JP4012410B2 - Musical sound generation apparatus and musical sound generation method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a musical sound generating apparatus and a musical sound generating method for reading musical sound waveform data sampled and stored in a waveform memory and generating a musical sound based on the musical sound waveform data.
[0002]
[Prior art]
2. Description of the Related Art A so-called waveform table (wave table) sound source that samples a sound of a musical instrument, stores it in a waveform memory as musical sound waveform data, and reads out the musical sound waveform data to generate a musical sound is conventionally known.
[0003]
[Problems to be solved by the invention]
In the above conventional waveform table sound source, in order to reduce the storage capacity of the waveform memory, the original sampling frequency of the sound source system, specifically, the tone waveform data stored in the waveform memory is read to generate a tone. Musical sounds may be sampled at a frequency lower than the frequency. Hereinafter, the original sampling frequency of the sound source system is referred to as “sampling frequency”, and the sampling frequency when the musical sound is sampled to generate musical sound waveform data is referred to as “recording sampling frequency”.
[0004]
Since the recording sampling frequency corresponds to the number of musical sound waveform samples (individual waveform samples constituting the musical sound waveform data) per unit time (1 second), the lower the recording sampling frequency, The capacity is reduced, and as a result, the storage capacity of the waveform memory is reduced.
[0005]
However, according to the sampling theorem, the recording sampling frequency determines the upper limit frequency of the reproducible musical sound (= half of the recording sampling frequency) (for example, when the recording sampling frequency is 16 KHz, the frequency up to 8 KHz). In other words, if the recording sampling frequency is kept low, only the music with the higher harmonic components lost can be reproduced, and the sound quality of the music reproduced in this case will deteriorate. Become.
[0006]
The present invention has been made paying attention to this point, and even when the musical sound waveform data is generated with the recording sampling frequency lower than the sampling frequency, the lost harmonic component is recovered to make it more natural. An object of the present invention is to provide a musical sound generating apparatus and a musical sound generating method capable of generating various musical sounds.
[0007]
[Means for Solving the Problems]
  In order to achieve the above object, the claims2The musical tone generator described in the above section uses musical sound waveform data collected by sampling the sound of musical instruments at the recording sampling frequency.Tone data memoryRememberTone data memoryIn a musical sound generation device that reads musical sound waveform data from a musical sound and generates a musical sound,Higher than the recording sampling frequencySampling by frequencyObtained by down-sampling the first musical sound waveform data stored in the first memory at the recording sampling frequency, the first storage means for storing in the first memory as the first musical sound waveform data A second storage means for storing data in the second memory as second musical sound waveform data, and a frequency spectrum in a frequency range lost by down-sampling the first musical sound waveform data at the recording sampling frequency. Formant data generating means for generating the formant data based on the first musical sound waveform data, and the second musical sound waveform data and the formant data generated by the formant data generating means are associated with each other and stored in the timbre data memory. And a second storage means for storing the second storage means stored in the timbre data memory. And tone waveform samples generating means for generating a musical tone waveform samples from the sound waveform data, stored in the tone color data memory, associated with the second musical sound waveform dataBased on formant dataTehuA synthesis means for synthesizing formant sounds;Before the generated musical sound waveform sampleAdd the synthesized formant soundThen outputAnd adding means.
[0008]
  GoodPreferably, the formant data is a center frequency and a level of the formant, and the synthesizing means includes white noise generating means for generating white noise, and spectral characteristics of white noise generated by the white noise generating means. A sine wave generating means for generating a sine wave having the center frequency, a multiplying means for multiplying the output from the changing means by the sine wave generated by the sine wave generating means, and the multiplication And adjusting means for adjusting the output level from the means so as to be the level described above.
