JPH0536797B2 - - Google Patents

Description

[Detailed description of the invention]

(Industrial Application Field) The present invention relates to a musical tone synthesis device that can obtain high-quality musical tones for playing an electronic musical instrument, particularly for legato playing. (Configuration of conventional examples and their problems) In general, various synthesis methods have been attempted as digital musical tone synthesis methods, and these techniques are
For example, JP-A-52-121313 or JP-A-Sho.
It is disclosed in Publication No. 55-28072. According to these methods, musical waveforms are stored in memory as digital waveform data in advance, and during synthesis, the waveform data stored in memory is sequentially read out at a speed (clock pulse) proportional to the pitch of the musical sound to be output. ), the desired musical tone is synthesized with high quality. However, storing all the waveform data from the start to the end of a musical tone in memory requires an enormous amount of memory, and is virtually impossible for a sustained tone such as a trumpet. Therefore,
Some kind of waveform compression technique is required, but for example, all the rising parts of musical tones are stored as so-called PCM waveform data, and the stationary parts are stored as one-cycle waveforms. At the time of synthesis, a musical tone is synthesized by sequentially reading out the waveform data of the rising portion of the musical tone from the memory, and then repeatedly reading and outputting the one-cycle waveform data of the steady portion. However, according to such a method, a newly specified musical tone is synthesized and output without any correlation with the previously played note. Among the waveform data described above, waveform data consisting of a one-period waveform will be referred to as phoneme piece data, and waveform data consisting of a multi-period waveform spanning the entire rising portion will be referred to as phoneme group data. When playing legato, click noise occurs when the output cycle is discontinuous during DA conversion in response to pitch changes, so the following method can be considered as a solution to this problem. Between two consecutive tones, the output period of the two tones is changed stepwise by linear interpolation or the like. The output is temporarily attenuated between two consecutive notes. In this case, the method (2) still has the disadvantage that click noise occurs when the output period ratio to be legato processed is large, and the method (2) can reduce the click noise, but the following This has the disadvantage that the rise of the sound is fixed, and it requires the addition of a gain control section. (Object of the Invention) It is an object of the present invention to provide a musical tone synthesis device which eliminates the above-mentioned conventional drawbacks and obtains high-quality musical tones corresponding to legato during legato performance. (Structure of the Invention) In order to achieve the above object, the musical tone synthesis device of the present invention includes an input unit for instructing the pitch of a musical tone to be output, start of generation, and stop of generation, and an input unit for instructing the pitch of a musical tone to be output, start of generation, and stop of generation, and an input unit for instructing the pitch of a musical tone to be output, start of generation, and stop of generation of sound, and an input unit for instructing the pitch of a musical tone to be output, start of generation, and stop of generation of sound, and an input unit for instructing the pitch of a musical tone to be output, start of generation, and stop of generation, A phoneme memory section that stores phoneme piece data or phoneme group data, and a plurality of tables that are composed of a plurality of table data for indicating phoneme piece data or phoneme group data to be used to synthesize musical tones. A table is selected depending on whether or not the table memory section, the pitch instruction to the input section, and the instruction to start producing two musical tones to the input section are temporally consecutive, and the selected table is configured. It is composed of a control section that synthesizes musical tones using phoneme piece data or phoneme group data corresponding to table data. (Description of Examples) Hereinafter, examples will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, in which 1 is an input section, 2 is a control section, 3 is a table memory section, 4 is a phoneme memory section, 5 is an output buffer section, and 6 is a DA section. This is the conversion section. First, when a desired type of musical tone is specified to an input section 1 such as a musical instrument keyboard, the input section 1 outputs a number N indicating the specified type of musical tone to the control section 2. Next, when the pitch of a desired musical tone is specified, the input section 1 outputs a number H indicating the specified pitch to the control section 2. Furthermore, when the start of sound generation of a desired musical tone is instructed, the input section 1 outputs a sound generation start signal to the control section 2, thereby starting musical tone synthesis. Musical tone synthesis is performed until an instruction is given to stop the production of a desired musical tone. The control section 2 performs a musical tone synthesis operation based on the above information, that is, the musical tone type number N, the musical tone pitch number H, the sound generation start signal, and the sound generation stop signal. Next, the operation of the control section 2 will be explained based on the table data shown in the table.

