JP2002311994A - Method and device for coding stereophonic audio signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that spectra are suddenly changed at the boundary of a non-intensity stereophonic process and an intensity stereophonic process while conducting an intensity stereophonic coding, distortion is generated by the change and the tone quality is deteriorated. SOLUTION: The stereophonic audio signal coding device is provided with time/frequency converting sections 110 and 111 which convert stereophonic audio signals to frequency spectra, an intensity signal generating section 120 which conducts a high efficiency coding using the intensity stereophonic coding, a switching section 130 and a quantization and coding section 150. Moreover, the device is provided with an intensity boundary spectrum interpolation section 140 to replace non-intensity stereophonic process frequency spectra in the vicinity of the boundary of the non-intensity stereophonic process and the intensity stereophonic process with the frequency spectra which are interpolated by the non-intensity stereophonic process frequency spectra and the intensity stereophonic process frequency spectra.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a method and apparatus for efficiently encoding a stereo audio signal using intensity stereo encoding.

[0002]

2. Description of the Related Art In recent years, low bit rates, for example, 64k
As a high-efficiency encoding method for stereo audio signals in bps, a method using intensity stereo encoding is widely used. Methods using intensity stereo include MPEG-1 audio and MPEG-2 audio, and MPEG-2 AAC (Adva
nced Audio Coding).

Conventionally, as a stereo audio signal encoding method using intensity stereo encoding, for example, the MPEG-2 AAC standard (ISO / IEC 13818-7,
Information technology-Generic coding of moving
pictures and associated audio information-Part
7: Advanced Audio Coding (AAC)).

FIG. 9 shows a block diagram of a conventional stereo audio signal encoding apparatus using intensity coding. In FIG. 9, 910 and 911 are time / frequency conversion units, 920 is an intensity signal generation unit, 930 is a switching unit that switches the frequency spectrum of left and right channels according to switching information of non-intensity / intensity stereo processing, 940 Is a quantization and coding unit. The operation of the stereo audio signal encoding device configured as described above will be described below.

An input audio signal (left time signal) on the time axis of the left channel is converted into a time / frequency converter 910.
In FIG. 10, orthogonal transform is performed as shown in FIG. 10 to transform the spectrum into a spectrum on the frequency axis (left spectrum), and the transformed spectrum is grouped in band units. As the orthogonal transform, MDCT (Modified Discrete Cosi
ne Transform: Modified discrete cosine transform) or FFT (Fast
Fourier Transform) is used. Also, the number of spectra per band is not uniform, and the number of spectra per band often increases as the frequency increases.

Similarly, the input right channel time axis audio signal (right time signal) is orthogonally transformed by a time / frequency converter 911 as shown in FIG. 10 to obtain a frequency axis spectrum (right spectrum). And the converted spectra are grouped in band units.

Next, an intensity signal generation section 920 generates a signal necessary for intensity stereo coding using the left spectrum and the right spectrum as inputs.

First, the intensity stereo coding method will be described. FIG. 11 shows an explanatory diagram of the intensity stereo coding method. Intensity stereo means that at a predetermined frequency (usually 3 kHz to 6 kHz) or higher, the spectral envelope (rough shape) is more important than the fine structure of the spectrum, and the signals of the left and right channels Utilizing the fact that the direction of the sound source is sensed by the level difference, the sum signal (MPEG) of the frequency spectra of the left and right channels above the predetermined frequency is used.
In the AAC of -2, the coding efficiency is improved by coding the ratio between the sum signal and the signal level of the left and right channels (corrected to match the signal level of the left channel) and the signal level of the left and right channels. It is. In intensity stereo coding, only one channel needs to be transmitted for frequency spectrum information that needs to be transmitted, so that high coding efficiency can be realized. However, sound quality may deteriorate due to the inability to reproduce the fine structure of the spectrum.

In the MPEG-2 AAC, as information indicating intensity stereo coding, switching information of non-intensity / intensity stereo processing indicating whether non-intensity stereo processing or intensity stereo processing is performed for each band, Left intensity spectrum that represents the spectrum during the intensity stereo processing of the channel, phase information that indicates whether the phase relationship between the spectrum of the left channel and the spectrum of the right channel is in phase or opposite phase, and the intensity stereo of the right channel from the left intensity spectrum Directivity information indicating the ratio of the signal level of the right channel to the left channel for obtaining the right intensity spectrum representing the spectrum at the time of processing is used.

