KR100655553B1 - Method of midi synthesizing based on wav table - Google Patents

Abstract

The present invention relates to a wave table-based MIDI (MIDI) synthesis, by compressing and storing the sound source samples used in the wave table-based MIDI synthesis to any sound quality, by decoding and synthesizing the sound source samples during MIDI synthesis The present invention relates to a wave table based MIDI synthesis method for efficiently reducing storage space.

The present invention provides a wave table based MIDI synthesis method comprising: an encoding step of performing encoding according to sound quality on a sound source sample of an existing wave table; A wave table reconstruction step of reconstructing and storing a wave table based on the encoded sound source sample; A MIDI synthesis step of decoding a sound source bit string of the encoded stored wave table and performing MIDI synthesis; Characterized in that comprises a.

MIDI, MIDI Synthesis, Encoding, Decoding

Description

Wave table-based MIDI synthesis {METHOD OF MIDI SYNTHESIZING BASED ON WAV TABLE}

1 is a flowchart illustrating a process of constructing a new encoded wave table by applying a compression method to sound source samples of an existing wave table.

2 is a flowchart of a method for encoding a sound source sample in the present invention.

Figure 3 is a view showing an example of the deviation interval of the continuous sound source sample in the present invention

4 illustrates an example of a bit string structure generated in the present invention.

5 is a flowchart illustrating a process of performing MIDI synthesis using an encoded wave table in the present invention.

6 is a flowchart showing a wave table decoding process encoded in the present invention.

The present invention relates to a wave table-based MIDI (MIDI) synthesis, by compressing and storing the sound source samples used in the wave table-based MIDI synthesis to any sound quality, by decoding and synthesizing the sound source samples during MIDI synthesis Wave table based MIDI synthesis method to reduce storage space efficiently.

Overview of MIDI Synthesis

MIDI (Musical Instrument Digital Interface) is a standard for describing musical information used to synthesize music. In addition to basic musical information such as the type, pitch, and length of the instrument you're trying to sound, MIDI includes information about how the instrument's sound should be represented. In this way, there are largely FM and wave table methods for synthesizing sounds based on musical information expressed in MIDI. The FM method is a method of generating a signal for a frequency when synthesizing sounds by extracting frequency information about sounds that a specific instrument should represent. The wave table method is a method of synthesizing sounds by acquiring and storing samples of actually played notes for each instrument in advance, and processing stored instrument samples according to the musical information described in the MIDI file.

[Middle Synthesis of Wave Table Method]

When synthesizing MIDI using the wave table method, as described above, a sample of a musical instrument actually played is used. This sample is used to describe the musical information described in the MIDI file, such as pitch, pitch, and musical expression. Based on the information, the sound height is modulated, the sound length is changed, and sound expression is applied. In addition, when acquiring and storing a sample of a musical instrument in a wave table, additionally, articulation data such as temporal change in sample volume and fine change in frequency, presence of a loop section, and the start and length of the loop section are described. Information is stored, and when the sound is synthesized, this information is also reflected to generate sound.

[Wave table storage method DLS]

The wave table, which is used for wave table-based MIDI synthesis, has a specific method for maintaining compatibility between various synthesis devices. A well-known method is a down loadable sound (DLS) method. The wave table storage format of the DLS method defines how to store the sample information and the articulation data of the instrument. In particular, the instrument samples are stored in the WAVE format, which is an audio data storage format.

[Loop section]

In the process of synthesizing the MIDI by the wave table method, the sound is reproduced through the sample of the sound actually played for each musical instrument and the additional information of the MIDI file or the wave table. If you store all the samples that are actually played, the size of the wave table will be huge. However, in the wave table, the played note is stable after a certain time and shows a repetitive shape, and it stores only the signal up to the repeated section when the sound source is stored. The repeated part is called a loop section. It will also store the start position and length. When playing an actual MIDI file, this loop section is played repeatedly. Depending on the characteristics of the instrument, there may be a loop applied, or a one-shot ending. This loop interval is applied to reduce the size of the wave table.

[Music source layer by instrument]

Synthesis of the MIDI using the wave table method generates a sound through the sample of the sound actually played for each instrument and additional information of the MIDI file or the wave table as described above. It would be nice to have all the notes in the wave table corresponding to each note in the MIDI file, but since the size is actually larger, it is generally divided into several areas for each instrument to create a note corresponding to the note in the MIDI file. oscillation).

