amplitude data K.sub.11 .multidot.SPD(t-j) before interval j temporarily stored in the register R1 are added together. The sum [K.sub.12 .multidot.SPD(t-j-1)+K.sub.11 .multidot.SPD(t-j)] is again temporarily stored in the register R1 and the register R2. Then, the amplitude data SPD(t-j-1) written at a time one sampling time (1.multidot.T.sub.o) before the present time t is again read out from the memory device SD and then multiplied with a predetermined coefficient K.sub.13 in the multiplier M13. The product K.sub.13 .multidot.SPD(t-j-1) thus formed is added to a value [K.sub.12 .multidot.SPD(t-j-1)] temporarily stored in the register R2 and the sum [K.sub.12 .multidot.SPD(t-j-1)+K.sub.11 .multidot.SPD(t-j)+K.sub.13 .multidot.SPD(t-j-1)] is again temporarily stored in the register R2. For the purpose of utilizing the value [K.sub.12 .multidot.SPD(t-j-1 )+K.sub.11 .multidot.SPD(t-j)] temporarily stored in the register R1 in the next sampling period (t+1), this value is stored in the memory device SD0.

By performing these operations at each sampling period T.sub.o an amplitude data SPD(t-j) before time j and removed with high frequency components in a predetermined bandwidth is outputted from the register R2 of the low pass filter LPF and sent to the high pass filter HPF.

Then the high pass filter HPF removes low frequency components in a predetermined bandwidth from the amplitude data SPD(t-j) before interval j in the same manner as in the low pass filter.

Then, the output data SPD(t-j) of the register R2 of the low pass filter LPF is supplied to the multiplier M14 to be multiplied with a predetermined coefficient K.sub.14, and the product K.sub.14 .multidot.SPD(t-j) is temporarily stored in the register R3. Then the amplitude data SPD(t-j-1) written one sampling time (1.multidot.T.sub.o) before is read out from the memory device SD1 having memory area of a single word and multiplied with a predetermined coefficient K.sub.15 in the multiplier M15. The product K.sub.15 .multidot.SPD(t-j-1) thus obtained is added to the amplitude data K.sub.14 .multidot.SPD(t-j) before time j and has been temporarily stored in the register K3 and the sum [K.sub.14 .multidot.SPD(t-j)+K.sub.15 .multidot.SPD(t-j-1)] is temporarily stored in the registers R3 and R4. The amplitude data SPD(t-j-1) written at a time before one sampling time (1.multidot.T.sub.o) than the present time t is again read out from the memory device SD1 and multiplied with a predetermined coefficient K.sub.16 in the multiplier M16 and the product K16.multidot.SPD(t-j-1) is added to [K.sub.14 .multidot.SPD(t-j)+K.sub.15 .multidot.SPD(t-j-1)] stored temporarily in the register R4 and the sum [K.sub.16 .multidot.SPD(t-j-1)+K.sub.14 .multidot.SPD(t-j)+K.sub.15 .multidot.SPD(t-j-1)] is temporarily stored in the register R4. For the purpose of using the value [K.sub.14 .multidot.SPD(t-j)+K.sub.15 .multidot.SPD(t-j-1)] temporarily stored in the register R3 in the next sampling period (t+1), this value is written into the memory device SD1.

These operations are performed in each sampling period T.sub.o so as to produce the amplitude data SPD(t-j) before time j and removed with low frequency components in a predetermined bandwidth from the register R4 of the high pass filter HPF.

Since the register R1 of the low pass filter LPF is not utilized until the next sampling period after writing its content into the memory device SD0, the register R3 of the high pass filter HPF can be used as the register R1.

The amplitude data SPD (t-j) before the interval j and removed with the low and high frequency components in a predetermined bandwidth is inputted to the first reverberation tone forming unit 2.

The first reverberation tone forming unit 2 is provided with three parallelly connected delay circuits 2A, 2B and 2C of the comb filter construction. With a single delay circuit of the comb filter construction the frequency characteristic becomes wavy as shown by A, B and C in FIG. 11 so that three delay circuits 2A, 2B and 2C are connected in parallel. More particularly, parallel connection of three delay circuits 2A, 2B and 2C having different delay times flattens the overall frequency characteristic as shown by D in FIG. 11. The degree of flatness can be improved as the number of parallelly connected delay circuits increase.

In this embodiment, the delay circuit 2A has the longest delay time, the delay circuit 2B has the next delay time, and the delay circuit 2C has the shortest delay time. Although delay circuits 2A, 2B and 2C have different delay times they have the same construction. Accordingly, the construction of only the delay circuit 2A is shown in detail, but delay circuits 2B and 2C are shown only with the reference characters of their multipliers registers and memory devices.

In the first reverberation tone forming unit 2 described above, the amplitude data SPD(t-j) before time j and passed through the bandpass filter BPF is multiplied with an amplitude level control coefficient K.sub.17 in a multiplier K.sub.17. The product K.sub.17 .multidot.SPD(t-j) thus produced is temporarily stored in a register R5 in the multiplier M17. For the purpose of reading out amplitude data SPD(t-x.sub.1) written in a memory device D1 having memory addresses for 2048 words x.sub.1 time before, an address of the memory device D1 corresponding to the delay time x.sub.1 is designated. The read out amplitude data SPD(t-x.sub.1) is applied to an adder SUM where it is added to the outputs of other memory devices D2 and D3 and to the outputs of the memory devices D4 through D6 and D7 through D9 of the delay circuits 2B and 2C, and the sum is temporarily stored in a register R11 in the adder SUM. In this case, the reading operations of the memory devices D1 through D9 are sequentially performed on the time division bases in the order of from D1 to D9. Accordingly, during the reading of the memory device D1, no data is outputted from other memory devices D2 through D9. As a consequence, the data written into the register R11 in the adder SUM is the data SPD(t-x.sub.1) read out from the memory device D1.

The amplitude data SPD(t-x.sub.1) read out from the memory device D1 is multiplied with an amplitude level control coefficient K.sub.13 in a multiplier 18 and then fed back to the input side of the memory device D1. The product K.sub.13 .multidot.SPD(t-x.sub.1) is added to data K.sub.17 .multidot.SPD(t-j) temporarily stored in the register R5 at the present time t and the sum [K.sub.17 .multidot.SPD(t-j)+K.sub.18 .multidot.SPD(t-x.sub.1)] in temporarilly stored in a register R6. Then the amplitude data [K.sub.17 .multidot.SPD(t-j)+K.sub.18 .multidot.SPD(t-x.sub.1)] stored in the register R6 is written into the same address which is storing the amplitude data SPD(t-x.sub.1) before time x.sub.1. Thereafter, the content of the register R6 is reset. The reason for resetting the register R6 is to use this register for the processing of the system including the memory device D2 in the next stage.