[0009]
  In order to achieve the above object, the claims1The musical sound generation method described inRecordMusical sound waveform data collected by sampling the sound of an instrument at the sound sampling frequencyTone data memoryRememberTone data memoryIn the musical sound generation method for generating musical sound by reading musical sound waveform data fromHigher than the recording sampling frequencySampling by frequencyThe first musical sound waveform data is stored in the first memory, and the data obtained by down-sampling the first musical sound waveform data stored in the first memory at the recording sampling frequency is stored in the second memory. Formant data is stored in the second memory as musical sound waveform data, and formant data based on a frequency spectrum in a frequency range lost by down-sampling the first musical sound waveform data at the recording sampling frequency is the first musical sound waveform data. And the second musical tone waveform data and the generated formant data are stored in the timbre data memory in association with each other, and when generating a musical tone, the second musical tone data stored in the timbre data memory is generated. A musical sound waveform sample is generated from the waveform data, and the second musical sound waveform data stored in the tone color data memory is stored. Associated with theBased on formant dataTehuSynthesizing formant sounds,Before the generated musical sound waveform sampleAdd the synthesized formant soundThen outputIt is characterized by doing.
[0010]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0011]
FIG. 1 is a block diagram showing a schematic configuration of a mobile phone to which a musical sound generating device according to an embodiment of the present invention is applied.
[0012]
As shown in the figure, the control unit 1 is used for playing a CPU that controls the entire mobile phone, a control program executed by the CPU, a ROM that stores various table data, and a ringing melody. It is composed of performance data (for example, composed of MIDI (Musical Instrument Digital Interface) data), various input information, calculation results, and the like in RAM.
[0013]
The control unit 1 includes an operation input unit 2 having a numeric keypad and an operator for inputting various information, a display unit 3 having, for example, a color liquid crystal display (LCD) and a light emitting diode (LED), and analog audio. The signal is converted into a digital audio signal and then compressed, and conversely, the compressed digital audio signal is expanded and then converted into an analog audio signal, and the output signal from the audio CODEC 5 is modulated, A communication unit 4 that transmits to a relay station (not shown), receives and demodulates a signal transmitted from the relay station via the antenna 7, outputs the signal to the voice CODEC 5, and a tone color data memory (waveform memory) to be described later ), The target musical tone waveform data is read from the tone color data memory, and various processes are performed to generate a digital musical tone signal. A waveform table tone generator 6 for converting the analog musical tone signal is connected by C (Digital-to-Analog Converter).
[0014]
Connected to the audio CODEC 5 are a speaker 8 for converting an analog audio signal output from the audio CODEC 5 into sound and a microphone 9 for converting audio into an analog audio signal.
[0015]
The waveform table sound source 6 is connected to a speaker 10 for converting an analog musical sound signal output from the waveform table sound source 6 into sound.
[0016]
The feature of the present invention resides in the waveform table sound source 6, particularly the musical tone signal generation processing thereof. Specifically, since the musical tone waveform sample collected at a recording sampling frequency lower than the sampling frequency is stored in the timbre data memory, if the musical tone waveform sample is read as it is, a frequency component more than half of the recording sampling frequency is stored. Cannot be reproduced, that is, only musical tones with high-order harmonic components lost can be reproduced. In order to deal with this problem, in the original sound, a plurality of formants (four in this embodiment) appearing in a high frequency more than half of the recording sampling frequency are detected, and the formant center frequency and formant of each formant are detected. Each level is extracted and stored in the tone color data memory in association with the musical sound waveform sample. When the target musical sound waveform sample is read out and a musical sound is generated, each formant center frequency and formant level associated with the musical sound waveform sample are read out, and a plurality of formants of each formant center frequency and formant level are provided. Formant sounds are generated approximately and added to the target musical sound waveform sample.
[0017]
As described above, as a precondition for performing the musical tone signal generation process that is a feature of the present invention, the plurality of formants are detected, the formant center frequency and formant level of each formant are extracted, and the recording frequency is lower than the sampling frequency. A musical sound sample is generated by sampling a musical sound at a sampling frequency, and musical sound waveform data including the musical sound waveform sample and the corresponding formant center frequency and formant level must be registered in the timbre data memory. . Note that the same original sound must be sampled in both the processing for detecting the plurality of formants and the processing for generating a musical sound waveform sample. However, the sampling frequency used for sampling differs between the two, and the former needs to be sampled at a sampling frequency that is twice that of the latter. For this reason, in the present embodiment, the original musical sound is sampled only once at the former sampling frequency, that is, twice the recording sampling frequency, and the latter processing uses the downsampled one. I have to.
