JP2004053932A - Waveform reproducing apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a musical sound waveform reproducing apparatus for reproducing high quality musical sound for auditory sensation. <P>SOLUTION: A sound source device 210 of a PSOLA (Pitch Synchronous Overlap Addition) method and a HPF 220 are arranged in a DSP 200, and the HPF 220 is arranged in the route where the musical sound waveform outputted from the sound source device 210 passes through. The further the musical waveform is away from an original pitch stored in waveform memory, the higher Q-value is set to this HPF 220, and a cut-off frequency Fc is set in the proximity of the fundamental frequency. High quality musical sound is obtained for auditory sensation by reproducing a musical sound waveform with emphasis on the fundamental frequency component and generating the musical sound waveform with emphasis on the fundamental frequency through a loudspeaker. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a musical tone waveform reproducing device for reproducing a musical tone waveform.
[0002]
[Prior art]
There are various sound source devices used in the musical tone waveform reproducing device. There is a method of reading a musical sound waveform from a waveform memory at an address interval according to a pitch and reproducing a musical sound, and a method described in Japanese Patent No. 3124349.
[0003]
The former sound source device has a drawback that the formant and the sounding time change when the reproduced pitch changes, but the latter sound source device has the feature that the drawback is improved and a tone is generated while the formant is maintained. . In the following description, in order to simplify the description, the former will be referred to as a simple sampling type sound source device, and the latter will be referred to as a PSOLA (Pitch Synchronous OverLap Addition) type sound source device.
[0004]
FIG. 6 is a diagram showing frequency characteristics of a tone waveform reproduced by the simple sampling method and a tone waveform reproduced by the PSOLA method.
[0005]
FIG. 6A is a diagram showing the spectrum of the fundamental frequency components and harmonics of the musical tone waveform of the original pitch stored in the waveform memory, and FIG. 6A shows the envelope connecting the peaks of the respective spectrum components. The formant of this instrument is shown.
[0006]
FIG. 6B is a diagram showing the frequency characteristics of a tone waveform reproduced at a reproduction pitch three times the original pitch in the simple sampling system, and FIG. 6C is a diagram showing the original pitch in the simple sampling system. FIG. 6D is a diagram showing the frequency characteristics of a musical tone waveform reproduced at a reproduction pitch of half the frequency of the original tone, and FIG. 6D shows a tone waveform reproduced at a reproduction pitch of three times the original pitch in the PSOLA method; FIG. 6E is a diagram showing the frequency characteristics of a musical tone waveform reproduced by the PSOLA method at a reproduction pitch of a half frequency of the original pitch. Also shows the spectrum of fundamental frequency components and harmonics.
[0007]
As shown in FIGS. 6B and 6C, when the pitch is changed by the simple sampling method, the formant changes, and a tone that is far from the tone originally possessed by the musical instrument is generated. I will.
[0008]
On the other hand, as shown in FIGS. 6D and 6E, in the PSOLA system, the pitch is changed while maintaining the formant of the original pitch.
[0009]
The principle of the PSOLA system will be described with reference to the waveform diagram of FIG.
[0010]
FIG. 7A shows a musical tone waveform of an original pitch composed of a plurality of periods and window functions window1, window2, window3, window4, and window5 for extracting the musical tone waveform. As shown in FIG. 7 (A), the first two periods of the musical tone waveform having a plurality of periods are cut out by the window function window1, and the next two-period musical tone waveform shifted by one period is cut out by the window2. Segmentation is performed by a window function. In order to show the situation where a musical tone waveform is cut out, FIG. 7B shows a musical tone waveform cut out by a window function window1 (here, the first two periods in FIG. ) Are excerpted and shown.
[0011]
In FIG. 7 (C), the reproduction cycle is extended by being partially superimposed so that the musical tone waveforms respectively cut out by the window functions window1, window2, and window3 are connected. This reduces the pitch but reproduces the tone waveform with the formants preserved.
[0012]
When a high pitch is reproduced, for example, as shown in FIG. 7 (D), the musical tone waveform of the first two cycles extracted by the window function window1 is repeated three times in a short cycle corresponding to the pitch. The musical tone waveform cut out by window2 that is repeatedly superimposed and repeated and the musical tone waveform cut out by window3 are further superimposed and synthesized three times each. In the example of FIG. 7 (D), a musical tone waveform cut out by window is superimposed three times in a row in order to match the sounding time.