Upon completion of the processing of the system including the memory device D1, the processing of the system including the memory device D2 is executed in the same manner.

More particularly, for the purpose of reading out the amplitude data SPD(t-x.sub.2) written x.sub.2 time before, into the memory device D2 having 2048 word addresses an address of the memory device D2 corresponding to the delay time x.sub.2 is designated, thereby to read out the amplitude data SPD(t-x.sub.2) sampled x.sub.2 time before from the memory device D2. This read out amplitude data SPD(t-x.sub.2) is added to the content SPD(t-x.sub.1) of a register R11 (the content read out from the memory device D1) by the adder SUM and the sum [SPD(t-x.sub.1)+SPD(t-x.sub.2)] is temporarily stored in the register R11.

The amplitude data SPD(t-x.sub.2) read out from the memory section D2 is multiplied with an amplitude level control coefficient K.sub.18 in a multiplier 19 and then fed back to the input side of the memory device D2. The product K.sub.19 .multidot.SPD(t-x.sub.2) is added to the value K.sub.17 .multidot.SPD(t-j) temporarily stored in a register R5, and the sum [K.sub.17 .multidot.SPD(t-j)+K.sub.19 .multidot.SPD(t-x.sub.2)] is temporarily stored in a register R6. The data [K.sub.17 .multidot.SPD(t-j)+K.sub.19 .multidot.SPD(t-x.sub.2)] to be stored in the register R6 is stored in the same address storing the data SPD(t-x.sub.2), x.sub.2 time before. Thereafter, the content of the register R6 is reset.

Thereafter, the processing of the system including the memory section D3 is executed in the same manner as that of the system including the memory device D2.

Denoting the delay time of the system including the memory device D3 by x.sub.3, at the time of completing the processings of the systems including memory sections D1, D2 and D3, the data to be stored in the register R11 is expressed by

whereas the data to be stored in the memory device D3 is expressed by

Similar processings are executed in the delay circuits 2B and 2C.

Denoting the delay times of the systems including memory sections D4, D5 and D6 of the delay circuit 2B by x.sub.4, x.sub.5 and x.sub.6 respectively, and the delay times of the systems including memory sections D7, D8 and D9 of the delay circuit 2C by x.sub.7, x.sub.8 and x.sub.9 respectively, then the content of the register 11 at the time when all processings of the delay circuits 2A, 2B and 2C have completed is expressed by the following equation ##EQU6##

Consequently, following the initial reflected tone, a reverberation tone can be obtained having a long delay time and in which the amplitude level and the delay time vary regularly as shown in FIG. 12, in which the reverberation tone of the delay circuit 2A alone is depicted for the sake of simplicity.

The reverberation tone data RVD.sup.1 thus formed and having a long delay time interval is supplied to the second reverberation tone forming unit 3.

The second reverberation tone forming unit 3 is provided with serially connected delay circuits 3A, 3B and 3C of the all pass type filter construction having a flat frequency characteristic.

The three delay circuits 3A, 3B and 3C are connected in series to form a reverberation tone data RVD.sup.2 having a shorter delay time interval than the reverberation tone data RVD.sup.1 formed by the first reverberation tone forming unit 2. For ths reason, the delay times of the delay circuits 3A, 3B and 3C of the second reverberation tone forming unit 3 are set to be shorter than the delay times of the delay circuits 2A, 2B and 2C of the first reverberation tone forming unit 2. The delay circuits 3A, 3B and 3C are set with different delay times but have the same construction. Accordingly, the construction of only the delay circuit 3A is shown in detail but delay circuits 3B and 3C are shown with the reference characters of their multipliers, registers and memory devices.

The reverberation tone data RVD.sup.1 outputted from the second reverberation tone forming unit 2 is supplied to a register R12 of the delay circuit 3A, but prior to store this data RVD.sup.1 in the register R12, for the purpose of reading out y.sub.1 time before data RVD.sup.1 (t-y.sub.1) written into a memory section MD0 having 512 word memory addresses, an address of the memory device MD0 corresponding to the delay time y.sub.1 is designated, thus reading out the data RVD.sup.1 (t-y.sub.1) from the memory device MD0 written before the time y.sub.1. The data RVD.sup.1 (t-y.sub.1) is multiplied with an amplitude level control coefficient K.sub.30 in a multiplier M30, and the product K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1) is fed back to the input side of the memory section MD0. Then the fed back data K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1) is added to data RVD.sup.1 (t) supplied from the first reverberation tone forming unit 2 at the present time and the sum [RVD.sup.1 (t)+K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)] is temporarily stored in the register R12. Thereafter, the address of the memory section MD0 corresponding to the delay time y.sub.1 is designated again and the data RVD.sup.1 (t-y.sub.1) written y.sub.1 time before is again read out from the memory section MD0. The read out data RVD.sup.1 (t-y.sub.1) is temporarily stored in the register R13. Then the data [RVD.sup.1 (t)+K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)] temporarily stored in register R12 is multiplied with an amplitude control coefficient K.sub.20 in a multiplier 29 and the product K.sub.29 .multidot.[RVD.sup.1 (t)+K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)] is added to a value RVD.sup.1 (t-y.sub.1) temporarily stored in the register R13. The sum RVD.sup.1 (t-y.sub.1)+K.sub.29 .multidot.[RVD.sup.1 (t)+K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)] is temporarily stored in the register R13. For the purpose of utilizing the data [RVD(t)+K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)] temporarily stored in the register R12 at a sampling time (t+y.sub.1) later than the present time t by an interval y.sub.1, the data [RVD.sup.1 (t)+K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)] is stored in the address in which the data RVD.sup.1 (t-y.sub.1) was stored.

When the processing executed by the delay circuit 3A is completed, the data RVD.sup.1 (t-y.sub.1)+K.sub.29 .multidot.[RVD.sup.1 (t)+K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)] is sent to the delay circuit 3B in which this data is processed in the same manner as in the delay circuit 3A.

Denoting the output data from the delay circuits 3A, 3B and 3C by RVD.sup.2A, RVD.sup.2B and RVD.sup.2C respectively and denoting the delay time of the delay circuit 3B by y.sub.2, and the delay time of the delay circuit 3C by y.sub.3, then the output data of the registers R13, R15 and R17 of the delay circuits 3A, 3B and 3C can be expressed by the following equations (4), (5) and (6).