[0018]
Hereinafter, processing from sampling to registration in the timbre data memory, that is, timbre data memory creation processing will be described.
[0019]
FIG. 2 is a flowchart showing the procedure of the timbre data memory creation process. The real tone color data memory creation process is executed on a personal computer, for example.
[0020]
In the figure, first, a musical sound waveform sample is collected, that is, the original sound is sampled (step S1).
[0021]
FIG. 3 is a diagram for explaining the processing for collecting a musical sound waveform sample in step S 1, and shows a method for collecting a musical sound waveform sample of a cymbal sound by the personal computer 100.
[0022]
As shown in the figure, first, a cymbal sound is generated by hitting a cymbal with a stick, and this cymbal sound is converted into an analog musical sound signal by a microphone 101.
[0023]
Next, an LPF (low-pass filter) 102 that passes only frequency components equal to or lower than the recording sampling frequency Wfs cuts frequency components equal to or higher than Wfs included in the analog musical tone signal. In this way, the frequency component of Wfs or higher is cut because the musical sound having a frequency component of half (= Wfs) or more of the sampling frequency (= 2 Wfs: double the recording sampling frequency) for generating the original musical sound waveform sample is cut. Is not reproducible as described above, and therefore a frequency component higher than Wfs is not necessary.
[0024]
Next, the output signal from the LPF 102 is converted into a digital musical tone signal by the A / D converter 103 having a sampling frequency of 2 Wfs, and this digital musical tone signal is stored in a memory (for example, RAM) 104.
[0025]
When the above processing is repeated for a predetermined time, for example, 1 second, 2 Wfs musical sound waveform samples are stored in the memory 104. A time series sample after down-sampling by thinning out one sample from the time series of the musical sound waveform samples is a musical sound waveform sample to be stored in the timbre data memory.
[0026]
In addition, since the musical tone waveform samples of a plurality of timbres are registered in the timbre data memory, the musical tone waveform samples that are the basis of the musical instrument sounds other than the cymbal sounds are collected in the same manner as described above. That's fine.
[0027]
Returning to FIG. 2, the musical sound waveform sample collected in step S1 is processed and edited (step S2). Examples of the processing / editing process include a zero level (or low level) waveform sample deletion process, an amplitude adjustment process of each musical sound waveform sample, the downsampling process, and a high frequency analysis process of the underlying musical sound waveform sample. be able to.
[0028]
FIG. 4 is a diagram showing the frequency spectrum of the basic musical sound waveform sample, the down-sampled musical sound waveform sample, and the formant after detection. FIG. 4A shows the frequency spectrum of the basic musical sound waveform sample. An example is shown, (b) shows the frequency spectrum after down-sampling the musical sound waveform sample of (a) by 1/2, and (c) shows four formants detected from the musical sound waveform sample of (a). The frequency spectrum is shown.
[0029]
As described above, since the original musical sound waveform sample is generated by sampling the original sound at a frequency twice the recording sampling frequency Wfs, the frequency spectrum is maximum, as shown in FIG. Appears up to frequency Wfs.
[0030]
If the time series of the sound waveform samples that are the basis of this is downsampled by cutting out frequency components of Wfs / 2 or more by a downsampling filter and then thinning out one sample at a time, this is exactly the case when the original sound is sampled at the recording sampling frequency Wfs Therefore, the frequency spectrum of the musical sound waveform sample after the down-sampling is a frequency spectrum of Wfs / 2 to Wfs in the frequency spectrum of (a) as shown in (b). . The musical tone waveform sample after the down-sampling is registered in the timbre data memory as will be described later.