In this way, the wave table method of MIDI synthesis stores the sound samples actually played for each instrument in the form of pulse code modulation (PCM). Therefore, there is a problem in that a very large storage space (storage device) is actually required.

The present invention, in addition to the method of dividing the loop section and the sound source layer for each instrument in order to solve the above problem, there may be a difference in the sound quality of the sound source sample required according to the environment or the device in which the wave table MIDI synthesis method is used. In view of this, it is an object of the present invention to provide a wave table based MIDI synthesis method that can effectively reduce the storage space by the compression method using the audio signal processing technique.

In addition, the present invention is a compression method according to the sound quality of the sound source sample used in the wave table-based MIDI synthesis, implementation difficulties due to the size of the instrument sample stored in the wave table, and the application of sound source samples of sound quality suitable for the environment or device It is an object of the present invention to provide a method of compressing and storing a sound source sample stored in a wave table by applying an audio signal processing technique, and decoding and synthesizing it during MIDI synthesis.

The present invention also provides a method for creating a beat string by compressing an instrument sample stored in the wave table according to an arbitrary sound quality when performing MIDI synthesis using a wave table method, and converting the beat string of the instrument sample into a sound source sample during MIDI synthesis. The goal is to provide a way to change it.

In accordance with another aspect of the present invention, there is provided a wave table based MIDI synthesis method comprising: an encoding step of performing encoding according to sound quality on a sound source sample of an existing wave table; A wave table reconstruction step of reconstructing and storing a wave table based on the encoded sound source sample; A MIDI synthesis step of decoding a sound source bit string of the encoded stored wave table and performing MIDI synthesis; Characterized in that comprises a.

In the wave table-based MIDI synthesis method according to the present invention, the maximum deviation value between successive sound source samples in the encoding step is obtained, and the sound source samples are encoded by applying different compression processes according to the maximum deviation value. do.

In the wave table-based MIDI synthesis method according to the present invention, in the encoding step, the maximum deviation value between successive sound source samples is obtained. When the maximum deviation value is smaller than a predetermined reference value, the maximum deviation value is compressed according to the signal-to-noise ratio. When the maximum deviation value is larger than the predetermined reference value, the sound source sample may be encoded by performing simple quantization.

In addition, in the wave table-based MIDI synthesis method according to the present invention, in the encoding step, it is characterized by minimizing the loss of sound quality by dividing into a plurality of areas according to the distribution of the sample deviation value.

In the wave table-based MIDI synthesis method according to the present invention, in the encoding step, a sample deviation value used for quantization is a difference between a value previously obtained by inverse quantization and a value previously quantized in the input signal. do.

In the wave table-based MIDI synthesis method according to the present invention, in the encoding step, the number of quantization bits closest to an arbitrary threshold signal-to-noise ratio is found through a signal-to-noise ratio between the input signal and the quantized and dequantized signal. It is done.

In the wave table-based MIDI synthesis method according to the present invention, the encoding step, characterized in that the nonlinear quantization of the sound source samples.

In the wave table-based MIDI synthesis method according to the present invention, in the encoding step, a bit string having additional information for expressing characteristics of the encoded sound source sample is generated.

In addition, in the wave table-based MIDI synthesis method according to the present invention, in the encoding step, as an additional information for expressing the characteristics of the encoded sound source sample, information representing an encoding method and a region distribution of deviation values between sound source samples And generating information representing the number of bits for one sound source sample and information representing the number of sound source samples.

Further, in the wave table-based MIDI synthesis method according to the present invention, in the MIDI synthesis step, by analyzing the bit string of the encoded wave table by decoding the corresponding bit string on the basis of additional information representing the characteristics of the encoded sound source sample It is characterized by obtaining a sound source sample.

Further, in the wave table-based MIDI synthesis method according to the present invention, in the MIDI synthesis step, when decoding the sound source bit string of the encoded wave table, a table or a calculation method is selected depending on the amount of operation or the importance of memory according to the device. It is characterized by decoding the sound source bit stream by applying a.