[0013]
As described above, when the tone is changed to a higher pitch and the musical sound waveforms cut out by the window function are superimposed and synthesized in a short cycle, the low-frequency components are reduced as shown in FIG. A high frequency component becomes a musical sound waveform with a relatively increased frequency component.
[0014]
[Problems to be solved by the invention]
As described above, when the PSOLA method is used, the formant is maintained, but if the reproduced pitch is much lower than the original pitch, the low-frequency component is insufficient, resulting in a musical sound with a sense of incongruity in hearing. This is because, as a characteristic of a voice or an acoustic instrument, the lower the pitch, the higher the proportion of low-frequency components.
[0015]
Further, when the musical tone waveform is reproduced at a reproduction pitch higher than the original pitch using the PSOLA method, the fundamental frequency component of the musical tone waveform is reduced, and the high-frequency component is relatively increased. As a result, there is a problem in that the musical tone becomes uncomfortable in terms of hearing and the quality of the musical tone cannot be guaranteed.
[0016]
SUMMARY OF THE INVENTION It is an object of the present invention to provide a musical sound waveform reproducing apparatus capable of reproducing musical sounds that are not uncomfortable in hearing.
[0017]
[Means for Solving the Problems]
The musical tone waveform reproducing device of the present invention includes a waveform storage unit that records a musical tone waveform of a predetermined original pitch.
A waveform reading unit that reads out the musical tone waveform stored in the waveform storage unit at a predetermined reading speed;
A multiplying unit for multiplying the tone waveform read by the waveform reading unit by a window function;
A waveform synthesizing unit that synthesizes a musical tone waveform at a cycle corresponding to a desired reproduction pitch after the window function has been multiplied by the multiplying unit;
And a filter section for applying a high-pass filter having a Q value for emphasizing a fundamental frequency component defining a reproduction pitch of the musical tone waveform to the musical tone waveform synthesized by the waveform synthesizing portion. .
[0018]
According to the musical sound waveform reproducing apparatus of the present invention, the reproduced sound of the musical sound waveform is converted into a musical sound waveform after being multiplied by the window function in the multiplying unit and synthesized in the waveform synthesizing unit at a cycle corresponding to a desired reproduced pitch. By applying a high-pass filter having a Q value that emphasizes the fundamental frequency component that defines the pitch, a tone waveform in which the fundamental frequency component that defines the reproduced pitch is emphasized is reproduced. If the musical sound waveform in which the fundamental frequency component is emphasized is emitted into a space through a speaker, a musical sound having excellent quality in terms of audibility is obtained.
[0019]
Here, the filter unit is
It is preferable that a high-pass filter having a high Q value is applied as the reproduced pitch departs from the original pitch.
[0020]
As described above, when a high-pass filter having a high Q value is applied to the vicinity of the fundamental frequency component as the pitch is further away from the original pitch, the fundamental frequency component is emphasized in a musical tone waveform having a pitch farther from the original pitch.
[0021]
As described above, as the pitch becomes higher than the original pitch, a high-pass filter having a high Q value is applied near the fundamental frequency component to emphasize the fundamental frequency component. Can be pronounced.
[0022]
Also, as the reproduced pitch becomes lower, a high-pass filter having a high Q value is applied to the vicinity of the fundamental frequency component to emphasize the fundamental frequency component. Moreover, it is possible to produce musical tones with excellent quality.
[0023]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described.
[0024]
FIG. 1 is a diagram showing the internal configuration of a musical tone waveform reproducing apparatus according to an embodiment of the present invention. FIG. 2 is a block diagram showing the configuration of the digital signal processor (hereinafter referred to as DSP) shown in FIG. In the DSP, a high-pass filter (hereinafter, referred to as HPF) is provided in a path through which a PSOLA type sound source device and a musical tone waveform output from the sound source device pass.
[0025]
An outline will be described with reference to FIG.
