JP4554629B2 - Waveform generator, synthesizer for sound source - Google Patents

Description

  The present invention relates to a waveform generation apparatus that reads waveform data stored in a waveform memory and generates a predetermined waveform, and a sound source synthesizer using the waveform generation apparatus.

  Generally, a sound source synthesizer has a waveform memory that stores waveform data obtained by sampling the sound of a musical instrument, and is configured to perform predetermined processing on the waveform data read from the waveform memory and output a musical sound. Has been. Waveform data is expressed as a digital value, and a typical modulation method for digitally expressing a wave is a PCM (Pulse Code Modulation) method. In the PCM method, the wave height at each time is expressed by a numerical value. Waves represented by such digital values can be converted in wavelength by a mathematical method. Wavelength conversion is a process required when raising or lowering the pitch of a musical sound.

  FIG. 10 is a diagram showing a waveform modulated by the PCM method, where W1 is a PCM waveform stored in the waveform memory, and W2 is a PCM waveform when W1 is wavelength-converted. The wavelength of W2 is converted to λ × n (n times) with respect to the wavelength λ of W1. Here, for convenience, the PCM waveforms W1 and W2 are drawn as continuous waves, but the actual PCM waveforms W1 and W2 are not continuous waves, but are discrete signals sampled at predetermined time intervals.

  As a typical method for performing such wavelength conversion, there is a method of reading a sample value by setting the sampling time interval of the original waveform to 1 / n. In this case, depending on the sampling timing, the sample value may not exist in the waveform memory, so the total number of samples decreases and the waveform error increases. For this reason, it is necessary to generate a virtual sample value by interpolation processing. Japanese Patent Application Laid-Open No. 2004-133620 describes a musical tone generating apparatus having such an interpolation processing function.

  The simplest interpolation method is two-point linear interpolation that geometrically finds an interpolation point between two points from the height and time interval of two waves on the waveform. However, this method is effective in the case of a simple waveform, but there is a drawback that the accuracy of interpolation is reduced when the waveform becomes complicated. Therefore, in the case of a complicated waveform, it is necessary to perform a slightly more advanced interpolation process. One method for this purpose is to apply Shannon's sampling theorem and perform interpolation by performing a convolution operation using an impulse response of a low-pass filter. However, there is a problem that a complicated circuit is required to perform the convolution operation, and the circuit becomes large and costs increase.

  On the other hand, as another method of wavelength conversion, there is a pitch conversion method using general harmonic analysis as shown in Non-Patent Document 1 described later. General harmonic analysis is an analysis method that extracts the sine wave with the smallest residual component from the original waveform within the observation window, and repeats the same processing for the residual component, enabling accurate frequency component extraction. It has the advantage of being. However, this method has a problem that the amount of calculation increases and the processing takes time.

JP 2003-233378 A "Time-axis companding and pitch conversion of high-speed 1-bit acoustic signals using general harmonic analysis" (Mayashi Hayashi, Yoshio Yamazaki, Akihiro Oikawa, Proc. Of the Acoustical Society of Japan, 603-604, September 1996)

  As described above, the conventional wavelength conversion method has a problem that the circuit becomes complicated and the processing takes time. Accordingly, an object of the present invention is to provide a waveform generation apparatus capable of performing wavelength conversion at high speed by simple means.

  A waveform generation device according to the present invention includes a waveform memory in which waveform data is stored, and a waveform generation unit having a pitch conversion unit for converting the pitch of a waveform generated based on the waveform data to N / M times The pitch conversion unit includes an interpolation unit, a total value calculation unit, and a quantization unit. The interpolation means interpolates N−1 zero values between each 1-bit signal constituting a PDM (Pulse Density Modulation) signal obtained from the waveform data. The total value calculation means sequentially reads out the data string after being interpolated by the interpolation means as a unit including a block including M bits, and for each block, the bit value at the bit position determined by the values of N and M A total value X obtained by summing the multiplication values with the value of N is calculated. The quantization means performs quantization using the value of M as a threshold for the total value Y obtained by adding the total value X in the block and the remainder value obtained by quantization in the previous block. , Assign a value of 1 or 0 to each block and output the value, and add the remainder value obtained by the quantization to the total value X of the next block.