FIG. 12 is a flowchart showing steps of intensity coding processing in the conventional audio signal coding method.

In step 1201, a band to be subjected to intensity stereo processing and a band not to be subjected to intensity stereo processing are determined. Here, the frequency is 6 kHz
The above bands are determined to be bands on which intensity stereo processing is performed. That is, a band from the intensity stereo start band Bs having a frequency of 6 kHz or more to the intensity stereo end band Be is determined as a band for performing intensity stereo processing.

Next, at step 1202, the energy of the sum signal of the left channel, the right channel, and the left and right channels, E _L (B), E _R (B), and E _L , for each band for which intensity stereo processing is performed. _{+ R} (B) is calculated. Here, B is a band number.

In step 1203, all phase information for each band is set to the same phase. Here, it is assumed that the phase information takes a value of 1 in the case of in-phase and a value of -1 in the case of out-of-phase.

In step 1204, direction information is calculated for each band. The direction information is the square root of the energy ratio of the left and right channels, that is, SQRT (E
_L (B) / E _R (B)). Where SQRT
(X) is a function representing the square root of X.

Next, at step 1205, a left intensity spectrum representing the spectrum of the intensity stereo processing of the left channel for each band is calculated by (Equation 1).

[0016]

(Equation 1)

In equation (1), Lspec (i) and Rspec
c (i) represents the spectrum of spectrum number i of the left and right channels, respectively, and is (B) represents the first spectrum number of band B.

The intensity signal generation section 920 performs the processing from step 1201 to step 1205 to switch information of non-intensity / intensity stereo processing required for intensity stereo processing, left intensity spectrum, phase information, Generate and output direction information.

Next, in step 1206 of FIG. 12, the spectrum of the left channel is replaced with a left intensity spectrum, and the spectrum of the right channel is replaced with a zero spectrum.

The above spectrum replacement process is performed by the switching unit 930 in FIG. According to the switching information of the non-intensity / intensity stereo processing, the switching unit 930 connects the switches of FIG. 9 to the upper side to output the left spectrum and the right spectrum during the non-intensity stereo processing, and outputs the intensity stereo processing. Sometimes a switch is connected to the lower side to output a left intensity spectrum and a zero spectrum. This zero spectrum is not transmitted as encoded data.

FIG. 13 shows the concept of the spectrum of the left and right channels obtained as described above.

The quantization and coding section 940 of FIG. 9 calculates the masking level of the spectrum, that is, the permissible quantization noise level, based on the auditory model, for the two spectra output from the switching section 930, and calculates the level. The spectrum is quantized on the basis of the permissible quantization noise level thus obtained, coding processing such as Huffman coding is performed, and highly efficient coded data is output.

FIG. 14 is a block diagram of a stereo audio signal decoding apparatus for decoding highly efficient encoded data using the intensity stereo encoding. FIG.
, 1410 is a decoding and inverse quantization unit, 1420
Is a right intensity spectrum reproducing section, and 1430 is a switch for switching between a spectrum of the non-intensity stereo processing and a spectrum of the intensity stereo processing (right intensity spectrum) of the right channel according to switching information of the non-intensity / intensity stereo processing. And 1440 and 1441 are frequency / time conversion units.
The operation of the stereo audio signal decoding device configured as described above will be described below.

The encoded data input to the stereo audio signal decoding device is decoded and dequantized by
Are decoded and dequantized. The right intensity spectrum reproducing unit 1420 includes a decoding and inverse quantization unit 1410
The right intensity spectrum is reproduced using the left intensity spectrum, the phase information, and the direction information from.

FIG. 15 is a block diagram showing a configuration of the right intensity spectrum reproducing section 1420. As shown in the figure, the right intensity spectrum has "1" in the left intensity spectrum when the phase information is in-phase,
In the case of opposite phase, multiply by “−1” and then directional information SQ
Reproduction is performed by dividing by RT (E _L (B) / E _R (B)).

In switching section 1430, according to the switching information of non-intensity / intensity stereo processing from decoding and inverse quantization section 1410, a switch is connected to the upper side during non-intensity stereo processing as the spectrum of the right channel. Then, the right channel spectrum from the decoding and inverse quantum lower part is output as it is, and at the time of intensity stereo processing, the switch is connected to the lower side to output the right intensity spectrum from the right intensity spectrum reproducing unit 1420.