[0031]
When FFT (Fast Fourier Transform) is performed on a part (one frame in this embodiment) of musical sound waveform samples selected from the time series of the musical sound waveform samples as the base, the frequency spectrum of (a) (Because the frequency spectrum of (a) is the frequency spectrum obtained from all the time series of the underlying musical sound waveform samples as described above). In this frequency spectrum, four formants are detected by analyzing the frequency range of Wfs / 2 to Wfs. (C) shows the frequency spectrum of the detected formant. Then, the center frequency and level are extracted from each formant of the detected four formants. Specifically, in (c), formant center frequencies of F1 to F4 and formant levels of L1 to L4 are extracted. The extracted formant center frequency and formant level are stored in the timbre data memory in association with a musical sound waveform sample for one frame.
[0032]
In this embodiment, the formant center frequency and the formant level are extracted for each of a plurality of predetermined musical sound waveform samples out of all the musical sound waveform samples. Only one set common to samples may be extracted.
[0033]
Returning to FIG. 2, the musical sound waveform data processed and edited in step S2 is registered in the timbre data memory (step S3).
[0034]
FIG. 5 is a diagram showing an example of a memory map of the timbre data memory. In the present embodiment, timbre data for one timbre is composed of basic timbre data and musical tone waveform data. Since a plurality of timbre data are registered in the timbre data memory, a timbre number (No) (for example, a timbre number in the GM (General MIDI) system format) is assigned to the head address of the timbre data memory. A conversion table for converting to the start address of the area where the basic timbre data corresponding to the timbre number is stored is stored.
[0035]
The basic timbre data for one timbre includes the timbre name, the data length of the basic timbre data, the recording sampling frequency Wfs, the waveform start address indicating the address of the first sample of the musical sound waveform data, and the loop reading of the musical sound waveform data. Waveform loop start address indicating the address of the first sample in the area to be read, waveform loop end address indicating the address of the last sample in the area to be read in the loop, waveform end address indicating the address of the last sample of the sound waveform data, and the sound waveform It consists of envelope data and other data that determine the data envelope.
[0036]
In the present embodiment, the basic tone color data and the musical tone waveform data are stored in different areas, and therefore, in the basic tone color data, from which area the musical tone waveform data of the relevant tone is stored. Data indicating how far it is, that is, a waveform start address and a waveform end address are included. However, if it is configured to store the corresponding tone waveform data following the basic tone data, the start address of the tone waveform data can be calculated from the data length in the basic tone data, so the waveform start address is stored. You may make it not.
[0037]
On the other hand, the musical tone waveform data for one tone color was extracted by down-sampling the time series of the basic musical tone waveform sample as described above and the high frequency analysis for the musical tone waveform sample for one frame. Each formant is composed of formant data consisting of formant center frequencies and formant levels. One frame means a musical sound waveform sample for 10 msec, for example, and if the recording sampling frequency is Wfs, it corresponds to Wfs × 0.01 musical sound waveform samples.
[0038]
In the timbre data registration process in step S3, musical tone waveform data including the downsampled musical tone waveform sample and the extracted formant data is registered so as to conform to the format of the timbre data memory of FIG. Specifically, the musical sound waveform data is stored in the musical sound waveform data storage area, the basic timbre data is created and stored in the basic timbre data storage area, and the head address of the basic timbre data is set to the timbre number. → Store in the corresponding position in the start address conversion table.
[0039]
In the real tone data registration processing, the memory of the personal computer 100 (may be the same as the memory 104 (however, the area needs to be different in this case) or may be a different storage medium). An area having the same capacity as the timbre data memory of FIG. 5 is generated above, and the musical tone waveform data is registered in this area.
[0040]
In the subsequent step S4, it is determined whether or not there is a tone of another tone color to be sampled. If there is, the process returns to step S1 and repeats from the waveform sampling process. If not, the process proceeds to step S5.
