JPH02146599A - Sound source data compressing and encoding method - Google Patents

Abstract

PURPOSE:To eliminate the discontinuity of looping points by storing words of straight PCM (pulse code modulation) as the specific number of starting words in one or plural starting compressed code blocks. CONSTITUTION:A looping section LP corresponds to specific periods of analog waveform and the specific starting words in starting compressed code blocks BL in the looping section LP are the straight PCM words WST. The number of the straight PCM words WST is larger than the degree of a filter at the time of compressive encoding. Namely, the words of straight PCM are stored as the specific number of starting words in at least one or plural starting compressed code blocks. Consequently, musical sound reproduction which does not generate any noise due to the discontinuance of folding-back points of looping is enabled without increasing the capacity of a storage medium such as a memory.

Description

[Detailed Description of the Invention] [Industrial Field of Application] [Prior Art] In general, sound sources used in electronic musical instruments, TV game machines, etc. are analog sound sources consisting of, for example, VCO, VCA, VCF, etc., and PSG (programmable sound). Digital sound sources are broadly classified into digital sound sources such as the digital sound source (GEW Rake) and the waveform RoM readout type. In recent years, as a type of digital sound source, a sampler sound source has become widely known, in which sound source data obtained by sampling live musical instrument sounds, etc. and digitally processing them is stored in a memory or the like.

This sampler sound source requires a large amount of memory for α to record the sound source data, so various methods have been proposed to save memory.For example, looping using the periodicity of the musical sound waveform Typical examples include processing and bit compression processing using nonlinear quantization and the like.

Here, the looping process is also a method for continuing to output a sound for a longer time than the original duration of the sampled musical tone. In other words, when considering a general musical tone signal waveform, for example, in parts other than the formant part where the periodicity of the waveform is unclear, such as immediately after the start of sound, the fundamental period corresponding to the musical interval (pitch, pitch) is the same. A waveform appears repeatedly, and n cycles of this repeated waveform (n is an integer)
By setting this as a looping section and reproducing it repeatedly as necessary, it is possible to play a long sustained tone with a small memory capacity.

On the other hand, as a system that performs bit compression processing on a normal audio PCM signal, a system using a feed forward type filter on the encoder side is typical of -C. This system sends attached information (parameters related to compression) in addition to compressed data, and the filter on the decoder side is
It has an R (cyclic) digital filter configuration. Such systems have already been adopted in digital optical disk standards and the like.

[Problem to be solved by the invention]

Here, when considering applying the above-mentioned normal audio PCM signal focus compression processing to a musical tone signal having a looping section, a problem arises when connecting the repeating waveform of the above-mentioned looping section.

That is, the values of the looping start point and the looping end point of the looping section are set to values close to each other, but are not exactly the same value. Therefore, in the feed-forward type data compression method, if there is a difference between the values at the looping start point and the looping end point, the IIR filter on the decoder side feeds back and refers to the past output value (restored data), so the loop The discontinuous points in the folding part of IIR cause errors in the subsequent output, and IIR
In other words, since it is a cyclic type, there is a problem in that the discontinuous connections are propagated to later stages.

In order to avoid the above-mentioned problem (error occurrence), it is possible to store the looping section data in a decoded and restored state and use that signal for looping processing, but in this case the buffer for the looping section is Since memory is required, the overall memory capacity increases, which is undesirable.

The present invention has been proposed in consideration of the above-mentioned circumstances, and aims to eliminate discontinuity at the beginning of a predetermined cycle (looping section) of sound source data, especially at the looping point. can be done, and written jQ
It is an object of the present invention to provide a sound source data compression encoding method that can prevent the maximum increase in volume.

[Means to solve the problem]

The present invention has been made to achieve the above-mentioned object, and is a digital waveform corresponding to a predetermined cycle of analog waveforms.
Generates compressed data words and parameters related to the compression of the data in blocks of a predetermined number of samples, and configures and stores one or more compressed code blocks containing the predetermined number of compressed data words and the parameters. A sound source data compression encoding method for storing on a medium, characterized in that straight PCM words are stored in a predetermined number of words at the beginning of at least the first block of the one or more compressed code blocks. There is.

That is, in FIG. 1, the looping section LP corresponds to a predetermined period of the analog waveform, and the first predetermined number of words of the first compression code PRO and BL in this looping section LP are defined as straight PCM words W and T. There is. The number of straight PCM words W and A may be greater than or equal to the order of the filter used in compression encoding.