The frequency / time conversion section 1440 performs an inverse orthogonal transform of the input left-channel frequency axis spectrum (left spectrum) to convert it into a left-channel time axis audio signal (left time signal). Output. Similarly, the frequency / time conversion unit 1441 performs an inverse orthogonal transform on the input right-channel frequency axis spectrum (right spectrum) to convert it into a right-channel time axis audio signal (right time signal) and output it. I do.

[0028]

However, in the above-mentioned conventional stereo audio signal encoding method and apparatus,
A sharp change in the spectrum at the band of the non-intensity stereo processing and the band boundary of the intensity stereo processing causes distortion components, and the sound quality is degraded, especially when a signal having a high tone property is near the boundary. There was a problem that there is.

If the phase information indicating whether the left and right channels of the intensity stereo coding are in phase or opposite phase changes frequently every time frame, or changes frequently every frequency band, a distortion component is generated. However, there is a problem that sound quality is deteriorated.

The present invention solves the above-mentioned problems and provides a method and apparatus for encoding a stereo audio signal using intensity stereo encoding, wherein a band for non-intensity stereo processing and a band for intensity stereo processing are used. It is an object of the present invention to provide a stereo audio signal encoding method and apparatus in which the generation of distortion components is reduced and the sound quality is improved as compared with the related art by smoothly connecting the spectra at the boundaries.

Further, by setting the phase information of the intensity stereo coding in consideration of the previous time frame phase information or the correlation between the left and right channels of the previous time frame, frequent changes in the phase information can be prevented. Hold down,
It is an object of the present invention to provide a stereo audio signal encoding method and apparatus capable of reducing the generation of distortion components and improving the sound quality as compared with the related art.

Further, by setting the phase information of the intensity stereo coding in consideration of the phase information of the adjacent frequency band or the correlation between the left and right channels of the adjacent frequency band, frequent changes in the phase information are performed. It is an object of the present invention to provide a stereo audio signal encoding method and apparatus capable of suppressing the occurrence of distortion components, reducing the occurrence of distortion components, and improving the sound quality as compared with the related art.

[0033]

In order to solve this problem, a stereo audio signal encoding method according to the present invention converts a stereo audio signal into a frequency spectrum, and performs high-efficiency encoding using intensity stereo encoding. A non-intensity stereo processing frequency spectrum near a boundary between the non-intensity stereo processing and the intensity stereo processing is replaced with a frequency spectrum interpolated by the non-intensity stereo processing frequency spectrum and the intensity stereo processing frequency spectrum. It has steps.

The stereo audio signal encoding method of the present invention is a method of converting a stereo audio signal into a frequency spectrum for each frame, and performing high-efficiency encoding using intensity stereo encoding. Determining a correlation between frequency spectra of the stereo audio signal, and representing a phase relationship of a frequency spectrum of the stereo audio signal during intensity stereo decoding according to the correlation between a current frame and a previous frame. Setting phase information.

Also, the stereo audio signal encoding method of the present invention converts a stereo audio signal into a frequency spectrum, groups the frequency spectrum in band units, and performs high efficiency encoding using intensity stereo encoding. Determining the correlation between the frequency spectra of the stereo audio signal for each band, and according to the correlation between a current band and an adjacent band, the stereo audio signal at the time of intensity stereo decoding. Setting phase information indicating the phase relationship of the frequency spectrum.

The stereo audio signal encoding apparatus according to the present invention is an apparatus provided with means for converting a stereo audio signal into a frequency spectrum and means for performing efficient encoding using intensity stereo encoding. Means for replacing the intensity stereo processing and the non-intensity stereo processing frequency spectrum near the boundary of the intensity stereo processing with a frequency spectrum interpolated between the non-intensity stereo processing frequency spectrum and the intensity stereo processing frequency spectrum. .

Further, the stereo audio signal encoding apparatus of the present invention is an apparatus comprising means for converting a stereo audio signal into a frequency spectrum for each frame, and means for performing high-efficiency encoding using intensity stereo encoding. Means for determining a correlation between the frequency spectra of the stereo audio signal for each frame, and, for the stereo audio signal at the time of intensity stereo decoding, according to the correlation between a current frame and a previous frame. Means for setting phase information indicating the phase relationship of the frequency spectrum.

Further, the stereo audio signal encoding apparatus of the present invention converts a stereo audio signal into a frequency spectrum, and groups the frequency spectrum in band units, and a high-efficiency encoding using intensity stereo encoding. Means for determining a correlation between frequency spectra of the stereo audio signal for each band, and intensity stereo decoding according to the correlation between a current band and an adjacent band. Means for setting phase information indicating the phase relationship of the frequency spectrum of the stereo audio signal at the time of conversion.