In addition, the wave table based MIDI synthesis method of the present invention for achieving the above object,

A method for synthesizing a wave table based on a MIDI, the method comprising: obtaining a maximum deviation value between consecutive sound source samples with respect to a sound source sample of an existing wave table, and quantizing and compressing the sound source sample if the value is larger than a predetermined value; Calculating a signal-to-noise ratio (SNR) between the input signal and the quantized and dequantized values when the maximum deviation value between the sound source samples is smaller than a predetermined set value; Comparing the signal-to-noise ratio (SNR) with a threshold to find the number of bits closest to any threshold signal-to-noise ratio and quantizing and compressing the sound source samples; Reconstructing a wave table by generating a bit string including additional information for expressing characteristics of the data in the compressed sound source sample; Analyzing a wave table based on the additional information in the reconstructed wave table to perform decoding on a corresponding sound source sample, and to synthesize it by MIDI; Characterized in that comprises a.

An embodiment of the wave table based MIDI synthesis method of the present invention made as described above will be described with reference to the accompanying drawings.

The wave table-based MIDI synthesis method according to the present invention encodes and reconstructs a wave table in consideration of sound quality and size of a sound source, and performs MIDI synthesis by decoding sound source sample information of the reconstructed wave table. An overview of the MIDI synthesis method according to the present invention will be described, followed by a method of encoding a sound source sample, a method of generating a bit string of encoded sound source samples, a method of reconstructing an encoded wave table based on the generated bit string, and the encoding. It will be divided into a method of decoding a sound source sample of a synthesized wave table and performing MIDI synthesis using the same.

[Overview of MIDI Synthesis Method Using Wave Table Encoding and Decoding Considering Sound Quality and Size of Sound Source]

In order to compensate for the shortcomings of using a large storage space of sound source samples in the wave table-type MIDI synthesis, the present invention uses a compression method that maintains an appropriate sound quality according to the environment or device of MIDI synthesis. We devised a way to reduce the size. In a general MIDI synthesis method using a wave table, MIDI is synthesized by applying additional information of a wave table to a sound source of the wave table. As mentioned above, the wave table method of MIDI synthesis has a disadvantage in that the storage space of the sound source sample is large. Thus, the present invention compensates for this by using an audio signal compression method. This method can be divided into first, compressing a musical sample of a wave table to an arbitrary sound quality, and converting a beat sequence of the musical instrument sample into a sound source sample at the time of MIDI synthesis.

1 shows a method of constructing a new encoded wave table by applying a compression technique to sound source samples of an existing wave table.

In step S10, an existing wave table is prepared. Existing wave table includes additional information necessary to synthesize the sound by acquiring and storing the sample of the actually played note for each instrument as described above, and processing the stored instrument sample according to the musical information described in the MIDI file. have.

Next step (S20) is a step of analyzing the existing wave table. This step is to analyze data and additional information for extracting sound source samples from the wave table.

The next step (S30) is to obtain a sound source sample as a result of the wave table analysis. This sound source sample corresponds to a sample of notes actually played for each instrument.

The next step (S40) is the step of encoding the sound source sample. Here, encoding refers to a process of compressing a sound source sample according to sound quality. That is, the sound source sample of the wave table is compressed based on an audio processing technique in consideration of the difference in the required level of sound quality according to the device or the environment.

The next step (S50) is to reconstruct and store the wave table using the encoded sound source samples obtained through the compression process. In other words, in addition to the newly compressed and encoded sound source sample, the necessary information related to the compression encoding is added to generate a bit string, and the wave table is reconstructed using this. The additional information required here includes information indicating what encoding process has been performed, information related to the distribution region of the deviation of the sound source samples, the number of bits for one sample of the sound source bit string, the number of sound source samples, and the like. This will be explained later in detail.

Next step S60 is a step in which a reconstructed, encoded wave table is completed and prepared. The encoded wave table is reconstructed by compressing the sound source samples extracted from the existing wave table and adding the necessary additional information as described above, and decoding the sound source samples of the reconstructed wave table based on the additional information during MIDI synthesis. To be used for MIDI synthesis.