  In the present invention, the pitch of the 1-bit signal constituting the PDM signal is shifted by interpolation processing, the interpolated data string is read out in units of blocks, and quantization is performed for each block, thereby maintaining the 1-bit signal. Can change the pitch. For this reason, complicated arithmetic processing such as convolution calculation and general harmonic analysis is not required, and pitch conversion processing can be performed at high speed by simple means. Further, by arbitrarily setting the values of N and M, it is possible to easily obtain a waveform having a desired pitch.

  In a preferred embodiment of the present invention, a ΔΣ modulator is used as the quantization means. According to this, even if a dedicated arithmetic circuit is not designed, it is possible to easily realize the quantization and the addition process of the remainder value by using the function of the existing ΔΣ modulator.

  A sound source synthesizer according to the present invention is processed by the above-described waveform generation device, a signal processing unit that performs predetermined signal processing on the sound waveform generated by the waveform generation device, and the signal processing unit. And an output unit that outputs sound based on the signal. The sound to be handled is not limited to a musical sound, and can be any sound such as a human voice or a pseudo sound.

  According to the present invention, there is an effect that wavelength conversion can be performed on a waveform of a PDM signal at high speed without requiring complicated processing.

  FIG. 1 is a block diagram showing the overall configuration of a sound source synthesizer according to an embodiment of the present invention. Reference numeral 1 denotes a waveform memory that stores musical tone waveform data. For example, waveform data of a PCM signal as shown in FIG. Here too, the PCM waveform is drawn as a continuous wave for convenience, but the actual PCM waveform is a sampled discrete signal. The waveform data stored in the waveform memory 1 is not limited to the PCM signal but may be a PDM signal described later.

  A PCM / PDM conversion unit 2 performs signal conversion, and converts the PCM signal of FIG. 3A read from the waveform memory 1 into a binary PDM signal as shown in FIG. This conversion is performed by a known delta-sigma (ΔΣ) modulation. When the waveform data stored in the waveform memory 1 is a PDM signal, the PCM / PDM conversion unit 2 is not necessary.

  Reference numeral 3 denotes a pitch converter which is a main part of the present invention, and performs pitch conversion on a 1-bit signal constituting a PDM signal output from the PCM / PDM converter 2. FIG. 2 is a diagram illustrating the configuration of the pitch conversion unit 3. Actually, each function of the pitch conversion unit 3 is realized by software, but here, it is represented by hardware for convenience. A weighting unit 3a sets the value of N specified by the user when the pitch of the PDM signal is converted to N / M times. Reference numeral 3b denotes a zero value interpolation unit which interpolates N−1 0 values as 1-bit signals between the 1-bit signals of the PDM signal. 3c is a total value calculation unit, which sequentially reads out the data string after the zero value interpolation as a unit including a block including M bits, and the bit value of the bit position determined by the values of N and M for each block And the sum of the product of N and the value of N is calculated. Reference numeral 3d denotes a quantization unit which assigns a value of 1 or 0 to each block and outputs the result by quantization processing described later.

  The waveform generation device 100 is configured by the waveform memory 1, the PCM / PDM conversion unit 2, and the pitch conversion unit 3 described above. Further, the zero value interpolation unit 3b, the total value calculation unit 3c, and the quantization unit 3d in the pitch conversion unit 3 constitute an embodiment of the interpolation unit, the total value calculation unit, and the quantization unit in the present invention, respectively.

  Reference numeral 4 denotes a signal processing unit that performs predetermined signal processing on the waveform generated by the waveform generation apparatus 100. The signal processor 4 converts the PDM signal pitch-converted by the pitch converter 3 into a PCM signal by passing through a low-pass filter. The signal processing unit 4 also includes a time-varying filter unit that controls the timbre by changing the cutoff frequency with respect to the PCM signal, and a gain that changes with time with respect to the output of the time-varying filter unit. A time-varying amplification unit that performs attenuation processing, a mixer that performs mixing processing for synthesizing a plurality of sounds, and the like are provided.

  Reference numeral 5 denotes an output unit that outputs sound based on the signal processed by the signal processing unit 4. For example, when the PCM signal output from the signal processing unit 4 is a musical sound signal, the musical sound is output from a speaker (not shown). To do.

