JP2913695B2 - Digital signal encoding method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a digital signal encoding method for encoding an input digital signal.

[Summary of the Invention]

The present invention divides an input digital signal into a plurality of frequency bands, selects a wider bandwidth for a higher frequency band, sets an allowable noise level for each band based on energy for each band, and In a digital signal encoding method for quantizing the components of each band with the number of bits according to the level of the difference between the energy and the set allowable noise level, the amount of output information after quantization is detected, and the detected output and the target value The number of bits allocated at the time of quantization is corrected based on the error of each band and the energy of each band, so that the amount of information in a predetermined period is made constant. Bit rate adjustment (bit packing) can be performed. In particular, a large number of bits can be allocated to a frequency band where energy is concentrated. , There is provided a digital signal encoding method which can improve the S / N of the hearing sense.

[Conventional technology]

In high-efficiency coding of signals such as audio and voice, a bit allocator that divides an input signal such as audio and voice into a plurality of channels along a time axis or a frequency axis and adaptively allocates the number of bits for each channel. There is an encoding technique based on a shion (bit allocation). For example, coding techniques based on the above-mentioned bit allocation of audio signals and the like include band division coding (sub-band coding: SBC) in which an audio signal or the like on the time axis is divided into a plurality of frequency bands and encoded. A so-called adaptive transform coding (AT) that converts a signal on the time axis into a signal on the frequency axis (orthogonal transform), divides the signal into a plurality of frequency bands, and adaptively codes each band.
C) or the above-mentioned SBC and so-called adaptive prediction coding (AP
C), a so-called adaptive bit allocation (APC-AB) in which a signal on the time axis is band-divided, each band signal is converted into a baseband (low band), and then multi-order linear prediction analysis is performed to perform predictive coding. ).

Here, for example, in the above-mentioned band division coding, in order to increase the compression efficiency, while keeping the bit rate per unit time block constant, the number of bits to be given to each band divided band is represented by the time variation of the signal spectrum intensity. Is dynamically (adaptively) changed according to the Also,
In the above adaptive transform coding, the number of allocated bits is dynamically changed on the frequency axis.

[Problems to be solved by the invention]

In the high-efficiency coding that adaptively allocates the number of bits as described above, a unit block (a unit time block or a unit frequency block) depends on how the number of bits is allocated.
The bit rate per hit may not be constant, which may result in excess or deficiency of bits. That is, for a given bit rate, a phenomenon may occur in which a bit is excessive at a place where the signal level is low on the time axis or the frequency axis, and a bit is insufficient at a place where the signal level is high.

When there is such an excess or deficiency in the number of bits, it is necessary to use a bit rate adjustment (so-called bit packing) technique for effectively adjusting the number of bits.

This bit rate adjustment is to reduce the number of bits given to this unit block when bits remain in the unit block, and to increase the number of bits given when there are not enough bits to keep the overall bit rate constant. It is to adjust. The bit rate adjustment is
For example, it is conceivable that this can be performed using functional blocks as shown in FIG. In FIG. 8, for the input digital signal, the number of allocated bits at the time of quantization is determined by an allocated bit number determining function block 90, and the spectrum intensity of a unit block of the input digital signal is determined by a bit rate adjusting function block 91. Of the bit rate (bit packing) according to. After that, the input digital signal is encoded (re-quantized) by the encoding function block 92 using the number of bits determined by adjusting the bit rate, and is output.

However, as the bit rate adjustment performed by the bit rate adjustment function block 92, there is still no effective method that is simple and signal deterioration is not conspicuous. The establishment of a method is desired. In particular, when adjusting the bit rate, a tone source (sine wave, tone burst wave, single instrument sound, etc.) in which signals (spectral intensity and energy) are concentrated in a certain band is generated in that band. Noise that occurs is often felt as deterioration, and there is a need for a countermeasure to prevent this deterioration.

Therefore, the present invention has been proposed in view of the above-described circumstances, and it is possible to perform bit rate adjustment (bit packing) in which signal deterioration is not noticeable with a simple configuration. It is an object of the present invention to provide a digital signal encoding method capable of performing signal encoding with little deterioration even with a certain source.