That is, as shown in FIG. 1, in the wave table-based MIDI synthesis method according to the present invention, a sound sample is extracted by analyzing information of an existing wave table, and the sound source sample is compressed through encoding by the method shown in FIG. Then, this encoded sound source sample is formed into a bit string of the form shown in Fig. 4 to form a new encoded wave table. The MIDI wave is finally synthesized using the encoded wave table as shown in FIG. 5, where the sound source sample decoding process S530 of FIG. 5 is performed as shown in FIG. 6.

[Sound sample encoding]

In FIG. 1, the method S40 for encoding the sound source sample separated from the wave table is performed as shown in FIG. 2.

The first step S400 is to separate N-bit sound samples from the existing wave table. This is done by analyzing the wave table as described above and separating the sound source samples based on the results.

Next step (S410) is a step of obtaining a deviation value (difference value) between consecutive samples using the separated N-bit sound source samples. That is, the maximum deviation value is obtained by comparing successive sound source samples. Obtaining the maximum deviation value by comparing successive sound source samples means comparing the successive sound source samples with each other to obtain a difference value (deviation value) between the samples, and using the largest difference value among them as the maximum deviation value. The reason for calculating the deviation value between consecutive sound source samples is to receive the actual N-bit signal as input, and the deviation between successive samples requires about N + 1 bits of storage space for signed signals. This is done to do the exclusion.

Next step S420 is a step of comparing the maximum deviation value (difference value) with a set threshold (here 2 ^N-1 ). This step is to perform different encoding processes according to the maximum deviation value. That is, the maximum deviation is divided into two cases where the maximum deviation is less than and twice the input signal, and when the maximum deviation is less than twice the input signal, the encoding process (S440) is performed according to the signal-to-noise ratio (SNR). Is larger than twice the input signal, it is to simply go through the step (S450) of quantizing and encoding the input sample signal X (n).

The quantization technique used here reduces quantization noise by using a nonlinear quantization technique that expresses the signal in detail when the size is small and performs sparse quantization for the case where the size is large.

First, in the case of simple quantization, that is, when the maximum deviation value (difference value) between successive sound source samples is larger than twice the input signal size, the sound source sample signal X (n) is simply N / 2 bits or 50% through quantization. The compression is performed at a compression rate of (step S451). Subsequently, as shown in the next step S460, the quantized sound source samples are stored in a bit string together with additional information. The additional information will be described later in detail with reference to FIG.

On the other hand, if the maximum deviation value (difference value) between consecutive sound source samples is less than twice, the encoding process (S440) is performed to determine an optimal quantization bit for the sound source samples by applying a threshold of the signal-to-noise ratio (SNR). Performs encoding.

First, in the difference value distribution checking step (S430), the deviation distribution region of the continuous sound source sample is checked. This task is to obtain a better signal-to-noise ratio by reducing the interval of the signal to be quantized depending on where the maximum value of the signal to be quantized is. In fact, as shown in Fig. 3, the deviation of the continuous sound source sample is considered to be divided into three intervals of 1/8, 1/4, and 1/2. Subsequently, the quantization of step S441 and the inverse quantization of step S442 are performed at the same time, which is performed to obtain a signal-to-noise ratio (SNR). In MIDI synthesis using a wave table, the encoding that compresses the sound source of the wave table is not really considered even if the amount of computation is large. This is because a real time implementation is required at the time of decoding.

In the quantization step S441 of FIG. 2, X _DIFF (n) = X (n) -X _REC (n-1), where X (n) is the current input source sample, and X _REC (n-1) is one sample. Each of the sound source samples obtained through the previous inverse quantization is shown. Note that the initialization is done with X _REC (-1) = 0, that is, X _DIFF (0) = X (0). The inverse quantization step S442 of FIG. 2 is a process of inverse quantization of the quantized signal again, which is the reverse order of the quantization process.

In the next step S443, a signal-to-noise ratio is obtained by comparing a sample obtained through quantization and inverse quantization with the original sound source sample as described above. If the signal-to-noise ratio (SNR) is greater than the threshold, the obtained signal-to-noise ratio is compared with the predetermined threshold in the next step (S444), where k is determined as an appropriate bit that satisfies the current threshold and quantized. Is formed into a sound source bit string containing necessary additional information (step S460). If the obtained signal-to-noise ratio (SNR) does not satisfy the threshold condition (S444), the same process is repeated while increasing the bits one by one (S446), applying up to N / 2 (step S445), If it is still not satisfied, the sample of the sound source is quantized to N / 2 bits, and a compression rate of 50% or less is maintained and stored as a bit string including additional information.