  The quantization unit 3d in the pitch conversion unit 3 can be configured by a known ΔΣ modulator that performs delta-sigma (ΔΣ) conversion. FIG. 4 is a block diagram illustrating an example of the ΔΣ modulator 50. Reference numeral 51 denotes an arithmetic unit that calculates the difference between the input value and the output feedback value, and reference numeral 52 denotes an arithmetic unit that adds the output of the arithmetic unit 51 and the output of the delay circuit 53. The delay circuit 53 holds the input signal for one sampling period, and outputs the held signal in the next sampling. The arithmetic unit 52 and the delay circuit 53 constitute an integrator 56. Reference numeral 54 denotes a quantization circuit that performs quantization on the output of the integrator 56 using M as a threshold. If the input value is M or more, “1” is output, and if the input value is less than M, “0” is output. Is output. A multiplier 55 has a feedback gain set to τ.

  FIG. 5 is a diagram showing the principle of pitch conversion according to the present invention. First, the principle of the present invention will be described with reference to FIG.

  FIG. 5A shows the waveform (wavelength λ1) of the original PCM signal stored in the waveform memory 1 and the waveform of the corresponding PDM signal. (B) shows the waveform of the PCM signal (wavelength λ2) after the pitch conversion and the corresponding PDM signal. In the figure, the waveform of the PDM signal is simplified. (C) is an enlarged view of a portion surrounded by a broken line of the PCM signal of (a), and Q1 to Q6 represent sample values at each sampling time point. A dotted line is a virtual sample value at each time point when the sampling interval is divided into M (for example, M = 5), and is not actual data. (D) has shown the bit value of the PDM signal corresponding to the PCM signal of (c), and the solid line represents 1 and the broken line represents 0. For convenience, each bit value is shown at the same timing as the sample values Q1 to Q6. The numbers shown by shading are the bit values in each sampling section, and coincide with the bit values shown by the solid line and the broken line.

  To obtain the waveform of FIG. 5B from the waveform of FIG. 5A, first, a zero value is interpolated between each 1-bit signal constituting the PDM signal. If the waveform pitch is converted to N / M times (λ2 = λ1 × N / M), N−1 “0” s are interpolated between 1-bit signals of the original PDM signal. For example, if N = 9, as shown in (f), eight “0” s are inserted between 1-bit signals. As a result, the bit values of the solid line and the broken line in (d) shift as shown in (f). However, the interval between each bit is constant (N = 9).

  Next, in FIG. 5F, the data string in each sampling interval (M data including the interpolation zero value) is taken as one block, and the data string is sequentially read out in units of blocks. Then, the calculation described later is performed using the bit values (1 or 0) of the solid line and the broken line included in each block and the value of N, and 1 or 0 is assigned to each block based on the calculation result. The numbers indicated by shading represent the assigned bit values. When extending the pitch, the wave height of the PCM signal at the same sampling time is reduced, and accordingly, the bit value of the PDM signal also changes from the bit value of the original PDM signal. The PCM signal corresponding to the PDM signal composed of the bit values after the change is as shown in FIG. 5E, which corresponds to the portion surrounded by the broken line of the PCM signal shown in FIG. Therefore, the PCM signal of (b) whose wavelength is converted to N / M times can be obtained by passing the PDM signal composed of the shaded data sequence obtained in (f) through the low-pass filter and returning it to the PCM signal. it can.

  FIG. 6 is a diagram showing in detail the state of zero value interpolation. Here, an example is given in which M = 8 and N = 9, and the pitch is extended to N / M = 9/8 times. (A) is a data string of bit signals constituting the original PDM signal, (b) is a data string obtained by interpolating N−2 = 7 “0” s between the bit signals of the data string of (a), (C) shows a data string obtained by interpolating N−1 = 8 “0” s between the bit signals of the data string of (a). (B) is given as a reference example, and this data string is not used in the present invention. Further, “0” interpolated in the data string of (c) is inserted only to shift the bit signal (shaded portion) of the PDM signal by a predetermined bit, and is not used in the arithmetic processing described later. As described in FIG. 5, the interpolated data sequence (c) is read in block units, with one block including M (= 8) bits corresponding to the sampling period.

  Next, details of the pitch conversion method of the present invention based on the above-described principle will be described with reference to specific examples. FIG. 7 is a flowchart showing a pitch conversion procedure in the pitch converter 3 of FIG. FIG. 8 is a diagram showing an example of data calculation when the pitch is extended, and FIG. 9 is a diagram showing an example of data calculation when the pitch is reduced.