[Means for Solving the Problems] A digital signal encoding method according to the present invention has been proposed to solve the above-described problems, and divides an input digital signal into a plurality of frequency bands and has a high frequency band. A noise level setting step of selecting a wider bandwidth for each band and setting an allowable noise level for each band based on the energy of each band; and a difference between the energy of each band and the set allowable noise level. And a quantization step of quantizing the components of each band with the number of bits according to the level of the digital signal, wherein the amount of output information obtained by the quantization step is detected, and the detected output and An error from a target value is detected, and based on the detected error and the energy of each band, the number of bits allocated in the quantization step is converted into a tone signal. The present invention is characterized in that a correction is made so that a large number of bits are assigned to a band where energy is concentrated in the band, and the information amount in a predetermined period is made constant.

[Operation] According to the present invention, the overall bit rate is kept constant,
At the same time, the number of allocated bits is increased or decreased by a certain amount based on the energy of each band, so that a larger number of bits is allocated to a band where energy is concentrated.

Embodiments Hereinafter, embodiments to which the present invention is applied will be described with reference to the drawings.

The digital signal encoding apparatus to which the digital signal encoding method of the present embodiment is applied converts an input digital signal such as audio or voice into, for example, band division encoding (SB
C), adaptive transform coding (ATC), adaptive bit allocation (AP
C-AB) or the like to perform high-efficiency coding. Therefore, in the present embodiment, the input digital signal is divided into a plurality of frequency bands, and the higher the frequency band, the wider the bandwidth is selected. That is, the input digital signal is divided by a so-called critical bandwidth (critical band) in consideration of human auditory characteristics described later. Also,
As shown in FIG. 1, a total sum detection circuit 14 and a filter as noise level setting means for setting an allowable noise level for each band based on the energy (or peak value, average value) of each band of the critical band. Circuit 15
And a quantization circuit 24 that quantizes the components of each band with the number of bits assigned according to the energy of each band and the level of the difference between the noise level setting means. Here, the output information amount of the quantization circuit 24 is detected by a data amount calculation circuit 26 described later, and the detected output and the terminal 3
An error with a predetermined target value (a target value of a bit rate for adjusting a bit rate) from the data is detected by an error detection circuit 27 described later. Thereafter, a correction value for correcting the number of bits allocated to the quantization circuit 24 is determined by a correction value determination circuit 28 described later based on the error output and the energy of each band, and quantization is performed based on the correction value. Have been done. In this way, in the present embodiment, the amount of information in a predetermined period is made constant (adjustment of the bit rate).

Thereafter, the quantized output from the quantizing circuit 24 is output from the output terminal 2 of the digital signal encoding device of the present embodiment via the buffer memory 25.

Therefore, when the signal supplied to the apparatus shown in FIG. 1 is, for example, a signal having a signal spectrum SS as shown in FIG. 2A, that is, a signal (spectral intensity,
In the case of a tonal source (sine wave, tone burst wave, single instrument sound, etc.) in which energy is concentrated, the noise spectrum NS is represented by a dashed line in FIG. It becomes something. Therefore,
In the apparatus of this embodiment shown in FIG. 1, for example, the frequency of a signal is analyzed to determine the ratio of energy for each band. A larger number of bits is given to a band having a large energy in proportion to, and a bit rate is adjusted with a small bit allocation to a small (small) band. Thus, as shown in FIG. 2 B, the noise spectrum NS is corrected to noise spectrum NS _B shown in dotted line in the figure in Figure 2 B. Conversely, when there are insufficient bits, bits are picked up from a band having a small energy in inverse proportion to the signal energy, and given to a band where the number of bits is insufficient. Thereby, as shown in FIG. 2C, the noise spectrum NS
Is corrected as noise spectrum NS _C shown in the figure in FIG. 2 C dotted line.

Here, the digital signal encoding apparatus of the present embodiment shown in FIG. 1 performs a fast Fourier transform (FFT) on an audio signal, a voice signal, and the like, and converts a time-axis signal into a frequency axis.
The encoding (requantization) is performed.

In FIG. 1, for example, an audio signal is supplied to an input terminal 1, and the audio signal on the time axis is transmitted to a fast Fourier transform circuit 11. In the fast Fourier transform circuit 11, the audio signal on the time axis is converted into a signal on the frequency axis at predetermined time intervals (unit blocks), and an FFT coefficient including a real component value Re and an imaginary component value Im is obtained. . These FFT coefficients are stored in the amplitude / phase information generation circuit 12
The amplitude and phase information generation circuit 12 obtains the amplitude value Am and the phase value from the real component value Re and the imaginary component value Im, and outputs information on the amplitude value Am.
That is, general human hearing is sensitive to the amplitude (power) in the frequency domain, but relatively insensitive to the phase. Therefore, in this embodiment, only the amplitude value Am is obtained from the output of the amplitude / phase information generation circuit 12. This is taken out as an input digital signal in the embodiment of the present invention.