[0041]
In step S5, the same memory map as the timbre data registered in the predetermined area of the memory is generated on a timbre data memory comprising, for example, a ROM, that is, converted into a timbre data memory.
[0042]
The outline of the control process executed by the mobile phone configured as described above will be described first, and then will be described in detail with reference to FIGS.
[0043]
In order to reduce the storage capacity of the timbre data memory, the cellular phone according to the present embodiment collects a musical sound waveform sample at a recording sampling frequency lower than the sampling frequency, stores it in the timbre data memory, and is lost thereby. High formant data (formant center frequency and formant level) from half the recording sampling frequency to the recording sampling frequency is extracted and stored in the tone color data memory in association with the musical sound waveform sample. When generating a musical tone, the target musical sound waveform sample is read from the timbre data memory, the formant data corresponding to the musical sound waveform sample is also read, and a formant having a plurality of formants, which is generated approximately based on the formant data. A sound signal is generated by adding a sound to the previously read musical sound waveform sample. As a result, the frequency component that is lost when the musical sound waveform sample is sampled, that is, the formant, which is more than half of the recording sampling frequency, is approximately recovered, so that a more natural musical sound can be generated.
[0044]
Here, as a method for generating formant sound from formant data, the method described in Japanese Patent Application Laid-Open No. Hei 2-271597 filed earlier by the present applicant is used as it is. Therefore, a detailed description thereof will be given in the publication, but will be described later within the scope necessary for carrying out the present invention.
[0045]
Next, this control process will be described in detail. Since the feature of the present invention is exclusively in the tone signal generation process, that is, the tone signal generation process in the waveform table tone generator 6 as described above, hereinafter, the control process (the tone signal generation process) performed in the waveform table tone generator 6 will be described. Processing) will be described.
[0046]
FIG. 6 is a block diagram showing a procedure of a musical sound signal generation process performed by the waveform table sound source 6. Since the waveform table sound source 6 is usually configured by a DSP (digital signal processor), most of the control processing is performed by software. Of course, all of the waveform table sound source 6 may be configured by hardware.
[0047]
In FIG. 6, MIDI data indicating a tone to be generated is input to the parameter generation unit 21, and a memory reading unit 24 that reads the stored contents of the timbre data memory 25 is connected to the parameter generation unit 21.
[0048]
When the MIDI data is input, the parameter generation unit 21 first analyzes the MIDI data and extracts information such as a tone color number and a key code, that is, information necessary for generating a musical tone generation parameter. Next, the parameter generation unit 21 accesses the head address of the timbre data memory 25 via the memory reading unit 24, and uses the timbre number → start address conversion table stored at that position to extract the above-described extraction. The head address of the basic tone color data corresponding to the selected tone color number is searched, the basic tone color data for one tone color stored at that position is read, and the tone generation parameter is generated based on the basic tone color data.
[0049]
The parameters generated at this time are specifically a waveform start address, a waveform end address, a waveform loop address (waveform loop start address and waveform loop end address), and envelope data. All of these parameters are data itself included in the basic tone color data.
[0050]
The parameter generation unit 21 also includes an F number (FN) that is a parameter representing the amount of advance of an address per musical tone waveform sample based on the extracted tone number and key code, the recording sampling frequency, and the sampling frequency. And an OCT parameter which is a parameter indicating octave.
[0051]
Further, the parameter generation unit 21 generates formant data (formant center frequency and formant level). As described above, since the formant data is stored differently for each frame, when the reading of the musical sound waveform sample for one frame is completed, the parameter generation unit 21 passes through the memory reading unit 24. The formant data corresponding to the next frame may be read from the timbre data memory 25 and generated.
[0052]
The various clock generators 22 generate and output clocks having a plurality of frequencies by, for example, dividing the supplied basic clock. The main clock is a clock of the sampling frequency Sfs inherent to the waveform table sound source 6, and the operation of each block 23 to 29 of the waveform table sound source 6 proceeds by this clock.