Note that the expression "at least the first block" also includes, for example, cases in which such processing is performed on all of the seven rounds.

According to the present invention, for example, a predetermined number of words at the beginning of at least the first block of l or a plurality of compressed code blocks obtained by compressing and encoding waveform data for a predetermined period corresponding to a looping section are stored (PCM). By using the above word, the word of the straight PCM can be used as it is as the data of the looping start point when connecting the looping points in the looping process, and there is no need to predict from the data near the looping end point. data.

[Example] Hereinafter, an example to which the present invention is applied will be described with reference to the drawings. It goes without saying that the present invention is not limited to the following examples.

First, FIG. 2 shows an example of an analog musical tone signal waveform before sampling. As is clear from the analog waveform in Figure 2, immediately after the sound of a general musical instrument is produced, the periodicity of the waveform is affected by the inclusion of non-pitch components such as piano keystroke noise and wind instrument press noise. In many cases, a formant portion PR, in which the waveform is unclear, occurs, and thereafter, the same waveform repeatedly appears at a fundamental period corresponding to the interval (pitch, pitch) of the musical tone.

A predetermined n period (n is an integer) of this repetitive waveform is defined as a looping section ALp, and the start point and end point of this section LP are respectively expressed as a looping start point L P s and a looping end point LPi (looping point). And the above Forman l partial PR and looping section L
By recording digital data corresponding to the analog waveform of P on a storage medium and repeatedly reproducing formant part FR followed by looping section LP during reproduction, musical tones can be generated over an arbitrary long period of time.

Here, the musical tone data including such a looping section LP is subjected to focus compression encoding before being stored in a storage medium such as a memory in order to reduce the amount of data, but in this embodiment, the amount of data is reduced. The applicant has previously published JP-A-61-158217 and JP-A-62-00862.
The focus compression encoding method proposed in Publication No. 9 or Japanese Unexamined Patent Publication No. 62-003516, etc., is a method in which each predetermined number of samples (h samples) of peak value data is bronchized and optimal bit compression is performed for each block. A high-efficiency encoding method such as the one shown in FIG.

As shown in FIG. 3, the high-efficiency compression encoding system is composed of an encoder 70 on the recording side and a decoder 90 on the reproduction side. The peak value data x (n) is supplied.

This human input signal (wave height data) χ (b) is the pre-reader 7
2 and an adder 73, and the predicted signal (peak value data) m(n) from the predictor 72 is supplied to the adder 73 as a subtraction signal. It is sent as. The adder 73 subtracts the prediction signal M(n) from the input signal x(n), thereby outputting a prediction error signal or a broadly defined difference output d(n).

Predictor 72 generally uses two past inputs x(n-p),
The predicted value Ma(n) is calculated by 1 frozen sum of x(n-p+1)+...X(n-1). In addition, the above F
The IR filter 74 is hereinafter referred to as an encode filter.

In this bit compression encoding system, the input signal (peak value data) r(n) is converted into a predetermined number of samples (h
The encoding filter 74 having the optimum characteristics is selected for each block. In this method, a plurality of (for example, four) encoding filters having different characteristics are provided in advance, and
This can be achieved by selecting among these filters a filter with optimal characteristics, that is, a filter that can obtain the highest compression rate. However, in the configuration of a general digital filter, a plurality of sets (for example, four sets) of coefficients for the pre-fAIRN12 of one encode filter 74 shown in FIG. 3 are stored in a coefficient memory or the like. By time-divisionally switching and selecting the coefficients of
In many cases, an operation substantially equivalent to selecting one of the plurality of encoding filters described above is performed. As a specific example, a total of four types of encoding filters, zero-order, first-order, and two types of second-order are prepared, and 2-bit filter selection information (mode information) indicating which of these four types of filters is selected is prepared. selection information) may be transmitted for each block. Note that the 0th order encoding filter is
This corresponds to selecting the straight PCM mode in which the input PCM signal is output as is.