The input digital signal such as the amplitude value Am obtained in this way is transmitted to the band dividing circuit 13. The band division circuit 13 divides the input digital signal represented by the amplitude value Am into a so-called critical bandwidth (critical band). This critical band takes into account human auditory characteristics (frequency analysis capability), and is, for example, 0
1616 kHz is divided into 24 bands, and the higher the frequency band, the wider the bandwidth is selected. That is, human hearing has characteristics like a kind of band-pass filter, and the band divided by each filter is called a critical band. Here, FIG. 3 shows the critical band. However, in FIG. 3, the number of the critical bands is represented by 12 bands (B _{1 to} B ₁₂ ) to simplify the illustration.

The amplitude value Am for each band (for example, 24 bands) divided into critical bands by the band division circuit 13 is transmitted to the sum detection circuit 14, respectively. This sum detection circuit
In 14, the energy for each band (spectral intensity in each band) is obtained by taking the sum of the respective amplitude values Am in each band (peak or average of the amplitude values Am or total energy). The output of the sum detection circuit 14, that is, the spectrum of the sum of each band is generally called a bark spectrum, and the bark spectrum SB of each band is as shown in FIG. 4, for example.

Here, in order to consider the influence of the bark spectrum SB on so-called masking, the bark spectrum SB
Is convolved with a function of a predetermined weight (convolution). Therefore, the output of the sum detection circuit 14, that is, each value of the bark spectrum SB is sent to the filter circuit 15. The filter circuit 15, as shown in FIG. 5, the delay elements for sequentially delaying input data from the input terminal _{^{100 (z -1) 101 1,}} 101 2 ‥‥ 101 m-2 ~101 m + 3 ‥‥ 101 ₂₃ , 101 ₂₄
(For example, 24 delay elements corresponding to the critical band) and, for example, 24 multipliers 102 ₁ , 102 ₂乗 算 for multiplying the outputs from these delay elements 101 _{1 to} 101 ₂₄ by filter coefficients (weighting functions). ‥ 102 _{m-3 to} 102 _{m + 3} ‥‥ 102 ₂₃ , 102 ₂₄ ,
And a sum adder 104. At this time,
In the multipliers 102 _{m−3 to} 102 _{m + 3} , for example, the multiplier 10
The filter coefficient 0.0000086 at 2 _m-3, the filter coefficient 0.0019 at multiplier 102 _m-2, filter coefficient 0.15 at multiplier 102 _m-1
The multiplier 102 the filter coefficient 1 at _m, a multiplier 102 the filter coefficients 0.4 _{m + 1,} the filter coefficients in addition multiplier 102 _{m + 2} 0.0
By multiplying by 6 and multiplying the output of each delay element by the filter coefficient 0.007 by the multiplier 102 _{m + 3} , the convolution processing of the Bark spectrum SB is performed. By the convolution process, the fourth
The sum of the parts indicated by the dotted lines in the figure is calculated. The masking refers to a phenomenon that a certain signal masks another signal and makes it inaudible due to human auditory characteristics.This masking effect includes a masking effect for an audio signal on a time axis. There is a masking effect on signals on the frequency axis. That is, even if there is noise in the masked portion due to the masking effect, this noise will not be heard. For this reason, in an actual audio signal, the noise in the masked portion is regarded as acceptable noise.

Thereafter, the output of the filter circuit 15 is sent to the subtractor 16. The subtracter 16 calculates a level α corresponding to an allowable noise level described later in the convolved area. The level α corresponding to the permissible noise level (permissible noise level) is, as described later, a level at which the permissible noise level of each critical band is obtained by performing inverse convolution processing. is there. Here, an allowance function (a function expressing a masking level) for obtaining the level α is supplied to the subtractor 16. The level α is controlled by increasing or decreasing the allowable function. The permissible function is supplied from a function generation circuit 29 described later.