[0026]
The tone waveform reproducing apparatus is controlled by the CPU 100, and the DSP 200 functioning as a tone generator is also controlled by the CPU 100 to generate a tone waveform. There is also provided a waveform memory 300 in which the tone waveform read out by the DSP 200 is stored. The waveform memory stores tone waveforms of the original pitch for a plurality of cycles.
[0027]
The keyboard 400 has a plurality of keys. When the keys are depressed or released, the information is detected by the CPU 100 and key depression information and key release information corresponding to the depressed keys are transmitted to the DSP 200. Can be The key press information includes the pitch information of the pressed key. When the key press information is transmitted to the DSP, the DSP 200 reads a tone waveform corresponding to the key press information from the waveform memory 300 until the key release information corresponding to the key press information is transmitted.
[0028]
The DSP 200 is provided with a PSOLA-type sound source and a high-pass filter. The D / A converter 500 converts the tone waveform into an analog tone waveform signal and outputs the signal to a speaker. The procedure for the CPU 100 to perform the processing is shown in the ROM 600, and the processing information used by the CPU 100 is temporarily stored in the RAM 700.
[0029]
FIG. 2 is a diagram showing the internal configuration of the DSP 200.
[0030]
A sound source device 210 of the PSOLA system is provided in the DSP 200 provided in the tone waveform reproducing device of the present embodiment, and an HPF 220 is provided on a path through which the tone waveform output from the tone source device 210 passes.
[0031]
FIG. 3 is a detailed diagram showing the internal configuration inside the DSP 200.
[0032]
First, an outline of the operation of the sound source device 210 will be briefly described with reference to FIG.
[0033]
In the detailed diagram shown in FIG. 3, a waveform memory 300 is shown in addition to the DSP, so that the relationship between the DSP 200 and the waveform memory 300 can be understood.
[0034]
FIG. 3 shows a case where a tone waveform is reproduced at an original pitch for the sake of simplicity. Here, since the pitch is the original pitch, the address signal output from the counter and the pulse can be described by two each, and the waveform readout unit can be described by two of 210a and 210b. When reproducing waveforms, three or more waveforms are required according to the pitch. Further, FIG. 3 shows a waveform memory 300a and a waveform memory 300b, which indicate that waveforms are read out independently from the waveform memory 300. The outputs are added by an adder 214a and output. Waveform reading units 210a and 210b are provided in front of the waveform memories 300a and 300b, respectively.
[0035]
The counter 211 generates a timing signal so that the window function is multiplied at a predetermined timing by the tone waveform read based on the read start address specified by each of the waveform read units 210a and 210b.
[0036]
In the counter 211, the pulse C1 and the pulse C2 are repeatedly generated at intervals of twice the period of the original pitch, and the pulse C2 is generated with a shift from the pulse C1 by one period. Based on these pulses C1 and C2, the counter 211 generates address signals Phase1 and Phase2 which are sequentially counted up based on the sampling frequency.
[0037]
One address signal Phase1 is an address signal that is sequentially counted up at the same frequency as the sampling frequency starting from the time when the pulse C1 is supplied, and is reset by the pulse signal C1 after a lapse of two periods of the original pitch. This address signal is repeatedly generated for two periods of the original pitch frequency. The read start address supplied from the waveform number table 215a is added to the sequentially counted up address signal, and the musical tone waveform for two cycles is stored in the waveform memory 300a from the waveform memory 300a at the same sampling frequency. Is read. The address signal Phase1 is also supplied to the cosine waveform generation circuit 212, and the cosine waveform COS1 is sequentially read by the address signal. The cosine waveform generation circuit 212 includes a plurality of cosine tables. One of the cosine tables generates COS1 corresponding to Phase1, and the other cosine table generates COS2 corresponding to Phase2. . In FIG. 7, the window function is indicated by a triangular wave for easy understanding, but actually, a window function of a cosine waveform is generated in the tone generator 210.
[0038]
The signal COS1 output from the cosine waveform generation circuit 212 becomes a window function, and a part of a musical tone waveform having a plurality of periods is cut out over two periods of the original high frequency. The multiplier 213a multiplies the musical tone waveform read from the waveform memory 300a by the window function COS1 output from the cosine waveform generating circuit 212 and cuts out the musical tone waveform shown in FIG. 7B. As shown in FIG. 7A, a window function such as window1, window3, window5... Corresponding to COS1 is supplied to one multiplier 213a every two cycles, and a plurality of cycles are read from the waveform memory 300a. A musical tone waveform of the original pitch is sequentially cut out every two periods.