[0053]
The address generation unit 23 receives the F number, OCT parameter, waveform start address, waveform end address, and waveform loop address generated by the parameter generation unit 21. The address generation unit 23, based on these parameters, An address of a musical sound waveform sample to be read from the timbre data memory 25 is generated.
[0054]
In the present invention, since the recording sampling frequency is lower than the sampling frequency, the address generated by the address generator 23 is usually a real value including a decimal part. That is, the address generator 23 generates an address indicating the position of the musical sound waveform sample between the musical sound waveform samples stored in the timbre data memory 25. Therefore, the musical sound waveform sample at this position is obtained by interpolation from the musical sound waveform samples stored in the timbre data memory 25, as will be described later.
[0055]
Of the addresses composed of real values generated by the address generation unit 23, the integer part is supplied to the memory reading unit 24 and the decimal part is supplied to the interpolation unit 26.
[0056]
The memory readout unit 24 reads out the musical sound waveform sample corresponding to the integer part of the supplied address and a predetermined number of musical sound waveform samples adjacent to the musical sound waveform sample from the timbre data memory 25 and outputs them to the interpolation unit 26. To do. Here, the predetermined number is determined according to the interpolation method employed in the interpolation unit 26. If the interpolation method is, for example, linear interpolation using two points, the predetermined number is one. The reading unit 24 reads out the musical sound waveform sample corresponding to the integer part of the supplied address and the musical sound waveform sample located at the next address, and outputs them to the interpolation unit 26.
[0057]
The interpolation unit 26 interpolates between the plurality of musical sound waveform samples read from the timbre data memory 25 on the basis of the decimal part of the supplied address by a preset interpolation method to obtain the target musical sound waveform sample. Is generated.
[0058]
The musical sound waveform sample output from the interpolation unit 26 is supplied to the addition unit 27. In addition, the noise formant sound output from the noise formant sound generation unit 28, more specifically, one waveform sample of the noise formant sound corresponding to the musical sound waveform sample is supplied to the adding unit 27. The musical sound waveform sample from the interpolation unit 26 and the noise formant sound waveform sample corresponding to the musical sound waveform sample are added, and the addition result is output to the envelope generation / giving unit 29.
[0059]
The envelope generation / giving unit 29 is supplied with the envelope data. The envelope generation / granting unit 29 generates an envelope based on the envelope data, adds the envelope to the musical sound waveform sample from the adding unit 27, and then the DAC. Output to.
[0060]
FIG. 7 is a block diagram showing the procedure of the noise formant sound generation process performed by the noise formant sound generation unit 28. The noise formant sound generation process applies the method described in the above publication.
[0061]
In FIG. 7, the white noise generated from the white noise generation unit 31 passes through an LPF (low pass filter) 32, thereby changing its spectral characteristics, and then generating a sine wave via the multiplication unit 33. Multiply by the sine wave of formant center frequency Fi (i is an integer value of 1 to 4) generated from the unit 34.
[0062]
In this way, the formant-shaped noise sound (noise formant sound) centered on the formant center frequency Fi is output from the multiplier 33 and supplied to the multiplier 35. In addition to this, the formant level Li generated by the parameter generation unit 21 is supplied to the multiplication unit 35, and the multiplication unit 35 outputs the noise formant sound of a predetermined level, for example, unit level, output from the multiplication unit 33 to the formant level. Change to the noise formant sound of Li.
[0063]
The multiplier 35 outputs the noise formant sound having the formant level Li to the accumulator 36.
[0064]
Blocks 31 to 35 perform a time-sharing operation to generate a plurality (four in this embodiment) of noise formant sounds. In other words, in the present embodiment, four formants are detected for a musical sound waveform sample for one frame, and the formant center frequency and formant level are extracted from the four formants, respectively. I'm stuck.