The output data RVD.sup.2C of the delay circuit 3C is outputted via a switch circuit SW as data for producing a reverberation tone following the initial reflected tone.

Where the relation among the delay times of the delay circuits 3A, 3B and 3C is selected as y.sub.1 >y.sub.2 >y.sub.3 it is possible to form a reverberation tone having a short delay time spacing as shown in FIG. 13. More particularly, based on the reverberation tone data RVD.sup.1 formed by the first reverberation tone forming unit 2 and having a long delay time spacing, the delay circuit 3A forms a first reverberation tone data RVD.sup.2A having a spacing shorter than the delay timne spacing of the first reverberation tone forming unit 2, while the delay circuit 3B forms a second reverberation tone data RVD.sup.2B having a spacing y.sub.2 shorter than the delay time spacing y.sub.1 of the delay circuit 3A. For this reason, as the forming processings of the reverberation tones by the delay circuits 3A, 3B and 3C proceed, reverberation tones having shorter delay time spacings would be formed.

Since the registers R12, R14 and R16 in the delay circuits 3A, 3B and 3C are not used until the next sampling period, once the processing executed by them are completed they can be used commonly on the time division basis.

Obviously, in the delay circuits 3A, 3B and 3C, the multiplier M29 may directly receive the data RVD' or the output of the first reverberation forming unit 2 as shown at dotted line and similarly, the multiplier M30 may be connected to receive the output of the register R13.

The detail of the construction and operation of the embodiment shown in FIG. 5 will now be described. In the following description, it is assumed that the circuit shown in FIG. 5 forms the reverberation tone according to the performances described in connection with FIG. 6.

The reverberation tone generating apparatus of the embodiment shown in FIG. 5 generally comprises a memory unit 10, a time information generator 20, an address information generator 30 and a calculating unit 40.

The memory unit 10 corresponds to the delay digital memory device DM shown in FIG. 8 and constituted by a data memory device 100 having a plurality of memory blocks and a latch circuit 101. By utilizing the plurality of memory blocks there are formed memory sections SD0 through SD15 each for one word (16 bits), memory sections MD0 through MD15 for 512 words (each 16 bits), and memory sections D0 through D15 for 2048 words (each 16 bits) as shown in FIG. 14. The data to be stored in these memory sections SD0 through Sd15, MD0 through MD15 and D0 through D15 are given from the calculating unit 40, and a data storing address and a data read out address are designated by address informations DM.multidot.ADR outputted from an address information generator 30. Data read out from respective memory sections SD0 through SD15 are supplied to the calculating unit 40 via the latch circuit 101.

The time information generator 20 corresponds to the delay length data memory device DDM and comprises a parameter designating circuit 200 and a delay length data memory device 201. The delay length data memory device 201 is constructed to select and output either one of the delay time informations DLD.sup.m (n) (where n designates memory sections D0 through D15 and MD0 through MD15, and m designates types 1 through 8) relating to respective data delay memory sections D0 through D15 and MD0 through MD15 respectively corresponding to 8 types of the reverberation tones (including the initial reflected tone) having different reverberation characteristics in accordance with a designation from the parameter designating circuit 200. More particularly, as shown in FIG. 15, the delay length data memory device 201 comprises memory blocks MB(D0) through MB(D15); MB(MD0) through MB(MD15) respectively corresponding to the data delay memory sections D0 through D15 and MD0 through MDL15. Each of the memory blocks MB(D0) through MB(MD15) comprises 8 memory addresses 0 through 7 corresponding to the 8 types of the reverberation tones. Respective memory addresses 0 through 7 of the memory blocks MB(D0) through MB(MD15) prestore different ones of the delay time informations DLD.sup.1 (D0) through DLD.sup.8 (D0), DLD.sup.1 (D1) through DLD.sup.8 (D1) . . . DLD.sup.1 (D15) through DLD.sup.3 (D15), DLD.sup.1 (MD0) through DLD.sup.8 (MD0), . . . DLD.sup.1 (MD15) through DLD.sup.8 (MD15). 3 bit parameter designation information PSL designating the reverberation tone characteristic of a reverberation tone to be generated is supplied from the parameter designating circuit 200 as a lower order address information when a bit memory number information DLn(n=0 through 15) that designate one of the memory numbers 0 through 15 of the memory sections MD0 through MD15 and a 2 bit memory type information DLK (k=D, MD, SD) that designates the type D, MD and SD of the memory sections are supplied from the address information generator 30 as upper order address informations, the delay time information DLD.sup.m (n) which has been stored in a memory address (one of 0 through n) designated by the information DSL in a memory block (one of MB(D0) through MB(MD15) designated by the informations DLn and DLk is read out and supplied to the address information generator 30 as an information that determines the delay time relation of a reverberation tone having a desired reverberation characteristic designated by the parameter designating circuit 200. The delay time of the memory sections SD0 through Sd15 is fixed to 1.multidot.T.sub.o, so that any delay time information is not necessary for these memory sections SD0 through Sd15. In addition to the parameter designation information PSL, the parameter designating circuit 200 produces a 3 bit program selection information PGS that selects desired one of the control programs prepared for forming 8 types of the reverberation tones.

Based on the delay time information DLD.sup.m (n) and the program selection information PGS outputted from the time information generator 20 and a master clock pulse .phi..sub.o that determines the one step period of the control program, the address information generator 30 produces an address information DM.multidot.ADR for the data memory device 100 necessary to form a reverberation tone of a desired reverberation characteristic and various control signals for controlling the operations of various circuits. The address information generator 30 comprises a program memory device 300, a program counter 301, a program decoding memory device 302, a control signal output register 303, a selector 304, an address counter 305, a latch circuit 306, a subtractor 307, a maximum value detector 308 and an address information output circuit 309.

8 types of the control programs are prestored in the program memory device 300 for forming 8 types of reverberation tones having different reverberation characteristics, and which one of the control programs is to be outputted is designated by a program selection information PGS outputted from the parameter designation circuit 200. The content of the designated program is sequentially read out at each step by the output information PC of the program counter 301 which counts the number of the master clock pulses .phi..sub.o.

In order to complete in one sampling period T.sub.o all processings of the initial reflected tone forming unit 1, the bandpass filter BPF, the first reverberation tone forming unit 2 and the second reverberation tone forming unit 3, when the sampling frequency is selected to be 25 KHz, and the frequency of the master clock pulse .phi..sub.o to be 4.8 MHz, then the number of steps of one control program becomes less than 4800/25=192 and the content of the control program having 192 steps is executed at each sampling period T.sub.o. As shown in the following Table II, as the control programs at respective steps, three types of contents are prepared, that is first, second and third types in which one step is constituted by a 16 bit information. The forming of the initial reflected tone, filter processing and the forming of the reverberation tone are implemented by approximately combining the sequence of these three type control programs and the contents of each bit information.