[Beat string generation method]

When actually decoding a sound source sample generated through the encoding as shown in FIG. 2, information on how to inverse quantize the quantized sound source data is required. For this purpose, additional information is generated together with the sound source data. do. The structure of the additional information is as shown in FIG. First, information 401 representing what encoding process (S440 or S450) of the encoding process in FIG. 2 has been required. In addition, information 402 representing the distribution area (see S430 and FIG. 3) of the deviation of the sound source sample described above is required. In addition, information 403 representing the number of bits for one sample of the encoded sound source bit string is required, information 404 representing the number of sound source samples, that is, length, and finally quantization data for the corresponding sound source sample. 405 is required. As described above, a bit string is generated by using corresponding sound source quantization data together with additional information related to an encoding process, an area distribution, a number of bits, and a length of a sample.

The bit string thus generated is used for the next wave table reconstruction.

[Encoded Wave Table Configuration (Wave Table Reconstruction)]

The process of storing the encoded bit stream (see FIG. 4) in the wave table after the encoding is performed. Existing wave tables contain the original uncompressed instrument samples, which instead store the strings generated by encoding. In this way, a new wave table in which sound sample data compressed in accordance with the present invention is stored instead of the existing wave table is constructed. As described above, the wave table is composed of encoded sound source data compressed based on an audio processing technique in consideration of sound quality required by an environment or a device, and the reconstructed wave table includes an encoding process as shown in FIG. The sound source quantization data is stored as additional information including the information representing the information, the region distribution information of the sound source sample deviation value (see FIG. 3), and the information representing the number of bits for one sample of the sound source bit string and the number of sound source samples. .

Thus, the wave table reconstructed according to the present invention is usefully used in the MIDI synthesis process to be described later.

[Encoded Wave Table Decoding and MIDI Synthesis]

When synthesizing a MIDI using the encoded wave table, the process is as shown in FIG. The first step S500 refers to the encoded wave table. That is, the step of reading the data of the wave table described with reference to Figures 2 to 4 for MIDI synthesis.

Since the encoded wave table includes encoded bit strings instead of the instrument samples of the existing wave table, in the next step S510, the encoded wave table is analyzed to determine what process (S440, S450) the encoding process has performed. It analyzes what the deviation value distribution is, how many bits are included in one sample of the sound source bit string, and how many (length) of the sound source samples are.

The next step (S520) is a step of obtaining a corresponding bit string through wave table analysis, and the next step (S530) is a method of compressing a bit string with the additional information when the sample encoding is mixed with various compression schemes. Decoded instrument samples are decoded together with the coding parameters stored in addition to the compression scheme because they are described, and the next step (S530) is to synthesize the instrument sounds using the decoded instrument sample data. Step.

The detailed decoding process is as shown in FIG. First, a bit string is obtained from the encoded wave table (step S531). In the next step S532, the obtained bit string is analyzed to determine which method (S440 or S450) has been encoded. Subsequently, it is determined whether to restore the mapping through the table or the calculation method according to whether the calculation amount or the storage space is considered important depending on the device or the environment (step S532).

If the simple quantization process S450 is performed among the encoding processes S440 or S450 shown in FIG. 2, the process proceeds to step S533 to dequantize and decode the corresponding data Y (n), and to decode the signal-to-noise ratio SNR. After the encoding process S440 is performed, the process proceeds to a step S534 where the corresponding data is decoded through the table mapping T (Y (n)).

By decoding as described above, an original sound source sample is obtained (step S535). By using the obtained sound source sample as described above by performing the MIDI synthesis process (S540) it is possible to generate the instrument sound signal.

According to the wave table-based MIDI synthesis method of the present invention, since the sound source samples used for MIDI synthesis are compressed according to sound quality, the wave table is reconstructed and the MIDI synthesis is performed based on this, which significantly reduces the amount of storage space required. have.