[0039]

Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1) FIG. 1 is a block diagram showing a configuration of a stereo audio signal encoding apparatus according to an embodiment of the present invention.

In FIG. 1, 110 and 111 are time / frequency converters, 120 is an intensity signal generator, 13
0 is a switching unit that switches the frequency spectrum of the left and right channels according to the switching information of the non-intensity / intensity stereo processing, 140 is an intensity boundary spectrum interpolation unit that interpolates the spectrum near the boundary of the intensity stereo processing, and 150 is It is a quantization and coding unit. The operation of the stereo audio signal encoding device configured as described above will be described below.

The input audio signal on the time axis of the left channel (left time signal) is converted into a time / frequency
In, the data is subjected to MDCT (Modified Discrete Cosine Transform), converted into a spectrum on the frequency axis (left spectrum), and the converted spectrum is grouped in band units. In the present embodiment, the number of spectra per band is not uniform, and the number of spectra per band increases as the frequency increases.

Similarly, the input right-channel audio signal on the time axis (right time signal) is MDCT-converted by the time / frequency converter 111 into a spectrum on the frequency axis (right spectrum) and converted. The spectra are grouped by band.

Next, the intensity signal generating section 120 receives the frequency spectrum of the left channel and the frequency spectrum of the right channel as input, and performs switching information indicating non-intensity stereo processing or intensity stereo processing for each band. , Left intensity spectrum indicating the spectrum of the intensity stereo processing of the left channel, phase information indicating whether the phase relationship between the left channel and the right channel is in phase or out of phase, and directionality indicating the ratio of the signal level of the right channel to the left channel. Generate and output information.

FIG. 2 is a block diagram showing a configuration of intensity stereo signal generating section 120. In FIG. 2, reference numeral 210 denotes a left intensity spectrum calculation unit that calculates a spectrum of the intensity processing of the left channel;
220 is a phase information calculation unit for calculating phase information, 230 is the sum signal and difference signal of the left and right channels and the left and right channels, and the phase of the left channel frequency spectrum (left spectrum) and the right channel frequency spectrum (right spectrum). The energy and correlation coefficient calculation unit 240 for calculating the number of relations is a direction information calculation unit. The operation of the intensity signal generator 120 configured as described above will be described below.

FIG. 3 is a flowchart showing processing steps of intensity stereo signal generating section 120 and switching section 130. In the figure, steps 301 to 306 represent the processing of the intensity stereo signal generation unit 120, and step 307 represents the processing of the switching unit 130.

First, in step 301, a band to be subjected to intensity stereo processing and a band not to be subjected to intensity stereo processing are determined. In the present embodiment, a band having a frequency of 6 kHz or more is determined as a band for performing intensity stereo processing. That is, a frequency of 6 kHz
The above-described band from the start band Bs to the end band Be is determined as a band for performing intensity stereo processing, and switching information of non-intensity / intensity stereo processing corresponding to this is output.

Next, in step 302, for each band, the energy of the sum signal and the difference signal of the left channel, the right channel, and the left and right channels of band B are E (B) _L ,
E (B) _R , E (B) _{L + R} , and E (B) _LR are calculated by (Equation 2). In the following, E (B) _L , E (B) _R , E (B) _{L + R} ,
The _sum of E (B) _LR is called energy information.

[0049]

(Equation 2)

In equation (2), Lspec (i) and Rspec
c (i) represents the spectrum of spectrum number i of the left and right channels, respectively, and is (B) represents the first spectrum number of band B.

Next, in step 304, the cross-correlation coefficient C (B) of the band B is calculated by (Equation 3) for each band.

[0052]

(Equation 3)

Next, the normalized cross-correlation coefficient Cn (B) is
Cn (B) = C (B) / SQRT (E (B) _L · E (B) _R ).

The energy and correlation coefficient calculating section 230 performs the processing from step 301 to step 303, outputs the normalized cross-correlation coefficient to the phase information calculating section 220, and makes the left intensity spectrum calculating section 210 The energy information for each band is output to the information calculation unit 240, and the switching information for non-intensity / intensity stereo processing is output to the switching unit 130, the intensity boundary spectrum interpolation unit 140, and the quantization and encoding unit 150.