In this example, the one step control programs each consisting of 16 bits can be classified into two types, one outputted through the control signal output register 303 as they are as informations OF.multidot.ADRn, RGn, DLn, and ADR(kn) and the other outputted through the control signal output register after being decoded by the program decoding memory device 302 as the memory write control signal WR1, the latter type signal being applied to the program decoding memory device 302 from the program memory device 300 to act as an operation code OPC. The content of Table II will be described later in detail together with the operation.

As shown in FIG. 16, the address counter 305 comprises address counters AC(D0) through AC(D15), AC(MD0) through AC(MD15) respectively corresponding to delay memory sections D0 through D15, MD0 through MD15. Respective counters AC(D0) through AC(D15) and AC(MD0) through AC(MD15) of the address counter 305 are selectively actuated by a memory number information DLn and a memory type information DLk. The count output informations ADR(n) of the address counters AC(n)(n: D0 through D15, MD0 through MD15) which are actuated by informations DLn and DLk are supplied to the address information output circuit 309 through the latch circuit 306 and also to the subtractor 307. In this example, the output informations ADR(n) of the address counters AC(n) is constituted by 11 bits so that they can designate an address range up to 2048 words, because memory sections D0 through D15 among the memory sections D0 through D15 and MD0 through MD15 are constructed to have an address information length of 2048 words. The address counter 305 is constituted by a RAM.

The subtractor 307 subtracts [1] from the output content ADR(n) of the address counters AC(n) inputted via the latch circuit 306 and feeds back the difference [ADR(n)-1] to the A input of the selector 304 for the purpose of using the difference in the next sampling period (t+1). At the same time, the difference is also supplied to the maximum detection circuit 308 which corresponds to the detector MXD shown in FIG. 8. When the maximum value detection circuit 308 detects the fact that an information [ADR(n)-1] obtained by subtracting [1] from the output information ADR(n) of the address counter AC(n) designated by the memory number information DLn and the memory type information DLk has reached the maximum value (all bits are "1"), the maximum value detection circuit 308 applies a selection control signal SLB to the selector 304 causing the same to select the input B. The output information [ADR(n)-1] of the subtractor 307 is inputted to the input A of the selector 304, and the output information DLD.sup.m (n) of the delay length data memory data memory device 201 is inputted to the input B of the selector 304 so that its output is supplied to one input of the address counter 305 so as to be written (preset) in an address counter AC(n) designated by informations DLn and DLK in accordance with a write control signal WR3. Consequently, under a condition in which the maximum value detection circuit 308 does not produce any selection control signal SLB, a value [ADR(n)-1] obtained by subtracting [1] from the present value ADR(n) would be written in the address counter AC(n) designated by informations DLn and DLk, at each sampling period, so that the output information ADR(n) of the address counter AC(n) decreases toward zero as the time elapses. However, when the value [ADR(n)-1] reaches the maximum value, the maximum value detection circuit 308 produces a selection control signal so that a delay time information DLD.sup.m (n) is applied to the address counter AC(n) via the selector 304 and written into the address counter AC(n) in accordance with the write control signal WR3. Consequently when the selector control signal SLB is generated, the content of the address counter AC(n) becomes DLD.sup.m (n) and then sequentially changes toward zero as the sampling time elapses. In other words, in a portion constituted by a selector 304, an address counter 305, the latch circuit 306, the subtractor 307 and the maximum value detection circuit 308, the address counter AC(n) designated by the informations DLn and DLk forms an address information ADR(n) that completes one cycle with a period equal to a delay time corresponding to the delay time information DLD.sup.m (n). The address information ADR(n) is supplied to the address information output circuit 309.

The purpose of the address information output circuit 309 is to output address informations for reading out and writing informations into the memory sections SD0 through SD15, D0 through D15 and MD0 through MD15. Where an information delayed by an interval in is read out from the memory section D0 to form an initial reflected tone ECH(t), the address information output circuit 309 formes one set of informations OF.multidot.ADRn, DLn, and DLk and outputs this set as an address information DM.multidot.ADR by utilizing an 11 bit address information OF.multidot.ADRn corresponding to respective delay times in of the first to 10th reflected tones ECH.sub.1 through ECH.sub.10 (outputted by the control signal output register 303) as a lower order address information and then adding a memory number information DLn and a memory type information DLk. Where an amplitude data SPD(t) sampled at the present time is to be written into the memory section D0, the address information output circuit 309 outputs a set of informations ADR(D0), DLn and DLk as an address information DM.multidot.ADR, the set being formed by utilizing the output information ADR(D0) of the address counter AC(D0) corresponding to the memory section D0 as a lower order address information and then adding informations DLn(=DL0) and DLk(=DL0) that designate the memory section D.sub.o to the upper order. When an amplitude data is written into and read out from the memory sections SD0 through SD15, all bits of a lower order address information are made "0" and informations DLn(=DL.sub.0 to DL.sub.15) and DLk (=DL.sub.SD) designating memory sections SD0 through SD15 are added to an upper order to form and output an address information DM.multidot.ADR. Where reverberation tones RVD.sup.1 and RVD.sup.2 are to be formed, the output informations ADR(D1) through ADR(D15) and ADR(MD0) through ADR(MD15) of respective address counters AC(D1) through AC(D15) and AC(MD0) through AC(MD15) respectively corresponding to memory sections D1 through D15 and MD0 through MD15 are utilized as the lower order informations and informations DLn and DLk are added to their upper orders. These one set of informations ADR(n), DLn and DLk are outputted as an address information DM.multidot.ADR. In this case, at a time when the information OF.multidot.ADRn is to be added to the lower orders of the information DLn and DLk, the control signal output register 303 outputs a control pulse GP1. When all bits of the lower address information to be added to the lower orders of the informations DLn and DLk are to be made to "0", the control signal output register 303 produces a control pulse GP2. The address information output circuit 309 contains therein a register that temporarily stores informations DLn and DLk.

The purpose of the calculating unit 40 is to effect amplitude level control of the data to be stored in memory sections D0 through D15, MD0 through MD15 and SD0 through SD15 and of the data read out from respective memory sections. The calculating unit 40 comprises a coefficient memory device 400, a selector 401, a calculating or operation circuit 402, a temporary register 403 and a latch circuit 404.