[0039]
Further, window functions such as window2, window4, window6... Shown in FIG. 7A are supplied to the other multiplier 213b, and a plurality of periods of original tone waveforms read out from the waveform memory 300b are sequentially output every two periods. It has been cut out.
[0040]
Each of the musical tone waveforms cut out by the multipliers 213a and 213b is added and combined by an adder 214a at the subsequent stage.
[0041]
These multipliers 213a and 213b correspond to a multiplying unit according to the present invention, and the adder 214a corresponds to a waveform synthesizing unit according to the present invention.
[0042]
Hereinafter, the waveform reading units 210a and 210b will be described.
[0043]
Since the waveform memories 300a and 300b store the musical tone waveforms of a plurality of cycles, the waveform reading sections 210a and 210b determine from which address of the musical tone waveforms of the plurality of cycles the reading is performed.
[0044]
Since the two waveform memories 300a and 300b are provided with the waveform reading units 210a and 210b having the same configuration, respectively, the following description focuses on the configuration of the one waveform memory 300a.
[0045]
The waveform reading section 210a has a waveform number table 215a for determining a read start address, and a waveform number counter 216a in the preceding stage of the waveform number table 215a is set with what number of periods of a musical tone waveform for a plurality of periods are to be read. Is provided. A pulse C1 is supplied from the counter 211 to the waveform number counter 216a, and the waveform number counter 216a determines which cycle of the musical tone waveform in a plurality of cycles is to be read according to the value counted up by the pulse C1. Is set.
[0046]
The read start address output from the waveform number table 215a is added to the address signal Phase1 output from the counter 211 by the adder 214b. Since the address signal Phase1 from the counter 211 is a signal that is sequentially counted up at the same frequency as the sampling frequency, the reading of the musical tone waveform is performed by the address sequentially counted up from the reading start address. Then, the address signal counted up by the pulse C1 repeatedly generated every two periods of the original sound high frequency is reset to the initial value, and the tone waveform is read from the waveform memory based on the next read start address.
[0047]
The address signal Phase1 from the counter 211 is added to the read start address supplied from the waveform number table 215a in this way, and the read start address is selected from among the plurality of cycles of the musical tone waveform stored in the waveform memory 300a. Tone waveforms of the corresponding cycle are sequentially read. On the multiplier 213a side, the musical tone waveform extracted by the window function COS1 is supplied to the adder 214a on the multiplier 213b side, and the musical tone waveform extracted by the COS2 is supplied to the adder 214a for each period of the reproduced pitch.
[0048]
FIG. 4A is a diagram showing the relationship between the cutoff frequency of the HPF 220 and the reproduced pitch, and FIG. 4B is a diagram showing the relationship between the Q value of the HPF 220 and the reproduced pitch.
[0049]
The cutoff frequency (Fc) and the Q value are given as coefficients from the CPU 100 to the HPF 220 in the DSP 200. With respect to the cutoff frequency Fc given as a coefficient of the HPF from the CPU 100, as shown in FIG. 4A, the cutoff frequency Fc corresponding to the reproduced pitch is given as a coefficient based on the frequency corresponding to the original pitch. As the reproduced pitch departs from the original pitch, a higher Q value is given as a coefficient.
[0050]
The cutoff frequency Fc and the Q value of the HPF 220 are set in the DSP 200 according to the pitch of the key pressed in this manner.
[0051]
FIG. 5A is a diagram in which the frequency characteristics of the HPF 220 are superimposed on the frequency characteristics of a musical tone waveform reproduced at a pitch higher than the original pitch by the sound source device 210, and FIG. FIG. 7 is a diagram showing the frequency characteristics of the HPF 220 superimposed on the frequency characteristics of a musical tone waveform reproduced at a pitch lower than the original pitch generated in FIG. The frequency characteristic of the HPF 220 is indicated by a one-dot chain line in the figure.