Next, in step 304, phase information is calculated. FIG. 4 shows a detailed flowchart of the phase information calculation step 304. The calculation of the phase information is performed for each band of the intensity stereo processing. Band 401 in step 401
Is set to the start band Bs of the intensity stereo processing. Next, at step 402, the normalized cross-correlation coefficient C
It is determined whether n (B) is equal to or greater than zero. When the normalized cross-correlation coefficient is equal to or greater than zero, the process proceeds to step 403, where the phase information is set to the same phase. When the normalized cross-correlation coefficient is negative, the procedure goes to step 404, where the phase information is set to the opposite phase. Next, in step 405, the band number B is incremented by one.
In step 406, it is determined whether or not the band number B exceeds the end band Be. If not, the process from steps 402 to 405 is repeated to set phase information of each band. If B exceeds Be, the process ends.

The phase information calculation processing of step 304 is performed by the phase information calculation section 220.

Next, in step 305, direction information is calculated. The direction information is, for each band, the square root of the ratio of the right channel energy to the left channel energy,
That is, it is calculated by SQRT (E _L (B) / E _R (B)). The direction information calculation unit 240 calculates and outputs the direction information as described above.

Next, in step 306, the frequency spectrum (left intensity spectrum) of the intensity stereo processing of the left channel is converted to the sum signal of the left and right channels using the phase information calculated in step 304 (when the phase information is in phase). ) Or the difference signal (when the phase information is in the opposite phase), the sum signal or the difference signal is corrected such that the energy of the original left channel is the same as that of the original left channel, and calculation is performed for each band. That is, when the phase information is in-phase, the left intensity spectrum is calculated by (Equation 4).

[0059]

(Equation 4)

When the phase information is out of phase, the left intensity spectrum is calculated by (Equation 5).

[0061]

(Equation 5)

Left intensity spectrum calculator 210
Then, the left intensity spectrum is calculated in step 306 by using as input the frequency spectrum of the left channel (left spectrum), the frequency spectrum of the right channel (right spectrum), and the energy information from the energy and correlation coefficient calculator 230. Outputs the left intensity spectrum.

The operation of the intensity signal generator 120 has been described above.

Next, in step 307, the frequency spectrum of the left channel is replaced with the left intensity spectrum and the spectrum of the right channel is replaced with the zero spectrum for the band of the intensity stereo processing.

The switching unit 130 performs a spectrum replacement process in step 307. That is, the switching unit 130 connects the respective switches to the upper side of FIG. 1 during the non-intensity stereo processing according to the switching information of the non-intensity / intensity stereo processing for each band from the intensity signal generation unit 120, and At the time of city stereo processing, connect the switch to the lower side. Accordingly, the left channel is switched between the left spectrum and the left intensity spectrum and the right channel is switched between the right spectrum and the zero spectrum according to the non-intensity / intensity stereo processing, as shown in FIG. Such a spectrum is generated and output.

Intensity boundary spectrum interpolation section 14
0, the spectrum from the switching unit 130 is used as an input,
The non-intensity stereo processing frequency spectrum in the band near the boundary between the non-intensity stereo processing and the intensity stereo processing is replaced with a spectrum interpolated by the non-intensity stereo processing frequency spectrum and the intensity stereo processing frequency spectrum. That is, the frequency spectrum of the non-intensity stereo processing of the band (Bs-1) in FIG. 13 is replaced with the interpolated frequency spectrum. Intensity stereo processing frequency spectrum L of left and right channels of band (Bs-1)
Ispec (i) and RIspec (i) are obtained as follows using the band B phase information.

That is, when the phase information of the band B is in phase, it is calculated by (Equation 6).

[0068]

(Equation 6)

If the phase information is in the opposite phase, it is calculated by (Equation 7).

[0070]

(Equation 7)

Next, the interpolation spectra L'spec (i) and R'spec (i) of the left and right channels to be replaced are shown.
(However, is (Bs-1) ≤i≤is (Bs) -1)
Is calculated by (Equation 8).

[0072]

(Equation 8)

The interpolation spectrum L'spec (i) and R'spec (i) calculated as described above are used for the band (B
The non-intensity stereo processing frequency spectra Lspec (i) and Rspec (i) of s-1) are respectively replaced.

The quantization and encoding section 130 calculates a masking level of the spectrum, that is, an allowable quantization noise level based on the auditory model using the frequency spectrum calculated as described above, and calculates the calculated allowable quantum noise level. Quantization of the spectrum based on the quantization noise level,
It performs encoding processing such as Huffman encoding and outputs highly efficient encoded data.