[Table] The control unit 2 selects the type N and pitch H of musical tones to be output.
The starting address A _H of the table M to be read from the table memory section 3 determined by is determined as follows. A _H =2・N・H・W (1) However, W represents the size of one table secured for synthesizing one musical tone. Also, N=
0,1,2,...,N _nax and H=0,1,2,...,
H _nax . In other words, N _nax is all kinds of musical tones,
H _nax corresponds to the number of keys, and usually H _nax =
87. Therefore, assuming that 256 words of memory are secured for each table, when there are four types of musical tones, 180,224 words (=2×4×88×256) of table memory are required. Note that the reason for first multiplying by 2 to calculate the address A _H is to enable separate tables to be used for legato and non-legato for each pitch. It is assumed that all data consists of one word. When a sound generation start signal is input after the musical tone type number N and musical tone pitch number H are input from the input section 1, the control section 2 starts the following operations. The control unit 2 reads the output time period T from the calculated head address A _H and outputs it to the output buffer unit 5. Next, the control unit 2 increases the address A _H in the table memory unit 3 by one word and adds the synthesis information S ₁ , that is, the address A ₁ in the phoneme memory unit 4 and the phoneme piece data or phoneme group data length.
Read L ₁ and repetition number R ₁ . Based on the read synthesis information _S1 , the control unit 2 sequentially reads phoneme pieces or pitch groups starting from the address _A1 in the phoneme memory unit 4 one word at a time, and outputs them as they are to the output buffer unit 5 while reading only _one word of data length L. . That is, one word of the data of the phoneme piece or phoneme group at address A ₁ in the phoneme memory section 4 is read out and output to the output buffer section 5, and then one word of the data at the address (A ₁ +1) in the phoneme memory section 4 is read out. The operation of reading out is repeated L _times to finish outputting one phoneme piece or one group of phonemes. The above output of the phoneme piece or phoneme group corresponding to the synthesis information S ₁ is repeated _once for the number of repetitions R, and the synthesis operation for the synthesis information S ₁ is completed. Next, the control unit 2 increases the address A _H in the table memory unit 3 by one word and adds the synthesis information S ₂ , that is, the address A ₂ in the phoneme memory unit 4 and the phoneme piece data or phoneme group data length L ₂ . Number of repetitions
Load R ₂ . With the loaded composite information S ₂ ,
The control section 2 sequentially reads out a phoneme piece or a phoneme group starting from address _A2 in the phoneme memory section 4 one word at a time by a data length of _L2 words, and outputs it to the output buffer section 5 as it is. That is, one word of the data of the phoneme piece or phoneme group at address A ₂ in the phoneme memory section 4 is read out and output to the output buffer section 5, and then one word of the data of the address (A ₂ +1) in the phoneme memory section 4 is read out. Repeat the reading operation _twice to finish outputting one phoneme piece or one group of phonemes. The number of repetitions R ₂ is to output the phoneme piece or phoneme group corresponding to the above synthesis information S ₂
The synthesis operation for the synthesis information _S2 is completed by repeating this process for the same number of times. After repeating the above operations in the same way, the number of repetitions Rn of the synthesis information Sn is a value indicating an infinite number of repetitions (it is desirable to apply the value = 0), so this phoneme piece or group of phonemes continues to be output repeatedly. It turns out. During the musical tone synthesis operation described above, when a signal related to stopping the output of the output musical tone is inputted from the input section 1, the control section 2 stops the musical tone synthesis operation. In the above, the phoneme piece data or phoneme group data stored in the phoneme memory section 4 is a PCM in which one word is made up of 8 or 16 bits sequentially from the address Ai corresponding to the table data.
(Pulse Code Modulation), data length Li
is stored sequentially, and consists of a data string of at least one cycle. When the next musical tone to be synthesized and output is further instructed while the musical tone instructed by the input section 1 is being synthesized and output, legato input is made. That is, when the sound generation start signal for the second musical tone is output from the input section 1 while the first musical tone is being generated, that is, before the sound generation stop signal for the first musical tone is output from the input section 1, the control section 2 is the type N and pitch H of the musical tone to be output, and the table size W mentioned above, in order to generate a musical tone corresponding to the legato playing style for the second musical tone after stopping the generation of the first musical tone. From,
Table in table memory section 3 to be read
The starting address A _H ′ of M′ is determined as follows. A _H ′=2・N・H・W+W …(2) Table memory section 3 determined as above
According to the first address _AH ' in the table M, the control section 2 performs the tone synthesis operation in the same manner as for the first address _AH of the table M described above. In the above, when the output time period T is output from the control section 2, the output buffer section 5 sequentially processes the phoneme piece data or phoneme group data output from the control section 2 in accordance with the output time period T. Outputs to the DA converter 6 at each timing, and also outputs the DA
The conversion unit 6 converts the input phoneme piece data or phoneme group data into an analog signal and outputs it.