  First, the pitch conversion procedure will be described with reference to FIG. 8A, (a) represents the first block, (b) represents the second block, (c) represents the third block, and (d) represents the fourth block.

  In step S1 of FIG. 7, the variables are initialized. As variables, an input value in input to the pitch conversion unit 3, an output value out output from the pitch conversion unit 3, a cursor value pos specifying a bit position in each block in FIG. 8, and a calculation for each block There is a sum value sum to be described later. By initialization, all these variables become zero.

  In step S2, it is determined whether or not the cursor value pos is greater than or equal to M. FIG. 8A shows a case where the pitch is 9/8 times, and the values of N and M are set to N = 9 and M = 8, respectively. Since pos = 0 initially, pos <M (step S2: NO), the process proceeds to step S9, and the data of the first block (a) is read into the input value in. In step S10, N (= 9) is added to the cursor value pos. Thereby, pos in the first block (a) becomes 0 + 9 = 9. Next, in step S11, the bit value “1” at the head (pos = 0) of the read first block (a) is multiplied by N (= 9), and this is added to the total value sum (= 0). The total value sum is updated. Thereby, sum in the first block (a) becomes 0 + 1 × 9 = 9. Thereafter, in step S7, the presence or absence of subsequent data is checked. If there is data (step S7: YES), the process returns to step S2, and if there is no data (step S7: NO), the process is terminated.

  In this case, since there is subsequent data, the process returns to step S2 to determine again whether pos is equal to or greater than M (= 8). As described above, pos = 9 at this time, so pos ≧ M (step S2: YES), and the process proceeds to step S3. In step S3, M (= 8) is subtracted from pos (= 9) to update the value of pos. As a result, pos = 1. Then, it progresses to step S4 and it is determined whether the total value sum is more than M (= 8). As described above, sum = 9 at this time, so sum ≧ M (step S4: YES), and the process proceeds to step S5. In step S5, “1” is written to the output value out, and M (= 8) is subtracted from sum (= 9) to update the total value. As a result, sum = 1. This value corresponds to the remainder value in the present invention. The remainder value is carried over to the next block and added. Further, the processes in steps S4, S5, and S8 correspond to quantization in the present invention. In the subsequent step S6, the output value out (= 1) is output. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). Since pos = 1 at this point, pos <M (step S2: NO), and the process proceeds to step S9. Then, the data of the second block (b) is read into the input value in. Subsequently, in step S10, N (= 9) is added to the cursor value pos (= 1). Thereby, pos in the second block (b) becomes 1 + 9 = 10. Next, in step S11, the bit value “0” of the second bit (pos = 1) from the top of the read second block (b) is multiplied by N (= 9), and this is summed (sum = 1). ) To update the total value sum. Thereby, sum in the second block (b) becomes 1 + 0 × 9 = 1. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). At this time, pos = 10, so pos ≧ M (step S2: YES), and the process proceeds to step S3. . In step S3, M (= 8) is subtracted from pos (= 10) to update the value of pos. As a result, pos = 2. Then, it progresses to step S4 and it is determined whether the total value sum is more than M (= 8). Since sum = 1 at this time, sum <M (step S4: NO), and the process proceeds to step S8. In step S8, “0” is written in the output value out, and in step S6, the output value out (= 0) is output. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is greater than or equal to M (= 8). Since pos = 2 at this time, pos <M (step S2: NO), and the process proceeds to step S9. Then, the data of the third block (c) is read into the input value in. In step S10, N (= 9) is added to the cursor value pos (= 2). Thereby, pos in the third block (c) becomes 2 + 9 = 11. Next, in step S11, the bit value “1” of the third bit (pos = 2) from the top of the read third block (c) is multiplied by N (= 9), and this is summed (sum = 1). ) To update the total value sum. Thereby, the sum in the third block (c) becomes 1 + 1 × 9 = 10. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is greater than or equal to M (= 8). Since pos = 11 at this time, pos ≧ M (step S2: YES), and the process proceeds to step S3. . In step S3, M (= 8) is subtracted from pos (= 11) to update the value of pos. As a result, pos = 3. Then, it progresses to step S4 and it is determined whether the total value sum is more than M (= 8). Since sum = 10 at this time, it becomes sum ≧ M (step S4: YES), and the process proceeds to step S5. In step S5, “1” is written to the output value out, and M (= 8) is subtracted from sum (= 10) to update the total value. As a result, sum = 2. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). Since pos = 3 at this point, pos <M (step S2: NO), and the process proceeds to step S9. Then, the data of the fourth block (d) is read into the input value in. Subsequently, in step S10, N (= 9) is added to the cursor value pos (= 3). As a result, pos in the fourth block (d) is 3 + 9 = 12. Next, in step S11, the bit value “0” of the fourth bit (pos = 3) from the top of the read fourth block (d) is multiplied by N (= 9), and this is summed sum (= 2). ) To update the total value sum. As a result, the sum in the fourth block (d) is 2 + 0 × 9 = 2. Thereafter, in step S7, the presence or absence of subsequent data is checked. If there is data (step S7: YES), the process returns to step S2, and if there is no data (step S7: NO), the process is terminated.