As described above, in the present embodiment, the spectrum near the boundary between the non-intensity stereo processing and the intensity stereo processing can be connected smoothly by providing the intensity boundary spectrum interpolation processing. It is possible to reduce the occurrence of distortion components due to abrupt switching between the intensity stereo processing frequency spectrum and the non-intensity stereo processing frequency spectrum in intensity stereo coding.

(Embodiment 2) The configuration of a stereo audio signal encoding apparatus and the configuration of an intensity signal generator according to Embodiment 2 of the present invention are the same as those of Embodiment 1, and are respectively shown in FIGS. Indicated by Embodiment 1
The difference between the second embodiment and the second embodiment resides in the processing contents of the phase information calculation section 220. Hereinafter, the processing contents of the phase information calculation section 220 in the second embodiment will be described, and the description of the other operations will be omitted. I do.

FIG. 5 shows a flowchart of the phase information calculation processing step in the second embodiment. The calculation of the phase information is performed for each band of the intensity stereo processing. In step 501, band B is set as a start band Bs for intensity stereo processing. Next, at step 502, the normalized cross-correlation coefficient Cn (B) is set to a predetermined threshold value TH (B).
It is determined whether or not this is the case. In the present embodiment, the initial value of the threshold value TH (B) is set to zero. When the normalized cross-correlation coefficient is equal to or larger than the threshold value, the process proceeds to step 503, where the phase information is set to the same phase and the value of TH (B) is set to a predetermined negative value −OFFS (where 0 ≦ OFFS ≦
Set to 1). When the normalized cross-correlation function is smaller than the threshold value, the process proceeds to step 504, where the phase information is set to the opposite phase and the value of TH (B) is set to the predetermined value OFF.
Set to S. Next, in step 505, the band number B is set to 1
To increase. In step 506, it is determined whether the band number B does not exceed the end band Be. If not, the process from step 502 to step 505 is repeated to set the phase information of each band. If B exceeds Be, the process ends.

As described above, in the second embodiment, the condition for judging the phase information is given a hysteresis, and if the phase information of the previous frame is in phase, it is easy to enter the same phase, and the phase is opposite to that of the previous frame. Frequent change of phase information for each time frame can be suppressed, and sound quality deterioration due to frequent switching of phase information for each time frame of intensity stereo coding can be reduced.

(Embodiment 3) The configuration of a stereo audio signal encoding device and the configuration of an intensity signal generator according to Embodiment 3 of the present invention are the same as those of Embodiment 1, and are respectively shown in FIGS. Indicated by Embodiment 1
The difference between the third embodiment and the third embodiment resides in the processing contents of the phase information calculating section 220. Therefore, the processing contents of the phase information calculating section 220 in the third embodiment will be described below, and the description of the other operations will be omitted. .

FIG. 6 shows a flowchart of the phase information calculation processing step in the third embodiment. The calculation of the phase information is performed for each band of the intensity stereo processing. In step 601, band B is set as a start band Bs for intensity stereo processing. Next, step 602
From the normalized cross-correlation coefficient Cn (B) of the current frame and the normalized cross-correlation coefficient Cnprev (B) of the previous frame, CC (B) = kCn (B) + (1-k) Cnprev (B ), A weighted normalized cross-correlation coefficient CC (B) is calculated. Where Cnp
The initial value of rev (B) is set to zero. Also, where
k (where 0 ≦ k ≦ 1) is a weighting coefficient, for example,
Set k to a value of 0.75.

Next, at step 603, it is determined whether or not the weighted normalized cross-correlation coefficient is equal to or larger than zero. When the weighted normalized cross-correlation coefficient is equal to or greater than zero, the process proceeds to step 604, and the phase information is set to the same phase. When the weighted normalized cross-correlation coefficient is negative, the procedure goes to step 605 to set the phase information to the opposite phase.

In step 606, the normalized cross-correlation coefficient Cnpr of the previous frame is prepared in preparation for the processing of the next frame.
ev (B) is the normalized cross-correlation coefficient Cn of the current frame.
Replace with (B).

Next, at step 607, the band number B is incremented by one. In step 608, it is determined whether the band number B does not exceed the end band Be. If the band number B does not exceed the end band Be, the processing of steps 602 to 607 is repeated to set phase information of each band. If B exceeds Be, the process ends.