Next, the difference output d (n) as the prediction error is sent to a shifter 75 with a gain of G and a quantizer 76 via an adder 2H81.
and a bit compressor, such as floating point (
A compression process or a ranging process is performed such that the exponent part in the display form (floating point) corresponds to the gain G and the mantissa part corresponds to the output from the quantizer 76, respectively. That is, the shifter 75 shifts, for example, 16-bit input data by the number of focuses corresponding to the gain G to switch the range, and the quantizer 76 shifts the focus-shifted data to a certain number of bits, so for example, 4 bits are shifted. Requantization is performed to extract the data. Here, noise shaping circuit (noise shaper) 7
7, the adder 78 obtains a so-called quantization error corresponding to the error between the output and the input of the quantizer 76, and this quantization error is added to the gain Q.
-1 shifter 79 to the predictor 80, and a so-called error feedbank is performed in which the predicted signal of the quantization error is fed back to the adder 81 as a subtraction signal. In this way, requantization by the quantizer 76 and error feedback by the noise shaping circuit 77 are performed,
Output ↑(n) is taken out from the output terminal 82.

By the way, the output d''(n) from the adder 81 is obtained by subtracting the predicted signal 7(n) of the quantization error from the noise shaper 77 from the difference output d(n), and is equal to the gain C. The output d'' (n) from the shifter 75 is the gain G and the output d' (n) from the output adder 81.
is multiplied by Further, the output 1(n) from the quantizer 16 is the quantization error e in the quantization process.
(n) and the output dl'(n) from the shifter 75, and the quantization error e(n) is extracted in the adder 78 of the noise shaper 77. This quantization error e (n) is the gain G−1
The predicted signal τ(n) of the quantization error is obtained by passing through the shifter 79 and the predictor 80 which takes the 1 frozen sum of r past inputs.

The sound source data is subjected to the encoding process as described above, and the output from the quantizer 76 is output from the quantizer 76. (n) and is taken out via the output terminal 82.

Next, the prediction/range adaptation circuit l\884 outputs mode selection information as optimal filter selection information and sends it to the predictor 72 and output terminal 87 of the encoding filter 74, and also outputs the gain G and gain G-
1 or range information for determining the focus shift amount is output and sent to each shifter 75, 79 and output terminal 86. These mode (filter) selection information and range information are also referred to as compression-related parameters.

Here, FIG. 4 shows an example of one bronc of such parameter information and the focus-compressed (encoded) sample data (the output cuff (n)). The data for 1 inch in FIG. 4 includes 1 byte of header information (parameter information) RF
and 8-byte sample data DAO to DM3. The header information RF includes the range information of 4 pints, the filter selection information of 2 bits, and two flag information of 1 bit each, for example, information indicating that the block includes the start point of a loop (loop start point).・Flag LSF) and information indicating the block at the end point of the loop (loop end flag LEF). Here, the peak value data of one sample is bit compressed and expressed in 4 points, and the above data DAO ~
D113 contains 16 samples of 4-bit data D,
. Contains 8-D szt.

Next, the output ↑(n) from the output terminal 82 of the encoder 70 is transmitted to the input terminal 91 of the decoder 90 on the playback side, or the signal ";(' ( This input signal ↑゛(n) is supplied to the adder 9 via a shifter 92 with a gain of G-1.
It has been sent to 3. Output x' (n
) is sent to the predictor 94 and the predicted signal ';i' (n)
Then, this predicted signal q' (n) is sent to the adder 93
The output from the shifter 92 +l (n)
is added. This addition output is the decode output ↑°(n
) from the output terminal 95.

Further, each output terminal 86 and 87 of the encoder 70
The range information and mode selection signal output, transmitted, recorded, and reproduced from the decoder 90 are manually inputted to input terminals 96 and 97 of the decoder 90, respectively. The range information from the input terminal 96 is sent to the shifter 92.
The mode selection information from the input terminal 97 is sent to the predictor 94 to determine the prediction characteristic. The prediction characteristic of this predictor 94 is selected to be equal to the characteristic of the predictor 72 of the encoder 70.

In this decoder 90, the output ↑"(n) from the shifter 92 is the product of the input signal ↑"(n) and the gain G-". Also, the output ↑"(n) from the adder 93 is the product of the input signal ↑"(n) and the gain G-". ) is the sum of the output from the shifter 92?'' (n) and the prediction signal x' (n).

In a bit compression encoding system with such a configuration,
Looping start point L of looping section LP in Figure 2 above
As shown in FIG. 1, the first predetermined number of words of the block starting from P5 (looping start block) are words WSt of straight PCM data. This predetermined number may be equal to or greater than the maximum value of the order of the encode filter 74 (order of the prediction filter), and in the above specific example, the maximum order of the encode filter 74 is second order. , 2 or more straight PCs
Words W and T of M data may be arranged.