That is, the level α corresponding to the allowable noise level is
Assuming that the number given in order from the low band of the critical band is i, the critical band can be obtained by Expression (1).

α = S− (n−ai) ‥‥‥‥ (1) In the equation (1), n and a are constants and a> 0, and S is the intensity of the convolution-processed bark spectrum. 1) (n-ai) in the equation is an allowable function. In the present embodiment, n = 38, a = 1, and there was no deterioration in sound quality at this time, and good encoding was performed.

Thus, the level α is obtained, and this data is transmitted to the divider 17. The divider 17 is for inversely convolving the level α in the convolved region. Therefore, by performing the inverse convolution processing, a masking spectrum can be obtained from the level α. That is, this masking spectrum becomes an allowable noise spectrum. Although the above inverse convolution process requires a complicated operation, in this embodiment, a simplified divider is used.
17 is used for inverse convolution.

Next, the masking spectrum is transmitted to the subtractor 19 via the synthesis circuit 18. Here, the output of the sum detection circuit 14, that is, the bark spectrum SB from the sum detection circuit 14 described above is supplied to the subtractor 19 via the delay circuit 21. Therefore, by performing the subtraction operation of the masking spectrum and the bark spectrum SB in the subtracter 19, as shown in FIG. 6, the burspectrum SB has a level lower than the level indicated by the level of the masking spectrum MS. It will be masked.

The output of the subtracter 19 is supplied to the quantization circuit 24 via the ROM 20.
In the quantization circuit 24, the subtracter
The quantization of the amplitude value Am supplied via the delay circuit 23 is performed with the number of bits at a rate corresponding to the output of 19.
In other words, in other words, the quantization circuit 24 quantizes the components of each band with the number of bits allocated according to the energy of each band of the critical band and the level of the difference between the noise level setting means. become. In addition, the delay circuit 21 delays the bark spectrum SB from the sum detection circuit 14 in consideration of the delay amount in each circuit before the synthesis circuit 18, and the delay circuit 23 performs the delay in each circuit before the ROM 20. It is provided to delay the amplitude value Am in consideration of the delay amount. Further, the ROM 20 is provided for temporarily storing and sending out the output of the subtractor 19 every predetermined time at the time of quantization.

Here, at the time of the quantization of the quantization circuit 24, as described above, the output information amount of the quantization circuit 24 is detected, and based on the error between the detected output and the target value and the energy of each band. The amount of allocated bits is corrected at the time of quantization, so that the amount of information in a predetermined period is made constant.

To do this, in the present embodiment, the data from the buffer memory 25 is sent to the error detection circuit 27 after the data amount is calculated by the data amount calculation circuit 26. The error detection circuit 27 detects an error between the data amount and a predetermined target value (a target value of the bit rate for adjusting the bit rate) from the terminal 3, and transmits the error data to the correction value determination circuit 28. Is done. The output (energy and spectrum intensity) of the sum detection circuit 14 is also supplied to the correction value determination circuit 28 via a memory 33 for storing data of a unit block. The correction value determination circuit 28 gives a larger correction value to a band with a larger energy if the error data is positive, and gives a smaller correction value to a band with a larger energy if the error data is negative. Is what you do. Such a correction value is transmitted to the bit number correction circuit 31. This bit number correction circuit
In step 31, by correcting the output of the ROM 20 with the correction value, correction of the number of allocated bits at the time of quantization (adjustment of the bit rate) is performed. Becomes constant.
For this reason, quantization is performed with a large number of bits in a high energy band, and quantization is performed with a small number of bits in a low energy band. Here, since the output of the bit number correction circuit 31 is supplied to the quantization circuit 24 via the rounding circuit 32, a fine change in the bit number is corrected.

In the case of the synthesizing by the synthesizing circuit 18, the so-called minimum audible curve (equal loudness curve) RC which is a human auditory characteristic as shown in FIG. The data and the masking spectrum MS can be synthesized. Therefore, by synthesizing the minimum audible curve RC and the masking spectrum MS together, the allowable noise level can be reduced to the portion indicated by the oblique line in the figure, and the portion indicated by the chain line in the diagram at the time of quantization. Can be reduced. FIG. 7 is represented by the critical band shown in FIG.
SS is also shown.