As described above, in the third embodiment, the determination of the phase information is performed using the weighted normalized cross-correlation coefficient, so that the current frame of the current frame is considered in consideration of the normalized cross-correlation coefficient of the previous frame. Phase information can be set, and if the normalized cross-correlation coefficient of the previous frame is positive or zero, it is easy to enter in-phase, and if the normalized cross-correlation coefficient of the previous frame is negative, the inverse This makes it easier to enter a phase, suppresses frequent changes in phase information for each time frame, and reduces sound quality degradation due to frequent switching of phase information for intensity stereo coding for each time frame.

In the above description, an example is shown in which only the normalized cross-correlation coefficient of the immediately preceding frame is considered, but the normalized cross-correlation coefficient of the previous frame may be considered.

In the above description, an example is shown in which an FIR filter is used to calculate the weighted normalized cross-correlation coefficient. However, an IIR filter may be used.

(Embodiment 4) The configuration of a stereo audio signal encoding apparatus and the configuration of an intensity signal generator according to Embodiment 4 of the present invention are the same as those of Embodiment 1, and are respectively shown in FIGS. Indicated by Embodiment 1
The difference between the second embodiment and the fourth embodiment lies in the processing contents of the phase information calculation section 220. Therefore, the processing contents of the phase information calculation section 220 in the fourth embodiment will be described below, and the description of the other operations will be omitted. .

FIG. 7 shows a flowchart of the phase information calculation processing step in the fourth embodiment. The calculation of the phase information is performed for each band of the intensity stereo processing. In step 701, band B is set as a start band Bs for intensity stereo processing. Next, step 702
And the normalized cross-correlation coefficient Cn (B) becomes a predetermined threshold TH.
(B) It is determined whether or not this is the case. In the present embodiment, the initial value of the threshold value TH (B) is set to zero. If the normalized cross-correlation coefficient is equal to or larger than the threshold, step 703
To set the phase information to the same phase and set the value of the threshold value TH (B + 1) of the adjacent band to a predetermined negative value −OFFS
(However, 0 ≦ OFFS ≦ 1) is set. When the normalized cross-correlation coefficient is smaller than the threshold value, step 7
04 and sets the phase information to reverse phase and
The value of (B + 1) is set to OFFS. Then step 7
At 05, the band number B is incremented by one. In step 706, it is determined whether the band number B does not exceed the end band Be. If the band number does not exceed end band Be, steps 702 to 705 are performed.
By repeating the above processing, the phase information of each band is set. When the band number B exceeds the end band Be, the process ends.

As described above, in the fourth embodiment, the determination condition of the phase information is given a hysteresis, and if the phase information of the adjacent band is in phase, it is easy to enter the same phase, and the phase information of the adjacent band is inverted. By making the phase easier to enter the opposite phase, it is possible to suppress frequent changes in phase information for each frequency band, and reduce sound quality degradation due to frequent switching of phase information for intensity stereo coding for each frequency band. it can.

In the above description, an example in which the influence of the phase information of the band immediately below is taken into consideration, but the influence of the phase information of the band immediately above may be taken into consideration.

(Embodiment 5) The configuration of a stereo audio signal encoding apparatus and the configuration of an intensity stereo signal generator according to Embodiment 5 of the present invention are described in Embodiment 1.
And are shown in FIGS. 1 and 2, respectively. The difference between the first embodiment and the fifth embodiment is that
Since the processing content is 0, the processing content of the phase information calculation unit 220 in the fifth embodiment will be described below, and the description of the other operations will be omitted.

FIG. 8 shows a flowchart of the phase information calculation processing step in the fifth embodiment. The calculation of the phase information is performed for each band of the intensity stereo processing. In step 801, band B is set as a start band Bs for intensity stereo processing. Next, step 802
And the normalized cross-correlation coefficient Cn (B) of the current band and the band-normalized cross-correlation coefficients Cn (B-1) and C
From n (B + 1), CC (B) = k1 (B) · Cn (B-1) + k2 (B) · Cn (B)
+ k3 (b) · Cn (B + 1) gives a weighted normalized cross-correlation coefficient C
Calculate C (B). However, k1 (B) + k2 (B) + k3 (B) = 1.

Here, k1 (B), k2 (B), k3
(B) is a weighting coefficient, which takes a positive or zero value. In the present embodiment, for example, for B = Bs, k1
(Bs) = 0, k2 (Bs) = 0.875, k3 (B
s) = 0.125, and for B where Bs <B <Be, k
1 (B) = 0125, k2 (B) = 0.75, k3 (B
s) = 0.125, for B = Be, k1 (Be)
= 0.125, k2 (Bs) = 0.875, k3 (B
e) Set to a value of = 0.