  In the manner as described above, the same processing is performed for the subsequent blocks after the fifth block. As a result, a bit signal “1010...” Constituting the PDM signal is output from the pitch conversion unit 3. Then, by passing this PDM signal through a low-pass filter, a PCM signal having a pitch (wavelength) converted to 9/8 times is obtained.

  Next, an example of FIG. 8B will be described. In FIG. 8B as well, (a) to (d) represent the first block to the fourth block.

  Initially, the variables are initialized in step S1 of FIG. In the next step S2, it is determined whether or not the cursor value pos is greater than or equal to M. FIG. 8B shows a case where the pitch is 17/8 times, and the values of N and M are set to N = 17 and M = 8, respectively. Since pos = 0 initially, pos <M (step S2: NO), the process proceeds to step S9, and the data of the first block (a) is read into the input value in. In step S10, N (= 17) is added to the cursor value pos. Thereby, pos in the first block (a) becomes 0 + 17 = 17. Next, in step S11, the bit value “1” at the head (pos = 0) of the read first block (a) is multiplied by N (= 17), and this is added to the total value sum (= 0). The total value sum is updated. Thereby, the sum in the first block (a) becomes 0 + 1 × 17 = 17. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). Since pos = 17 at this time, pos ≧ M (step S2: YES), and the process proceeds to step S3. In step S3, M (= 8) is subtracted from pos (= 17) to update the value of pos. As a result, pos = 9. Then, it progresses to step S4 and it is determined whether the total value sum is more than M (= 8). Since sum = 17 at this time, sum ≧ M (step S4: YES), and the process proceeds to step S5. In step S5, “1” is written to the output value out, and M (= 8) is subtracted from sum (= 17) to update the total value. As a result, sum = 9. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). Since pos = 9 at this time, pos ≧ M (step S2: YES), and the process proceeds to step S3. To do. In step S3, M (= 8) is subtracted from pos (= 9) to update the value of pos. As a result, pos = 1. Then, it progresses to step S4 and it is determined whether the total value sum is more than M (= 8). Since sum = 9 at this time, sum ≧ M (step S4: YES), and the process proceeds to step S5. In step S5, “1” is written in the output value out (in the flowchart of FIG. 7, if pos ≧ M (= 8), then the output of out is continued until pos <M (= 8). ). Further, M (= 8) is subtracted from sum (= 9) to update the total value. As a result, sum = 1. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). Since pos = 1 at this point, pos <M (step S2: NO), and the process proceeds to step S9. Then, the data of the third block (c) is read into the input value in. Subsequently, N (= 17) is added to the cursor value pos (= 1) in step S10. Thereby, pos in the third block (c) becomes 1 + 17 = 18. Next, in step S11, the bit value “0” of the second bit (pos = 1) from the top of the read third block (c) is multiplied by N (= 17), and this is summed (sum = 1). ) To update the total value sum. As a result, the sum in the third block (c) is 1 + 0 × 17 = 1. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  The same processing is performed for the subsequent blocks in the manner described above. As a result, a bit signal “1100...” Constituting the PDM signal is output from the pitch converter 3. Then, by passing this PDM signal through a low-pass filter, a PCM signal whose pitch (wavelength) is converted to 17/8 times is obtained.