For this purpose, for example, a changeover switch may be provided just before the encode output terminal 82 so that the compressed data from the quantizer 76 and the straight PCM data can be selectively output. For the predetermined number of words, the selector switch may be used to select and output straight PCM data. Note that the range ↑ information is independent in each block, so instead of outputting straight PCM data of the word length of the original sample, for example 16 bits, as is, it is compressed by the bits corresponding to the range information of the same block, for example 4 bits. Straight PCM data may be output.

Next, FIG. 5 is a block circuit diagram showing a specific example when the decoder 90 of the bit compression encoding system described above is used in an embodiment of the sound source data compression encoding method according to the present invention.

In FIG. 5, the memory 1 stores the compression-encoded sample data, variation information (header information rIIRF, attached information in FIG. 4), etc. (sound source data), and also stores the looping start address. (the in-memory address of the looping start block) etc. are also stored as directory information, for example. At least this looping start address is sent to the address register 2 via the data bus of the memory 1, where it is temporarily stored, and is preset into the address counter 3 immediately before the end of looping. Decoder 4 outputs the output from address counter 3 and the loop start flag L
According to the SF, a straight PCM switching control signal is output during the first predetermined number of words in the looping start block (while straight PCM data is output). Note that when multiple types of sound source data are stored in the memory 1, the start address and looping start address of each sound source data are stored in the memory 1 as the above directory information, and in response to key-on etc. The start address of the sound source data corresponding to the sound source selected in advance is read from the directory of the memory 1, and is presented to the address counter 3 via the address register 2, so that the data is sequentially read from the start address. Reading is performed. When there is only one type of sound source data, the address counter 3 may start counting from a predetermined initial value (the first address of the sound source data) in response to key-on or the like.

Of the sound source data read out from memory l in this way,
The above parameter information (header information RF and attached information in FIG. 4) is sent to the attached information register 5 and temporarily stored.
The 4-bit compressed data (
The sample data DA0□ to DI13L) in FIG. 4 are sent to the data register 6 and temporarily stored. Also,
Stray 1-1'CM of a predetermined number of words at the beginning of the looping start block in the sound source data from the memo 121
The data is sent to multiplexer 8.

The compressed data temporarily stored in the data register 6 is sent to the bit shift circuit 7 of the decoder 20 corresponding to the shifter 92 of the decoder 90 in FIG. The signal is then sent to adder 10 corresponding to adder 93 in FIG. The output from the adder IO is sent to the output register 16 and also to the predictor 21 corresponding to the predictor 94 in FIG. 3, and the output from this predictor 21 is fed to the adder IO. . The predictor 21 includes two delay registers 14.15, a coefficient multiplier 12.13 that multiplies the outputs from these delay registers 14.15 by coefficients, and the outputs from these coefficient multipliers 12.13. Adder 1 to add
1, and each multiplication coefficient of the coefficient multipliers 12 and 13 is determined by the coefficient generation circuit 9. The predictor 21, addition 3I2i10, and bit shift circuit 7 constitute a decoder 20 that encodes the bit compression encoded data into 1y code.

The above header information F temporarily stored in the attached information register 5
Of R, range information is sent to the focus shift circuit 7, filter selection information is sent to the coefficient generation circuit 9, and the loop
The end flag LEF is connected to address counter 3 (preset control terminal), and the loop start flag LSF is
are sent to the decoder 4, respectively.

In the decoder 20 shown in FIG. 5, the 4-pint compressed data is subjected to the same decoding process as the decoder 70 shown in FIG. D via 16
The signal is sent to the /A converter 17 and taken out from the output terminal 18 as an analog tone signal.

Here, when performing looping playback, the operation when returning from the looping end point to the looping start point will be described. When reading the sound source data of the block including the looping end point, the loop end flag LEF of the attached information is set, and based on this loop end flag LEF, when the looping start point is reached, the address register is set. The looping start address from 2 is preset in address counter 3. Therefore, the address counter 3 starts accessing the looping start block of the memory 1 via the address bus, and the first predetermined number of words read out the straight PCM data.

While reading this stray (stray) PCM data, the decoder 4 outputs the straight PCM switching control signal and sends it to the multiplexer 8 and coefficient generation circuit 9, and controls to send this straight PCM data as it is to the output register 16. It will be done. That is, multiplexer 8
selects and outputs the above-mentioned straight PCM data from memory l, and the coefficient generation circuit 9 sends multiplication coefficients constituting a zero-order filter to the coefficient multipliers 12 and 13, so that the output from the adder 10 The above straight PCM data can be obtained as is.