Similar to the delay length data memory device, the coefficient memory device 400 includes 8 memory blocks corresponding to 8 types of the reverberation tones having different reverberation characteristics and respective memory blocks prestore a set of coefficients Kn (n=1 to 32) necessary to form reverberation tones of different types. When supplied with a parameter designation information PSL from the parameter designation circuit 200, and an address information ADR(Kn) designating the coefficient Kn, among the memory blocks designated by the information PSL, a coefficient Kn is read out from an address designated by the information ADR(Kn) and supplied to an input A of the calculating circuit 402.

The amplitude data SPD(t) of the input musical tone sampled by a samle and hold circuit SPH is inputted to the input A of the selector 401 data MRD read out from the memory device 10 is inputted to the input B and the output data RGD of the temporary register 403 is supplied to the input C via the latch circuit 404. Either one of these input data SPD(t), MRD and RGD is selected by a selection control signal SL1 (2 bits) and then applied to the input X of the calculating circuit 402.

A coefficient Kn read out from the coefficient memory device 400 is applied to the input A of the calculating circuit 402, and the output data RGD from the temporary register 403 is inputted to the input B through the latch curcuit 404 and data SPD(t), MRD, RGD selected by the selector 401 are applied to the input X so that the calculating circuit 402 performs the following calculations in accordance with a calculation control signal CTL (3 bits) outputted from the control signal output registre 303.

The results (Y) of calculations are supplied to the temporary register 403, the memory device 10 and the output register 500.

The temporary register 403 temporarily stores the values calculated by the calculating circuit 402 while the initial reflected tone ECH(t) and the reverberation tones RVD.sup.1 and RVD.sup.2 are being formed, and feeds back its content to the input C of the selector 401 and the input B of the calculating circuit 402 to act as the register output data RGD. The temporary resistor 403 has 32 registers R0 through R31 designated by register designation informations RGn (n=1 to 32) of a 5 bit construction and the input data are written into registers R0 through R31 designated by the informations RGn under the control of the write control signal WR1.

The output resistor 500 stores the instantaneous value ECH(t) of the initial reflected tone obtained as a value Y calculated by the calculating circuit 402 and the instantaneous value RVD(t) of a reverberation tone following the initial reflected tone under the control of a control signal WR2 and supplies the data thus stored to a D/A converter 502 via an attenuator 501.

The selection control signal SL1 applied to the calculation control signal CTL applied to the calculation circuit 402 are contained in an operation code OPC outputted from the control signal output register 303.

The reverberation tone generating apparatus described above operates as follows.

(a) Formation of the initial reflected tone FCH(t).

(1) For the purpose of writing the amplitude data SPD(t) of an input musical tone sampled at the present time t into the memory device D0, a selection control signal SL1 and the calculation control signal CTL respectively represented by

SL1: SELECT(A)

CTL: (Y)=(X)

are outputted from the control signal output register 303 as an operation code OPC, whereby the selector 401 supplies the amplitude data SPD(t) outputted from the sample and hold circuit SPH to the input X of the calculating circuit 402. The calculating circuit 402 outputs the amplitude data SPD(t) inputted to its input A as a calculated value (Y).

(2) Then after an address of the memory device D0 corresponding to the present sampling time t has been designated, for the purpose of writing the output data SPD(t) of the calculating circuit 402 into the designated address a memory type information DLk, a write control signal WR4 and a latch signal L3 respectively represented by

DLn: DL.sub.0

DL.sub.K : DL.sub.D

WR4: "1" (WRITE)

L3: "1" (LATCH)

are outputted from the control signal output register 303 together with a memory number information DLn.

Accordingly, the output information ADR(D0) of the address counter AC(D0) corresponding to the memory device D0 is latched by the latch circuit 306 as a lower order address information for writing the amplitude data SPD(t) at the present time t. In the address information output circuit 309, to the upper order of the lower order address information ADR(D0) is added the memory number information DLN (=DL.sub.0) and the memory type information DLk (=DL.sub.D) to form a write address information DM.multidot.ADR of the amplitude data SPD(t) for the memory device D0, and the formed write address information is outputted. As a consequence the amplitude data SPD(t) at the present time t applied to the data input of the memory device D0 via the calculating circuit 402 is written into an address corresponding to the present time t by the write control signal WR4.

(3) Then, for the purpose of clearing the register R0 storing the synthesized value of the initial reflected tones at respective sampling times, an operation control signal CTL, a write control signal WR1 acting as operation codes and respectively expressed by

CTL: (Y)=0

WR1: "1" (WRITE)

are outputted from the control signal output register 303 together with a register number information RGn expressed by RGn=R0.

Consequently [0] is written into the register R0. In other words, the register R0 is cleared.

(4) Then, for the purpose of forming the first reflected tone ECH.sub.1, a memory type information DLk, a control pulse GP1 and a latch control signal L2 acting as operation codes and respectively expressed by

DLk: DL.sub.D

GP1: "1"

L2: "1" (LATCH)

are outputted from the control signal output register 303 together with an address information OF.multidot.ADRn=OF.multidot.ADR.sub.1 corresponding to the delay time i.sub.1 of the first reflected tone ECH.sub.1. In this case, the address information output circuit 309 is storing the memory number information DLn (=DL.sub.0) at the step (3).

Consequently, the address information output circuit 309 outputs an address information DM.multidot.ADR for reading out from the memory device D0 the amplitude data SPD(t-i.sub.1) written before an interval i.sub.1 by utilizing the address information OF.multidot.ADR.sub.1 corresponding to the delay time i.sub.1 as a lower order information, and by utilizing the memory number information DLn (=DL.sub.0) and the memory type information DLk (=DL.sub.D) as the upper order address information, whereby an amplitude data SPD(t-i.sub.1), an interval i.sub.1 before is read out from the memory device and the amplitude data SPD(t-i) thus read out is latched by the latch circuit by a latch control signal L2.

(5) For the purpose of transferring the present value of the register R0 to the latch circuit 404 a latch control signal L1="1" (LATCH) acting as an operation code and a register number information RGn=R0 are outputted from the control signal output register 303, whereby the present value of the register R0 is transferred to the latch circuit 404 to be stored therein.

(6) Then for the purpose of obtaining an instantaneous value K.sub.1 .multidot.SPD(t-i.sub.1) regarding the first reflected tone ECH.sub.1 by multiplying an amplitude data SPD(t-i.sub.1) before time i.sub.1 with an amplitude level controlling coefficient K.sub.1, a selected control signal SL1=SELECT(B) and an calculation control signal CTL=(A).multidot.(X)+(B)=(Y) which constitute and operation code are outputted from the control signal output register 303 together with a constant reading out address information ADR (Kn)=ADR.multidot.(K.sub.1).