[0065]
The accumulator 36 accumulates four noise formant sounds that are sequentially supplied in a time division manner, and outputs the accumulated noise formant sound to the adder 27 at a predetermined timing. Note that the same noise formant sound is output from the accumulator 36 for musical tone waveform samples of the same frame. FIG. 8 is a diagram showing the transition of the level of the noise formant sound, and the accumulator 36 outputs different discrete values for each frame as shown in FIG. Note that, for example, by performing a filtering process on the output of (a), a continuous value may be output as shown in FIG. Further, the same processing as in (b) may be applied to other data of formant data, that is, the formant center frequency.
[0066]
In the present embodiment, the formant data is extracted and stored for each of a plurality of predetermined musical sound waveform samples among all the musical sound waveform samples, and when generating a musical sound, However, the present invention is not limited to this, but only one set of formant data is extracted and stored for all musical sound waveform samples, and when generating musical sounds, all musical sound waveforms are generated. A noise formant sound common to the samples may be generated. In this case, it is desirable to detect the level of the musical sound to which the noise formant sound should be added, and to change the center frequency and level of the formant included in the noise formant sound in accordance with this detection level. Further, the target to be detected is not limited to the tone level but may be other characteristics.
[0067]
In general, when natural sounds are viewed on the frequency axis, the spectrum of frequencies below 8 KHz is information that recognizes the sound itself, that is, the sound such as piano and violin sounds, human voices, etc. Contains information to know if it exists. On the other hand, the spectrum of about 8 KHz or more to 16 KHz (audible upper limit) often includes information that brings out the naturalness of the sound. Even if the spectrum of this region is completely lost, it is possible to recognize what the sound is, but the naturalness is greatly lost. If the information that brings out the naturalness, that is, the frequency spectrum from 8 KHz to 16 KHz is approximately recovered as in the present embodiment, a more natural musical tone can be generated.
[0068]
Further, in the present embodiment, the musical sound generating device of the present invention is applied to a mobile phone, particularly a sound source device thereof. However, in a normal sound source device provided in an electronic keyboard instrument or the like, a recording sampling frequency lower than the sampling frequency is used. This is because it is not necessary to reduce the storage capacity of the timbre data memory (waveform memory) until the musical sound is sampled and the reproduced musical sound is degraded. Therefore, in other words, the apparatus is not limited to a mobile phone as long as it is a device that needs to sample a musical sound at a recording sampling frequency lower than the sampling frequency.
[0069]
As described above, in this embodiment, the frequency component lost when the musical sound waveform data is generated by sampling at the recording sampling frequency lower than the sampling frequency, that is, the frequency component more than half of the recording sampling frequency, Based on formant data corresponding to the formant of the region and stored in the timbre data memory, a formant sound having a formant that approximates the lost formant is synthesized, and the generated musical sound waveform data is synthesized. By performing the addition, it is possible to generate a more natural musical tone because it is approximately recovered.
[0070]
A program in which a storage medium storing software program codes for realizing the functions of the above-described embodiments is supplied to a system or apparatus, and a computer (or CPU or MPU) of the system or apparatus is stored in the storage medium. It goes without saying that the object of the present invention can also be achieved by reading and executing the code.
[0071]
In this case, the program code itself read from the storage medium realizes the novel function of the present invention, and the storage medium storing the program code constitutes the present invention.
[0072]
As a storage medium for supplying the program code, for example, a flexible disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.
[0073]
Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also the OS running on the computer based on the instruction of the program code performs the actual processing. It goes without saying that a case where the functions of the above-described embodiment are realized by performing part or all of the above and the processing thereof is included.
[0074]
【The invention's effect】
  As described above, according to the invention described in claim 1 or 3,, RecordMusical sound waveform data collected by sampling the sound of an instrument at the sound sampling frequencyTone dataStore it in memoryTone dataWhen the musical sound waveform data is read from the memory to generate the musical sound, formant data corresponding to the formant of the frequency range lost by sampling at the recording sampling frequency,Tone data memoryThe formant sound having a formant that approximates the lost formant is synthesized based on the stored formant, and the synthesized formant sound is added to the read musical sound. The formant in the frequency range lost when sampling is recorded at the recording sampling frequency is approximately recovered, so that a more natural musical tone can be generated.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a schematic configuration of a mobile phone to which a musical sound generating device according to an embodiment of the present invention is applied.