However, in the present embodiment, a configuration may be adopted in which the above-described minimum audible curve synthesis processing is not performed. That is, in this case, the minimum audible curve generating circuit 22 and the synthesizing circuit 18 become unnecessary, and the output from the subtracter 16 is deconvolved by the divider 17 and immediately thereafter,
Will be transmitted.

That is, in the digital signal encoding device of the present embodiment, when encoding an audio signal, the bit rate is adjusted as described above, and since the shape of the noise spectrum does not change, even when the number of bits is reduced, The hearing deterioration is small. In particular, it is possible to perform signal encoding with little deterioration even for a source having a tone characteristic. Also, since the algorithm for adjusting the bit rate is easy, hardware can be easily configured.

In the present invention, in addition to the so-called adaptive transform coding for processing an input digital signal by performing a fast Fourier transform as in the embodiment of FIG.
The present invention can be applied to an apparatus that performs C). In this case, the signal is band-divided by a band-pass filter or the like, and the number of bits assigned to each channel is increased or decreased based on the error between the detection output of the output information amount of the quantization means and the target value and the energy of each band. It will be. In the case of the band division coding, the same effect as described above can be obtained.

〔The invention's effect〕

In the digital signal encoding method of the present invention, the amount of output information after quantization is detected, an error between the detected output and a target value is detected, and based on the detected error and the energy of each band, By correcting the number of bits allocated in the quantization step so as to allocate a large number of bits to a band where energy is concentrated in the tone signal, so as to stabilize the information amount in a predetermined period, With a device having a simple configuration, it is possible to perform bit rate adjustment (bit packing) with no noticeable signal deterioration. In particular, it is possible to perform signal encoding with little deterioration even for a source having a tone characteristic. Therefore,
S / N deterioration on hearing can be reduced.

[Brief description of the drawings]

FIG. 1 is a block circuit diagram showing a schematic configuration of a digital signal encoding apparatus to which a digital signal encoding method according to an embodiment of the present invention is applied, and FIG. 2A shows a noise spectrum before bit rate adjustment. FIG. 2B is a characteristic diagram showing a noise spectrum after bit rate adjustment (when the number of bits is too small), and FIG. 2C is a characteristic diagram showing a noise spectrum after bit rate adjustment (when the number of bits is insufficient). 3 is a diagram showing a critical band, FIG. 4 is a diagram showing a bark spectrum, FIG. 5 is a circuit diagram showing a filter circuit, FIG. 6 is a diagram showing a masking spectrum, FIG. Figure that combines masking spectrum,
FIG. 8 is a functional block diagram for adjusting a bit rate. 11 Fast Fourier transform circuit 12 Amplitude / phase information generation circuit 13 Band division circuit 14 Sum detection circuit 15 Filter circuit 16 Divider 17 Divider 18 Synthesis circuit 19 … Subtractor 20… ROM 21,23… Delay circuit 22… Minimum audible curve generation circuit 24… Quantization circuit 25… Buffer memory 26… Data amount calculation circuit 27… Error detection circuit 28… Correction value determination circuit 29 Function generation circuit 31 Bit number correction circuit 32 Rounding circuit 33 Memory

Continuation of front page (56) References JP-A-3-117921 (JP, A) JP-A-60-96041 (JP, A) JP-A-62-57621 (JP, A) JP-A-61-110200 (JP, A) , A) Proceedings. IEEE International Conferencing on Acoustics, Speech and Signal Processing (1988) Vol. 5, p. 2524-2527 Proceedings. IEEE International Conferencing on Acoustics, Speech and Signal Processing (1989) Vol. 3, p. 1993-1996 Proceedings of the National Meeting of the Institute of Electronics, Communication and Communication Engineers, 1984-1984 [Part 1], pp. 1-389-390

Claims (1)

(57) [Claims] 1. A noise level for dividing an input digital signal into a plurality of frequency bands, selecting a wider bandwidth for a higher frequency band, and setting an allowable noise level for each band based on the energy of each band. A digital signal encoding method, comprising: a setting step; and a quantization step of quantizing components of each band with a number of bits corresponding to a level of a difference between the energy of each band and the set allowable noise level. Detecting the amount of output information obtained in the quantization step, detecting an error between the detected output and a target value, and based on the detected error and the energy of each of the bands, The number of allocated bits is corrected so as to allocate a large number of bits to a band where energy is concentrated in the tone signal, and the amount of information in a predetermined period is corrected. Digital signal encoding method is characterized in that so as to kept constant.