Particularly, according to the wave table-based MIDI synthesis method of the present invention, it is possible to apply a sound source sample having a sound quality suitable for an environment or an apparatus due to the difficulty of implementation due to the size of the instrument sample stored in the wave table, and thus requires a large storage space. You'll be able to achieve a high level of MIDI synthesis without having to.

Claims (13)

A wave table based MIDI synthesis method, comprising: an encoding step of encoding sound source samples of an existing wave table according to sound quality of a sound source sample required by a device using a wave table MIDI synthesis method; A wave table reconstruction step of reconstructing and storing a wave table based on the encoded sound source sample; A MIDI synthesis step of decoding a sound source bit string of the encoded stored wave table and performing MIDI synthesis; Wave table based MIDI synthesis method characterized in that comprises a. The method of claim 1, wherein in the encoding step, successive sound source samples are compared to obtain a difference value (difference value) between the samples, and a maximum deviation value, which is the largest difference value among the values, is obtained, and according to the maximum deviation value. Wave table-based MIDI synthesis method characterized in that the encoding of the sound source samples by performing either compression or quantization of the input sample according to the noise ratio (SNR). The method of claim 1, wherein in the encoding step, a difference value (deviation value) between samples is obtained by comparing successive sound source samples, a maximum deviation value that is the largest difference value among the values is obtained, and the maximum deviation value is a reference value. If smaller, compress according to the signal-to-noise ratio, and if the maximum deviation value is larger than the set reference value, input sample quantization is performed to encode a sound source sample.  The wave table-based MIDI synthesis method according to claim 1, wherein in the encoding step, loss of sound quality is minimized by encoding and dividing the data into a plurality of regions according to a distribution of sample deviation values. The method of claim 1, wherein in the encoding step, a sample deviation value used for quantization is a difference from a value obtained through inverse quantization of a value previously quantized in an input signal again. 2. The wave table-based MIDI synthesis method of claim 1, wherein in the encoding step, the number of quantized bits closest to an arbitrary threshold signal-to-noise ratio is found through a signal-to-noise ratio between the input signal and the quantized and dequantized signal. . The method of claim 1, wherein in the encoding step, non-linear quantization of a sound source sample. The wave table-based MIDI synthesis method of claim 1, wherein in the encoding step, a bit string having additional information for expressing a characteristic of the encoded sound source sample is generated. The method according to claim 1, wherein in the encoding step, information representing an encoding method, information representing an encoding method, information representing an area distribution of deviation values between sound source samples, and a single sound source sample are additional information for representing characteristics of the encoded sound source sample. Wave table-based MIDI synthesis method characterized in that for generating information representing the number of bits, information representing the number of sound source samples. The method of claim 1, wherein in the MIDI synthesis step, a sound source sample is obtained by analyzing a bit string of the encoded wave table and decoding the corresponding bit string based on additional information representing characteristics of an encoded sound source sample. Wave table based MIDI synthesis method. The method of claim 1, wherein in the MIDI synthesis step, when decoding the sound source bit stream of the encoded wave table, the bit stream encoded according to the signal-to-noise ratio (SNR) is decoded according to the table mapping. And decoding the sound source bit stream based on inverse quantization of the bit stream encoded through simple quantization. In a wave table-based MIDI synthesis method, comparing continuous sound source samples with respect to sound source samples of an existing wave table, obtaining a difference value (difference value) between samples, and obtaining a maximum deviation value which is the largest difference value among the values. ; Quantizing and compressing a sound source sample when the maximum deviation value is larger than a set value; Calculating a signal-to-noise ratio (SNR) between an input signal and a quantized and dequantized value when the maximum deviation value between the sound source samples is smaller than a set value; Comparing the signal-to-noise ratio (SNR) with a threshold to find the number of bits closest to any threshold signal-to-noise ratio and quantizing and compressing the sound source samples; Reconstructing a wave table by generating a bit string including additional information for expressing characteristics of the data in the compressed sound source sample; Analyzing the wave table based on the additional information in the reconstructed wave table to perform decoding on a corresponding sound source sample, and to synthesize them; Wave table based MIDI synthesis method characterized in that comprises a. The method of claim 12, wherein the additional information for expressing the characteristics of the encoded sound source sample includes information representing an encoding method, information representing an area distribution of deviation values between sound source samples, and number of bits for one sound source sample. And information representing the number of sound source samples.

Images