FIG. 2 is a flowchart showing a procedure of timbre data memory creation processing;
FIG. 3 is a diagram for explaining a musical sound waveform sample collection process of FIG. 2;
FIG. 4 is a diagram showing a state of each frequency spectrum of a basic musical sound waveform sample, a musical sound waveform sample after downsampling, and a formant after detection;
FIG. 5 is a diagram showing an example of a memory map of a timbre data memory.
6 is a block diagram showing a procedure of a musical tone signal generation process performed by the waveform table sound source of FIG. 1. FIG.
7 is a block diagram showing a procedure of noise formant sound generation processing performed by the noise formant sound generator of FIG. 6; FIG.
FIG. 8 is a diagram illustrating a transition of the level of a noise formant sound.
[Explanation of symbols]
6 waveform table sound source (musical sound waveform sample generation means),25 Tone Data MemoRe28 Noise formant sound generator (synthesizer), 31 Adder (adder), 104 memory (first memory, second memory)

Claims (3)

In Recording sampling frequency, stores the musical sound waveform data collected by sampling sound from the instrument to the tone color data memory, in the tone generating method of generating a musical tone from the tone color data memory reads musical tone waveform data,
The sound of the instrument is sampled and collected at a frequency higher than the recording sampling frequency, and stored in the first memory as first musical sound waveform data,
Storing the data obtained by down-sampling the first musical sound waveform data stored in the first memory at the recording sampling frequency in the second memory as second musical sound waveform data;
Generating formant data from the first musical sound waveform data based on a frequency spectrum in a frequency range lost by down-sampling the first musical sound waveform data at the recording sampling frequency;
Storing the second musical tone waveform data and the generated formant data in association with each other in the timbre data memory;
When generating music,
Generating a musical sound waveform sample from the second musical sound waveform data stored in the timbre data memory;
The tone color is data stored in memory, to synthesize a formant sound on the basis of the formant data associated with the second musical sound waveform data,
Tone generating method characterized by adding and outputting pre Symbol synthesized formant sound on the generated tone waveform samples. In recording sampling frequency, stores the musical sound waveform data collected by sampling sound from the instrument to the tone color data memory, the tone generation device that generates a musical tone by reading the musical tone waveform data from the timbre data memory,
First storage means for sampling and collecting the sound of the instrument at a frequency higher than the recording sampling frequency, and storing the sampled sound as first musical sound waveform data in a first memory;
Second storage means for storing data obtained by down-sampling the first musical sound waveform data stored in the first memory at the recording sampling frequency as second musical sound waveform data in the second memory. When,
Formant data generation means for generating formant data from the first musical sound waveform data based on a frequency spectrum in a frequency range lost by down-sampling the first musical sound waveform data at the recording sampling frequency;
Third storage means for associating and storing the second musical sound waveform data and the formant data generated by the formant data generation means in the timbre data memory;
A musical sound waveform sample generating means for generating a musical sound waveform sample from the second musical sound waveform data stored in the timbre data memory;
Synthesizing means for synthesizing a formant sound the tone color is data stored in memory, based on the formant data associated with the second musical sound waveform data,
Tone generation apparatus characterized by having an adding means for adding and outputting pre Symbol synthesized formant sound on the generated tone waveform samples. The formant data is the center frequency of the formant and its level,
The synthesis means includes
White noise generating means for generating white noise;
Changing means for changing the spectral characteristics of white noise generated by the white noise generating means;
Sine wave generating means for generating a sine wave of the center frequency;
Multiplying means for multiplying the output from the changing means by the sine wave generated by the sine wave generating means;
3. The musical tone generating apparatus according to claim 2 , further comprising an adjusting unit that adjusts the level of the output from the multiplying unit to be the level.

Images