  Next, the pitch conversion procedure will be described with reference to FIG. 9A also, (a) to (d) represent the first block to the fourth block.

  Initially, the variables are initialized in step S1 of FIG. In the next step S2, it is determined whether or not the cursor value pos is greater than or equal to M. FIG. 9A shows a case where the pitch is 7/8 times, and the values of N and M are set to N = 7 and M = 8, respectively. Since pos = 0 initially, pos <M (step S2: NO), the process proceeds to step S9, and the data of the first block (a) is read into the input value in. In step S10, N (= 7) is added to the cursor value pos. As a result, pos becomes 0 + 7 = 7. Next, in step S11, the bit value “1” at the head (pos = 0) of the read first block (a) is multiplied by N (= 7), and this is added to the total value sum (= 0). The total value sum is updated. Thereby, sum becomes 0 + 1 × 7 = 7. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). Since pos = 7 at this time, pos <M (step S2: NO), the process proceeds to step S9 and the first block ( The data of a) is read into the input value in (in the flowchart of FIG. 7, the data reading is continued until pos ≧ M (= 8)). In step S10, N (= 7) is added to the cursor value pos. Thereby, pos in the first block (a) becomes 7 + 7 = 14. Next, in step S11, the bit value “0” of the eighth bit (pos = 7) from the top of the read first block (a) is multiplied by N (= 7), and this is summed (sum == 7). ) To update the total value sum. As a result, sum in the first block (a) becomes 7 + 0 = 7. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is greater than or equal to M (= 8). Since pos = 14 at this time, pos ≧ M (step S2: YES), and the process proceeds to step S3. To do. In step S3, M (= 8) is subtracted from pos (= 14) to update the value of pos. As a result, pos = 6. Then, it progresses to step S4 and it is determined whether the total value sum is more than M (= 8). Since sum = 7 at this point, sum <M (step S4: YES), and the process proceeds to step S8. In step S8, “0” is written in the output value out, and in step S6, the output value out (= 0) is output. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). Since pos = 6 at this time, pos <M (step S2: NO), and the process proceeds to step S9. Then, the data of the second block (b) is read into the input value in. Subsequently, in step S10, N (= 7) is added to the cursor value pos (= 6). Thereby, pos in the second block (b) becomes 6 + 7 = 13. Next, in step S11, the bit value “1” of the seventh bit (pos = 6) from the top of the read second block (b) is multiplied by N (= 7), and this is summed (sum == 7). ) To update the total value sum. As a result, the sum in the second block (b) becomes 7 + 1 × 7 = 14. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). At this time, pos = 13, so pos ≧ M (step S2: YES), and the process proceeds to step S3. . In step S3, M (= 8) is subtracted from pos (= 13) to update the value of pos. As a result, pos = 5. Then, it progresses to step S4 and it is determined whether the total value sum is more than M (= 8). At this time, sum = 14, so sum ≧ M (step S4: YES), and the process proceeds to step S5. In step S5, “1” is written to the output value out, and M (= 8) is subtracted from sum (= 14) to update the total value. As a result, sum = 6. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is equal to or greater than M (= 8). Since pos = 5 at this time, pos <M (step S2: NO), and the process proceeds to step S9. Then, the data of the third block (c) is read into the input value in. Subsequently, in step S10, N (= 7) is added to the cursor value pos (= 5). Thereby, pos in the third block (c) becomes 5 + 7 = 12. Next, in step S11, the bit value “0” of the sixth bit (pos = 5) from the top of the read third block (c) is multiplied by N (= 7), and this is summed (sum == 6). ) To update the total value sum. Thereby, the sum in the third block (c) is 6 + 0 × 7 = 6. Thereafter, the presence or absence of subsequent data is checked in step S7, and there is data (step S7: YES), the process returns to step S2.

  In step S2, it is determined again whether or not pos is greater than or equal to M (= 8). At this time, pos = 4, so pos <M (step S2: NO), and the process proceeds to step S9. Then, the data of the fourth block (d) is read into the input value in. Subsequently, in step S10, N (= 7) is added to the cursor value pos (= 4). Thereby, pos in the fourth block (d) is 4 + 7 = 11. Next, in step S11, the bit value “1” of the fifth bit (pos = 4) from the head of the read fourth block (d) is multiplied by N (= 7), and this is summed (sum == 6). ) To update the total value sum. As a result, the sum in the fourth block (d) is 6 + 1 × 7 = 13. Thereafter, in step S7, the presence or absence of subsequent data is checked. If there is data (step S7: YES), the process returns to step S2, and if there is no data (step S7: NO), the process is terminated.