Therefore, since the above Stray 1-PCM data is used as is as data at the looping start point, there is no need to make predictions from the data at the looping end point, and errors caused by discontinuities at the looping point can be effectively avoided. It can be prevented.

In this case, the sound source data includes the looping starting point LP, the straight P of the part, and
Since the compression ratio as a whole remains almost the same even though only the CM data increases, an increase in memory capacity can be suppressed.

By the way, the above-mentioned bit compression encoding process and other digital signal processing for generating sound source data are often implemented in software using a digital signal processing device (DSP). A software-like configuration using a DSP is often adopted for the reproduction of sound source data. As an example, FIG. 6 shows an example of the overall configuration of a system including an audio processing unit (APU) 107 as a sound source unit that receives sound source data and its surroundings.

In FIG. 6, a host computer 104 installed in, for example, a general personal computer device, a digital electronic musical instrument, a TV game machine, etc. is connected to the APUIO 7 as the sound source unit, and the host computer 104 is connected to the APUIO 7 as the sound source unit. Sound source data etc. are APU10
7. This APU107
is the CPU (central processing unit) of MicroProsensor, etc.
103, a digital signal processing device (DSP) 101, and a memory 1 in which sound source data and the like as described above are recorded.
02. In other words, this memory 102 corresponds to the memory 1 in FIG. 5 above, and stores the compressed data and attached information (parameters related to compression), as well as the start address and looping start address of each sound source of multiple types in. has been done. DSPLol executes various processes, such as looping processing, the above-mentioned decoding (bit expansion) processing, pitch conversion processing, envelope addition, echo (reverberation) processing, etc., while controlling the reading of this memory 102. The memory 102 is also used as a buffer memory for these various processes.

The CPU 103 controls the operations and contents of these various processes of the DSP 101.

Furthermore, D is applied to the sound source data from the memory 102.
The digital musical tone data finally obtained by performing the above various processing using SPIOI is converted into digital/analog (D
/A) The signal is converted into an analog signal by the converter 105 and supplied to the speaker 106.

It should be noted that the present invention is not limited to the above-described embodiments; for example, not only the first block of the looping section but also the first predetermined number of words ' of the next block and subsequent blocks are converted into words of straight PCM data. In this case, it is possible to prevent errors from occurring when an arbitrary block is extracted and reproduced. In addition, the number of bits of the straight PCM data may be the same as bit h of the original sampling peak value data, for example, the same number of bits as the compressed data, and the straight PCM data compressed using the range information of the block at this time may be used. You can also do this. Further, in the above-described embodiment, the formant portion and the looping section are connected to form sound source data, but the present invention can easily be applied to the case where sound source data consisting only of the looping section is formed. Further, the decoder side configuration and the external memory for sound source data may be supplied as a ROM cartridge or an adapter. Moreover, it is applicable not only to a sound source of musical tone signals but also to other uses such as voice synthesis.

[Effects of the Invention] According to the sound source data compression encoding method of the present invention, l or a plurality of compressed code blocks obtained by compressing and encoding digital data corresponding to a predetermined cycle of analog waveforms corresponding to, for example, a looping section. At least a predetermined number of words at the beginning of the first block are straight PCM words, so even if the looping points are discontinuous during looping processing, the looping start point is not influenced by past output values. I don't accept it. Therefore, without increasing the capacity of a storage medium such as a memory, it is possible to prevent errors due to discontinuity of looping return points, and it is possible to reproduce music without generating looping noise.

FIG. 6 is a block circuit diagram for explaining a specific example of the apparatus. FIG. 6 is a block diagram showing an example of the configuration of a system including an audio processing unit and its surroundings.

Claims (1)

[Claims] Digital data corresponding to an analog waveform for a predetermined period is generated into compressed data words and parameters related to the compression in block units for each predetermined number of samples, and a predetermined number of compressed data words are generated. A sound source data compression encoding method comprising configuring one or more compressed code blocks including parameters thereof and storing the same in a storage medium, the method comprising: forming one or more compressed code blocks including parameters thereof, and storing the same in a storage medium; A given number of words in straight P
A method for compressing and encoding sound source data, characterized in that words of commercials are stored.