[0052]
FIG. 5A shows an example in which an HPF 220 having a high Q value in the vicinity of the fundamental frequency is formed. Here, an example is shown in which a tone waveform having a reproduction pitch of three times the original pitch is reproduced. ing. Therefore, the conventional problem that the fundamental frequency component is reduced is solved by adding the HPF 220, and a tone waveform in which the fundamental frequency component is emphasized is reproduced. When the musical tone waveform in which the fundamental frequency component is emphasized is sounded through a speaker, the musical tone has a high audibility.
[0053]
FIG. 5B shows a frequency spectrum when a musical tone waveform is reproduced at a reproduction pitch of half the original pitch. As shown in FIG. 4B, when the reproduced pitch is lower than the original pitch, a considerably high Q value is set in the HPF 220, and the fundamental frequency component is emphasized. Then, even if the musical tone has a lower pitch than the original pitch and cannot be originally produced by the musical instrument, the musical tone waveform which would have been produced if the musical instrument can produce the low tone is reproduced. . When the tone waveform in which the fundamental frequency component is emphasized is emitted as a tone through a speaker, the sense of incongruity is eliminated, and the tone becomes high in audibility.
[0054]
As described above, even if the reproduced pitch is higher than the original pitch or lower than the original pitch, the tone waveform in which the fundamental frequency component is emphasized by the HPF is reproduced. Further, the HPF having a higher Q value is set as the reproduced pitch is further away from the original pitch, so that the fundamental frequency component is further emphasized as the playback pitch is further away from the original pitch, and the musical tone waveform is reproduced. When a musical sound waveform in which the fundamental frequency component is emphasized in this way is generated through a speaker, the musical sound has excellent audibility.
[0055]
【The invention's effect】
As described above, according to the present invention, it is possible to provide a musical tone waveform reproducing apparatus capable of reproducing musical sounds having high quality in terms of hearing.
[Brief description of the drawings]
FIG. 1 is a diagram showing a configuration of a musical tone waveform reproducing device according to an embodiment of the present invention.
FIG. 2 is a block diagram showing an internal configuration of the DSP shown in FIG.
FIG. 3 is a diagram showing an internal configuration of a PSOLA type sound source device.
FIG. 4 is a diagram illustrating a relationship between a reproduced pitch and a filter coefficient of an HPF.
FIG. 5 illustrates how the HPF operates when a tone waveform having a pitch higher than the original pitch is reproduced and how the HPF operates when a tone waveform having a pitch lower than the original pitch is reproduced. FIG.
FIG. 6 is a diagram showing a spectrum of a fundamental tone and harmonics of a musical tone waveform reproduced by the simple sampling method and the PSOLA method.
FIG. 7 shows a state where a part of a musical tone waveform cut out by a window function is multiplexed and synthesized when a reproduced pitch is changed based on an original pitch and reproduced by a PSOLA type sound source device. FIG.
[Explanation of symbols]
100 CPU
200 DSP
210 sound source device 210a 210b waveform readout unit and address selection unit 211 counter 212 cosine waveform generation circuit 213a 213b multipliers 214a to 214c adders 215a 215b waveform number table 216a 216b waveform number counter 220 HPF
300 Waveform memory 300a 300b Waveform memory 400 Keyboard 500 D / A
600 ROM
700 RAM

Claims (2)

A waveform storage unit for recording a musical tone waveform of a predetermined original pitch,
A waveform reading unit that reads out the tone waveform stored in the waveform storage unit at a predetermined reading speed;
A multiplication unit that multiplies the musical tone waveform read by the waveform reading unit by a window function,
A waveform synthesizing unit that synthesizes the tone waveform after being multiplied by the window function in the multiplying unit at a cycle corresponding to a desired reproduction pitch;
A musical sound waveform reproducing apparatus comprising: a filter section for applying a high-pass filter having a Q value that emphasizes a fundamental frequency component defining a reproduced pitch of the musical sound waveform to the musical sound waveform synthesized by the waveform synthesizing section. The filter unit includes:
2. The musical sound waveform reproducing apparatus according to claim 1, wherein a high-pass filter having a high Q value is applied as the reproduced pitch is away from the original pitch.

Images