Consequently, a coefficient K.sub.1 regarding the first reflected tone ECH.sub.1 is read out from the coefficient memory device 400 and supplied to the input A of the calculating circuit 402. On the other hand, the selector 401 selects the amplitude data SPD(t-i.sub.1), i.sub.1 time before and supplied to its input B from the latch circuit 101 and applies the selected data SPD(t-i.sub.1) to the input X of the calculating circuit 402 which performs the following calculation.

Since the content of the register R0 has been cleared at step (3) described above, at this time, an instantaneous value K.sub.1 .multidot.SPD(t-i.sub.1) regarding the first reflected tone ECH is obtained as the calculated value (Y) of the calculation circuit 402.

(7) Then, for the purpose of transferring the instantaneous value K.sub.1 .multidot.SPD(t-i.sub.1) of the first reflected tone ECH.sub.1 to the register R0 and to store therein a write control signal WR1="1" (WRITE) acting as the operation code OPC is, outputted from the control signal output register 303 together with a register number information RGn=R0, whereby the output data (Y)=K.sub.1 .multidot.SPD(t-i.sub.1) of the calculation circuit 402 is written into the register R0.

When various steps described above are completed, the instantaneous value K.sub.1 .multidot.SPD(t-i.sub.1) of the first reflected tone ECH.sub.1 can be obtained.

(8) Then the instantaneous values K.sub.2 .multidot.SPD(t-i.sub.2) through K.sub.10 .multidot.SPD(t-i.sub.10) respectively regarding the second to 10th reflected tones ECH.sub.2 through ECH.sub.10 are formed at steps (4) through (7) as a consequence at a time when the step regarding the 10th reflected tone ECH.sub.10 has completed, the register R0 stores the total sum ##EQU7## of the instantaneous values of the first to 10th reflected tones ECH.sub.1 through ECH.sub.10, and the total sum is written into the output register 500 by the write control signal WR2 and then transferred to the attenuator 501.

(b) Filter operation

(1) For the purpose of reading out from the memory device D10 the amplitude data SPD(t-j), j time before, a memory type information DLk=DL.sub.D, latch control signals L3="1" (LATCH) and L2="1" (LATCH) which constitute the operation code OPC are outputted from the control signal output register 303 together with a memory number information DLn=DL.sub.10. From the address counter AC(D10) corresponding to the memory section D10 is latched by the latch circuit 306 as a lower order address information for reading out the amplitude data SPD(t-j), j time before. The lower order address information ADR(D10) thus latched is added to its upper order the memory number information DLn (=DL.sub.10) and the memory type information DLk (=DL.sub.D) in the address information output circuit 309 to form a read address information DM.multidot.ADR for reading out the amplitude data SPD(t-j) from the memory section D10 of the data memory device 100, whereby the amplitude data SPD(t-j), an interval j before, is read out from the memory section D10 and the read out data is latched by the latch circuit 101 according to the latch control signal L2.

(2) For the purpose of writing the amplitude data SPD (t) sampled at the present time t into the same address from which the amplitude data SPD(t-j) has been read out, a selection control signal SL1=SELECT(A) and a calculation control signal CTL=(Y)=(X) which constitute the operation code are outputted from the control signal output register 303. Consequently, the selector 401 supplies to the input X of the calculation circuit 402, amplitude data SPD(t) outputted from the sample and hold circuit SPH. Furthermore, the calculation circuit 402 outputs the amplitude data SPD(t) inputted to its input X as a calculated value (Y).

(3) For the purpose of writing the amplitude data SPD(t) into the memory section D10 a memory type information DLk=DL.sub.D, a write control signal WR4="1" (WRITE), and a latch control signal L3="1" (LATCH) which constitute the operation code OPC, and a memory number information DLn=DL10 are outputted from the control signal output register 303. Accordingly, the output information ADR(D10) of the address counter AC(D10) corresponding to the memory section D10 is latched by the latch circuit 306 as a lower order address information for writing the amplitude data SPD(t) at the present time t. In the address information output circuit 309, to the lower order address information ADR(D10) thus latched are added the memory number information DLn (=DL.sub.10) and the memory type information DLk (=DL.sub.D) to form and output an address information DM.multidot.ADR for writing the amplitude data SPD(t) in the memory section D10. As a consequence, the amplitude data SPD(t) at the present time t applied to the data input of the memory section D10 via the calculation circuit 402 is written into an address corresponding to the present time t by the write control signal WR4.

(4) Then, in the low pass filter LPF, the following equation

is calculated according to the content of the register R1, the coefficient K.sub.11 and the amplitude data SPD(t-j), j time before. For storing again the calculated value in the register R1, a latch control signal L1="1" (LATCH) acting as the operation code OPC and a register number control signal RGn=R1 are outputted from the control signal output register 302 and the content of the register R1 is transferred to the latch circuit 404.

(5) For the purpose of calculating K.sub.11 .multidot.SPD(t-j), the control signal output register 303 outputs a selection control signal SL1 (SELECT (B)) and a calculation control signal CTL which constitute the operation code OPC and a constant read out address information ADR (Kn).

Consequently, a coefficient K.sub.11 is read out from the coefficient memory device 400 and supplied to the input A of the calculating circuit 402. The selector 401 selects the amplitude data SPD(t-j) which was latched in the latch circuit 101 at the preceding step b-(1) and supplies the selected data SPD(t-j) to the input X of the calculation circuit 402. Accordingly, the calculation circuit 402 calculates the following equation ##EQU8## At this time, since the content of the register R1 has been cleared at a time when the filtering processing at the previous sampling time (t-1) has completed, data K.sub.11 .multidot.SPD(t-j) is obtained as the calculated value (Y) at this step.

(6) For the purpose of storing this calculated value (Y)=K.sub.11 .multidot.SPD(t-j) in the register R1, the control signal output register 303 outputs a write control signal WR1="1" (WRITE) utilized as the operation code OPC and a register number information RGn=R1, whereby the output data K.sub.11 .multidot.SPD(t-j) of the calculation circuit 402 is stored in the register R1.

(7) Then for the purpose of reading out the amplitude data SPD(t-j-1) interval before, the control signal output register 303 outputs a memory type information DL.sub.K =DL.sub.SD, a latch control signal L2="1" (LATCH), and a gate pulse signal GP2="1" which constitute the operation code OPC, and a memory number signal DLn=DL.sub.0. Then the address information output circuit 309 changes to "0" all bits of the lower order address information and adds the memory type information DLk (=DL.sub.SD) and the memory number information DLN (=DL.sub.o) to the upper order to form and output an address information DM.multidot.ADR for the memory section SD0, whereby the amplitude data SPD(t-j-1), a (j-1) time before, is read out from the memory section SD0 and latched by the latch circuit 101.