A synthesized musical tone output will be obtained. Next, the characteristic content of the present invention in this embodiment will be explained with reference to FIGS. 2 and 3. FIG. 2 shows that when the musical tone generation start instruction to the input section 1 is intermittently executed twice, the control section 2 receives the first musical tone generation start instruction from the table memory section 3 according to equation (1). Select the table _M1 starting from the obtained head address A _H1 , and read the phoneme piece or phoneme group data from the phoneme memory section 4 by the same operation as described above according to the synthesis information stored in this table _M1 . while synthesizing synthesized musical tone _P1 . When the first musical tone generation stop instruction is given to the input section 1, the control section 2 stops the musical tone synthesis operation and outputs a zero value. When the input section 1 is instructed to start generating the second musical tone, the control section 2 performs the same operation as for the first musical tone, and starts from the start address A _H2 obtained from the table memory section 3 by equation (1). Select table _M2 starting from this table _M2
The synthesized musical tone P2 is synthesized while reading out phoneme pieces or phoneme group data from the phoneme memory unit 4 according to the synthesis information stored in the synthesized musical tone _P2 . In the figure, a is the rising part,
b indicates a steady portion after the rising portion. FIG. 3 shows that when the instruction to start musical tone generation to the input section 1 is executed twice consecutively, that is, during legato performance, the control section 2 receives the first musical tone generation start instruction from the table memory section 3. Correspondingly, the table starting from the first address A _H1 obtained by equation (1)
M ₁ is selected, and the above operation is performed in a similar manner according to the composition information stored in this table M ₁ .
While reading the phoneme piece or phoneme group data from the phoneme memory section 4, a synthesized musical tone _P1 is synthesized. When the input unit 1 is instructed to start generating the second musical tone, the control unit 2 stops the first musical tone synthesis operation,
By performing the same operation for the first musical tone, the start address obtained from the table memory section 3 by equation (2)
A table M′ ₂ starting from A _H ′ ₂ is selected, and the synthesized musical sound P′ ₂ is synthesized while reading phoneme pieces or phoneme group data from the phoneme memory section 4 according to the synthesis information stored in this table M′ ₂ . do. In the figure, a' indicates a rising portion, and b indicates a steady portion after the rising portion. The characteristic difference between FIG. 2 and FIG. 3 in this example is that the rise times and composite waveform shapes of the rising portions a and a' of the second musical tone are different in non-legato and legato playing styles. There it is. In this embodiment, the constant portion b after the rising portion of the second musical tone in the non-legato performance style and the legato performance style share the same phoneme piece or phoneme group data. (Effects of the Invention) As described above, according to the present invention, there is provided an input unit for instructing the pitch of a musical sound to be output, the start and stop of sound generation, and the phoneme piece data or phoneme group necessary for synthesizing the musical sound. A phoneme memory unit that stores data;
A table memory section that stores multiple tables composed of multiple table data for indicating phoneme piece data or phoneme group data to be used to synthesize musical tones, and a table corresponding to pitch instructions of the input section. and a control unit for synthesizing a musical tone using phoneme piece data or phoneme group data corresponding to the table data constituting the table, and the control unit receives two musical tone generation start instructions to the input unit sequentially in time. The table selected for the second musical tone is different in both cases of legato and non-legato playing, that is, for legato and non-legato playing. In the case of different playing styles, the phoneme piece data or phoneme group data used to synthesize the second musical tone will be different. Therefore, in the case of legato playing, it is necessary to use a table in which the synthesized waveform rises rapidly to synthesize phoneme piece data or phoneme group data.
By preparing a table in advance to form a composite waveform corresponding to the legato playing style, it is possible to eliminate click noise without adding a gain control section to the rise of the second musical waveform and controlling the gain as in the past. A musical tone synthesis device that reduces noise can be realized. In addition, since the table data and phoneme piece data or phoneme group data used for synthesizing the second musical tone in the case of legato playing style can be set arbitrarily, an appropriate legato tone can be created depending on the type and pitch of the musical tone. Can be synthesized.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention, and FIGS. 2 and 3 are output waveform diagrams for explaining the operation of the present invention. 1... Input section, 2... Control section, 3... Table memory section, 4... Phoneme memory section, 5... Output buffer section, 6... DA conversion section.

Claims (1)

[Scope of Claims] 1. An input section for instructing the pitch of a musical tone to be output and the start and stop of sound generation; a phoneme memory section for storing phoneme piece data or phoneme group data necessary for synthesizing the musical tones; , a table memory section that stores a plurality of tables constituted by a plurality of table data for indicating phoneme piece data or phoneme group data to be used to synthesize the musical tones; and a pitch instruction of the input section; a control unit that selects the corresponding table and synthesizes a musical tone using the phoneme piece data or phoneme group data corresponding to the table data constituting the table; A musical tone synthesis device characterized in that the table to be selected for the second musical tone is different depending on whether the musical tone generation start instruction is continuous in time or intermittently.