  In the manner as described above, the same processing is performed for the subsequent blocks after the fifth block. As a result, the pitch converter 3 outputs a bit signal “0101...” Constituting the PDM signal. Then, by passing this PDM signal through a low-pass filter, a PCM signal whose pitch (wavelength) is converted to 7/8 times is obtained.

  Similarly, in the case of FIG. 9B in which the pitch is 3/8 times, pitch conversion can be performed according to the same procedure as in FIG. 9A described above (details are omitted).

  As described above, in the above-described embodiment, the pitch of the 1-bit signal constituting the PDM signal is shifted by the interpolation process, the interpolated data sequence is read out in units of blocks, and quantization is performed for each block. The pitch can be converted without changing the bit signal. For this reason, complicated arithmetic processing such as convolution calculation and general harmonic analysis is not required, and pitch conversion processing can be performed at high speed by simple means. Further, by arbitrarily setting the values of N and M, it is possible to easily obtain a waveform having a desired pitch.

  In the above-described embodiment, since the ΔΣ modulator 50 is used as the quantizing means, the quantization and the remainder can be obtained by utilizing the function of the existing ΔΣ modulator without designing a dedicated arithmetic circuit. Value addition processing can be easily realized.

  The present invention is not limited to the case of outputting an analog tone signal, but can also be applied to the case of outputting a digital tone signal.

  In the above embodiment, the PCM signal is taken as an example of the musical sound signal stored in the waveform memory 1, but the present invention can also be applied to the case where a musical sound signal other than the PCM signal is used.

  Furthermore, the present invention is not limited to musical sounds, and can be applied to the generation of all sounds including human voices and pseudo sounds, and is also applied to the generation of waveforms other than sounds. be able to.

It is a block diagram which shows the whole structure of the synthesizer for sound sources by embodiment of this invention. It is a figure showing the structure of the pitch conversion part. It is a figure which shows a PCM signal and a PDM signal. It is a block diagram which shows an example of a delta-sigma modulator. It is the figure which showed the principle of the pitch conversion by this invention. It is the figure which showed the mode of the zero value interpolation in detail. It is the flowchart which showed the procedure of pitch conversion. It is the figure which showed the example of the data calculation in the case of extending pitch. It is the figure which showed the example of the data calculation in the case of shortening a pitch. It is a figure explaining wavelength conversion.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Waveform memory 2 PCM / PDM conversion part 3 Pitch conversion part 3b Zero value interpolation part 3c Total value calculation part 3d Quantization part 4 Signal processing part 5 Output part 50 ΔΣ modulator 100 Waveform generator

Claims (3)

In a waveform generation apparatus having a waveform memory in which waveform data is stored, and a pitch conversion unit for converting the pitch of a waveform generated based on the waveform data to N / M times,
The pitch converter is
Interpolating means for interpolating N−1 zero values between each 1-bit signal constituting a PDM (Pulse Density Modulation) signal obtained from the waveform data;
The data string after being interpolated by the interpolating means is sequentially read out with a block containing M bits as one unit, and for each block, the bit value at the bit position determined by the values of N and M and the value of N A total value calculating means for calculating a total value X obtained by summing the multiplication values;
By performing quantization using the value of M as a threshold for the total value Y obtained by adding the total value X in the block and the remainder value obtained by quantization in the previous block, 1 is assigned to each block. Or a quantization unit that assigns a value of 0 and outputs the value, and adds the remainder value obtained by the quantization to the total value X of the next block;
A waveform generation apparatus comprising: The waveform generation device according to claim 1,
A waveform generating apparatus characterized in that the quantizing means comprises a ΔΣ modulator. The waveform generation device according to claim 1 or 2,
A signal processing unit that performs predetermined signal processing on the waveform of the sound generated by the waveform generation device;
An output unit that outputs sound based on the signal processed by the signal processing unit;
A synthesizer for sound sources characterized by comprising

Images