(8) Then an equation

is calculated in accordance with the content K.sub.11 .multidot.SPD(t-j) of the register R1, the coefficient K.sub.12, and the amplitude data SPD(t-j-1) latched in the latch circuit 101, and for the purpose of storing again the calculated value in the register R1, the control signal output register 303 outputs a latch control signal L1="1" (LATCH) acting as the operation code OPC and a register number information RGn=R1 so as to transfer the content K.sub.11 .multidot.SPD(t-j) of the register R1 to the latch circuit 404.

(9) Then, for the purpose of calculating an equation

the control signal output register 303 outputs a signal SL1=SELECT (B) and a signal CTL=(Y)=(A).multidot.(X)+(B) which constitute the operation code OPC and an address information ADR(Kn)=ADR(K.sub.12), whereby a coefficient K.sub.11 is read out from the coefficient memory device 400 and then applied to the input of the calculation circuit 402. The selector 401 selects the amplitude data SPD(t-j-1) latched in the latch circuit 101 and supplies it to the input x of the calculation circuit 402. Hence this circuit 402 outputs the result of calculation (Y) of the following equation ##EQU9## This calculated value Y is stored in the registers R1 and R2 in the next step so that their contents are changed as follows.

(10) For the purpose of calculating an equation K.sub.13 .multidot.SPD(t-j-1)+[R2] by utilizing the content of the register R2, the coefficient K.sub.12, and the amplitude data SPD(t-j-1), a (j-1) before and stored in the memory section SD0, the amplitude data SPD(t-j-1) is read out from the memory section SD0 and latched in the latch circuit 101 in the same manner as in steps (b)-(7).

(11) In the same manner as the step (b)-(8), the content K.sub.12 .multidot.SPD(t-j-1)+K.sub.11 .multidot.SPD(t-j) of the register R2 is transferred to the latch circuit 404.

(12) Then for the purpose of reading out the coefficient K.sub.13 to calculate an equation K.sub.13 .multidot.SPD(t-j-1)+[R2], the control signal output register 303 outputs a signal SL1=SELECT(B), and CTL=(Y)=(A).multidot.(X)+(B) which constitutes the operation code OPC and an address information ADR(Kn)=ADR(k.sub.13), whereby a coefficient K.sub.11 is read out from the coefficient memory device 400 and supplied to the input A of the calculation circuit 402. The selector 401 selects the amplitude data SPD(t-j-1) latched in the latch circuit 101 and supplies it to the input X of the calculation circuit 402, whereby it calculates an equation ##EQU10##

This calculated value (Y) is stored in the register R2 at the next step and then applied to the high pass filter HPF from this register R2.

(13) At the last step of the low pass filter LPF, for the purpose of writing the content of the register R1 in the memory section SD0 for use at the next sampling time (t+1), at first the content K.sub.12 .multidot.SPD(t-j-1)+K.sub.11 .multidot.SPD(t-j) of the register R1 is transferred to the latch circuit 404 in the same manner as the step (b)-(8) for causing the operation circuit 402 to calculate (Y)=(B). The calculated value (Y)=K.sub.12 .multidot.SPD(t-j-1)+K.sub.11 .multidot.SPD(t-j) is written in the memory section SD0. This writing operation is performed by outputting the operation code OPC constituted by DLk=DL.sub.SD, GP2="1" and WR4="1", and a memory number information DLn=DL.sub.o from the control signal output register 303.

After the operation of the low pass filter LPF completes, the high pass filter HPF operate in the same manner.

The formation of a reverberation tone RVD.sup.1 having a large delay time interval will now be described.

(c) Formation of the reverberation-tone RVD.sup.1

(1) At first, the data SPD(t-j) stored in the register R4 of the high pass filter HPF is multiplied with a coefficient K.sub.17 and for the purpose of storing the product K.sub.17 .multidot.SPD(t-j) in the register R5, a latch control signal L1="1" (LATCH), and a register number information RG.sub.n are outputted from the register 303 and the content SPD(t-j) of the register R4 is transferred to the latch circuit 404.

(2) Then for the purpose of calculating data K.sub.17 SPD(t-j) the control signal output register 303 outputs a selection control signal SL1=SELECT(c), a calculation control signal CTL=(Y)=(A).multidot.(X) and a coefficient read address information ADR(Kn)=ADR(K.sub.17), whereby a coefficient K.sub.17 is read out from the coefficient memory device 400 and applied to the input A of the calculation circuit. The selector 401 selects the data SPD(t-j) latched in the latch circuit 404 and applies it to the input X of the calculation circuit 402, whereby the calculation circuit calculates the following equation

and this calculated value Y is stored in the register R5 in the next step.

(3) Then, for the purpose of reading out the amplitude data SPD(t-x.sub.1), x.sub.1 interval before from the memory device D1, adding the read out data to the present value of the register R11 and storing again the sum in the register R11, the control signal output register 303 outputs a latch control signals L3="1" (LATCH) and L2="1" (LATCH) a memory number information DLn=DL.sub.1 and a memory type information DLk=DL.sub.D, whereby the output information ADR(D1) of the address counter AC(D1) corresponding to the memory section D1 would be latched in the latch circuit 306 as a lower order information for reading out the amplitude data SPD(t-x.sub.1). The memory number information DLn and the memory type information DLk are added to the upper order of the lower order address information ADR(D1) to form an address information DM ADR for the memory section D1 of the data memory device 100. Consequently, the amplitude data SPD(t-x.sub.1), x.sub.1 time before is read out from the memory section D1 and latched by the latch circuit 101.

(4) To add together the read out data SPD(t-x.sub.1) and the present value of the register R11, the content of the register R11 is transferred to the latch circuit 404 and thereafter the control signal output register 303 outputs a selection control signal SL1=SELECT(B) and a calculation control signal CTL=(Y)=(X)+(B).

Then the selector 401 selects the amplitude data SPD(t-x.sub.1) latched in the latch circuit 101 and supplies it to the input X of the calculation circuit 402, whereby the calculation circuit 402 calculates an equation

Before this time, the content of the register R11 has been cleared at a time when the operation at the preceding sampling time (t-1) was completed, so that the calculated value Y at this step (4) is equal to SPD(t-x.sub.1). Thereafter, the calculated value Y is transferred to the register R11 to be stored therein.

(5) Then, for the purpose of reading out the amplitude data SPD(t-x.sub.1) from the memory device D1, multiplying the amplitude data with a coefficient K.sub.18, and for storing again the sum of the product K.sub.18 SPD(t-x.sub.1) and the content K.sub.17 .multidot.SPD(t-j) of the register R5 in the register R6, the content K.sub.17 .multidot.SPD(t-j) of the register R5 is transferred to the latch circuit 404 in the same manner as in the step (c)-(1).

(6) Then, for the purpose of calculating equation

based upon the amplitude data SPD(t-x.sub.1) latched by the latch circuit 101, the data latched by the latch circuit 404 and the coefficient K.sub.18, the control signal output register 303 outputs a selection control signal SL1=SELECT(B), a calculation control signal CTL=(Y)=(A).multidot.(X)+(B), and a coefficient read out information ADR(Kn)=ADR(K.sub.18), whereby the coefficient K.sub.18 is read out from the coefficient memory device 400 and applied to the input A of the calculation circuit 402. On the other hand, the selector 401 selects the amplitude data SPD(t-x.sub.1) latched by the latch circuit 101 and applies it to the input X of the calculation circuit 402, whereby this calculation circuit calculates an equation ##EQU11## This calculated value Y is written into an address of the memory section D1 corresponding to the present time t via the register R6 in the next step. Thereafter, the register R6 is cleared for processing a system including the memory section D2.

(7) Then the processings regarding the systems respectively including memory sections are executed in the same manner as steps (c)-(3) through (c)-(6). Upon completion of the processings of respective systems respectively including memory sections D1 through D9, the register R11 produces an information regarding a reverberation tone RVD.sup.1 represented by ##EQU12##

The operation of forming a reverberation tone RVD.sup.2 having a short delay time interval will now be described.

(d) Formation of the reverberation tone RVD.sup.2

(1) For the purpose of reading out the amplitude data RVD.sup.1 (t-y.sub.1) from the memory section MD0, the control signal output register 303 outputs a latch control signal L3="1" (LATCH) and L2="1" (LATCH), a memory number information DLk=DL.sub.MD and a memory type information DLk=DL.sub.MD. Consequently, in the same manner as the step (c)-(3), an address information DM.multidot.ADR for the memory section MD0 is formed in the addess information output circuit 309 for reading out the amplitude data RVD.sup.1 (t-y.sub.1), a y.sub.1 time before, from the memory section MD0. This read out data RVD.sup.1 (t-y.sub.1) is latched by the latch circuit 101.

(2) The, data

is calculated by using the amplitude data RVD.sup.1 (t-y.sub.1) latched by the latch circuit 101, the output data RVD.sup.1 (t) of the register R11 and the coefficient K.sub.30. then to store the calculated value Y in the register R12, after transferring the output data RVD.sup.1 from the register R11 to the latch circuit 401, the control signal output register 303 outputs a selection control signal SL1=SELECT(B), a calculation control signal SL1=SELECT(B), a calculation control signal CTL=(Y)=(A).multidot.(X)+(B), and a coefficient read address information ADR(Kn)=ADR(K.sub.30). As a consequence, the coefficient K.sub.30 is applied to the input A of the calculation circuit 402 in the same manner as the step (c)-(6) described above, and the data RVD.sup.1 (t-y.sub.1) is supplied to the input X of the calculation circuit 402, with the result that the calculation circuit calculates an equation ##EQU13## and this calculated value (Y) is stored in the register R12 at the next step.

(3) Then, for the purpose of multiplying the content [K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)+RVD.sup.1 (t)] of the register R12 with a coefficient K.sub.29, the content of the register R12 is transferred to the latch circuit 404 and the control signal output register 303 outputs a selection control signal SL1=SELECT(C), a calculation control signal CTL=Y=(A).multidot.(X) and a coefficient read address information ADR(Kn)=ADR(K.sub.29).

Accordingly, the coefficient K.sub.30 is applied to the input A of the calculation circuit 402, while the data [K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)+RVD.sup.1 (t)] is being supplied to the input X of the calculation circuit 402, with the result that the calculation circuit 402 calculates an equation ##EQU14## This calculated value Y is stored in the register R13 at the next step.

(4) Then, for the purpose of adding the content of the register R13 to the data RVD.sup.1 (t-y.sub.1), y.sub.1 time before, and storing again the sum in the register R13, in the same manner as the step (d)-(1), the data RVD.sup.1 (t-y.sub.1), y.sub.1 time before is read out from the memory section MD0 and latched by the latch circuit 101. After transferring the content K.sub.29 .multidot.[K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)+RVD.sup.1 (t)] of the register R13 to the latch circuit 404, the control signal output register 303 produces a selection control signal SL1=SELECT(B) and a calculation control signal CTL, whereby the calculating circuit 402 produces a calculated value Y shown by ##EQU15## This calculated value Y is stored in the resistor R13 and then outputted as a reverberation tone information RVD.sup.2A at the next step.

(5) Then the content of the register R12 is written into an address of the memory section MD0 corresponding to the present time t for the purpose of utilizing the content [K.sub.30 .multidot.RVD.sup.1 (t-y.sub.1)+RVD.sup.1 (t)] of the register R12 at a sampling time (t+y.sub.1) later by a time y.sub.1.

(6) Thereafter a reverberation tones RVD.sup.2B and RVD.sup.2C having shorter intervals than (t) are formed in the same manner.

Although in the embodiment shown in FIGS. 5 and 6, a bandpass filter 16 provided, it may be omitted in a certain case. Furthermore, as shown in the performance block diagram shown in FIG. 17, the data may be divided into 3 frequency bands by using a high pass filter HPF, a bandpass filter BPF and a low pass filter LPF for forming different reverberation tones for respective frequency bands in the first reverberation tone forming unit 2. This can readily be realized by changing the content of a control program.

It should be understood that the invention is not limitted to the specific embodiment described above and that various changes and modifications may be made without departing from the true spirit and scope of the invention as defined in the appended claims.

For example, in the foregoing embodiment although the delay circuit was constituted by a digital memory device, any type of delay circuit can be used, for example such an analog delay element as an BBD or a CCD.

As above described since the reverberation tone generating apparatus is constituted by a combination of a delay circuit forming a reverbertion tone having irregular delay time and level and a delay circuit forming a reverberation tone having regular delay time and level it is possible to produce such reverberation tones rich in naturality as those produced in a concert hall with apparatus of small scale. Where a digital memory device is used as the delay circuit, high quality reverberation tones having excellent S/N ratio can be produced even when the reverberation time is elongated. In this case since the reverberation time can be varied to any length by varying the address spacing of the digital memory device, it is possible to simulate various types of reverberation tones which vary depending upon room conditions or ambient conditions.