Next, at step 803, it is determined whether or not the weighted normalized cross-correlation coefficient is equal to or greater than zero. When the weighted normalized cross-correlation coefficient is equal to or greater than zero, the process proceeds to step 804, and the phase information of band B is set to the same phase. When the weighted normalized correlation function is negative, the procedure goes to step 805 to set the phase information to the opposite phase.

Next, at step 806, the band number B is incremented by one. In step 807, it is determined whether the band number B does not exceed the end band Be. If the band number B does not exceed the end band Be, the process of steps 802 to 806 is repeated to set phase information of each band. If B exceeds Be, the process ends.

As described above, in the fifth embodiment, the determination of the phase information is performed using the weighted normalized cross-correlation coefficient, and the current band of the current band is considered in consideration of the normalized cross-correlation coefficient of the adjacent band. Since the phase information is set, if the normalized cross-correlation coefficient of the adjacent band is positive, it is easy to enter the same phase, and if the normalized cross-correlation coefficient of the adjacent band is negative, it is likely to enter the opposite phase. That is, frequent changes in phase information for each frequency band can be suppressed, and sound quality deterioration due to frequent switching of phase information for intensity stereo coding for each frequency band can be reduced.

[0097]

As described above, according to the present invention, the spectrum near the boundary between the non-intensity stereo processing and the intensity stereo processing can be connected smoothly by interpolation, and the spectrum of the non-intensity stereo processing can be smoothly connected. It is possible to reduce the generation of distortion components due to abrupt switching of the spectrum of the intensity stereo processing.

Also, by setting the phase information of the intensity stereo coding in accordance with the phase information of the previous time frame or the correlation between the left and right channels of the previous time frame, the phase information of each time frame of the phase information is set. Frequency degradation due to frequent switching can be reduced.

Also, by setting the phase information of the intensity stereo coding according to the phase information of the adjacent frequency band or the correlation between the left and right channels of the adjacent frequency band, the phase information of each frequency band of the phase information is set. Frequency degradation due to frequent switching can be reduced.

As described above, according to the present invention, it is possible to realize a stereo audio signal encoding method and apparatus in which the sound quality deterioration of intensity stereo encoding is reduced.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a stereo audio signal encoding device according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of an intensity signal generation unit in the stereo audio signal encoding device.

FIG. 3 is a flowchart showing steps of an intensity encoding process in the stereo audio signal encoding method according to the first embodiment of the present invention.

FIG. 4 is a flowchart showing phase information calculation processing steps in the stereo audio signal encoding method.

FIG. 5 is a flowchart showing phase information calculation processing steps in a stereo audio signal encoding method according to Embodiment 2 of the present invention;

FIG. 6 is a flowchart showing phase information calculation processing steps in the stereo audio signal encoding method according to the third embodiment;

FIG. 7 is a flowchart showing phase information calculation processing steps in the stereo audio signal encoding method according to the fourth embodiment;

FIG. 8 is a flowchart showing phase information calculation processing steps in the stereo audio signal encoding method according to the fifth embodiment;

FIG. 9 is a block diagram showing a configuration of a conventional stereo audio signal encoding device.

FIG. 10 is an explanatory diagram of time / frequency conversion for explaining an operation in the conventional example.

FIG. 11 is an explanatory diagram of a method of intensity stereo coding for explaining an operation in the conventional example.

FIG. 12 is a flowchart showing intensity encoding processing steps in the conventional stereo audio signal encoding method.

FIG. 13 is an explanatory diagram of a spectrum of non-intensity / intensity stereo processing for explaining an operation in the conventional example.

FIG. 14 is a block diagram showing the overall configuration of a conventional stereo audio signal decoding device.

FIG. 15 is a block diagram showing a configuration of a right intensity spectrum reproducing unit in the stereo audio signal decoding device of the conventional example.

[Explanation of symbols]

 110, 111 Time / frequency conversion unit 120 Intensity signal generation unit 130 Switching unit 140 Intensity boundary spectrum interpolation unit 150 Quantization and encoding unit 210 Left intensity spectrum calculation unit 220 Phase information calculation unit 230 Energy and correlation coefficient calculation Unit 240 Direction information calculation unit

Claims (8)

[Claims] 1. A method for converting a stereo audio signal into a frequency spectrum and performing high-efficiency coding using intensity stereo coding, wherein the non-intensity stereo processing and the non-intensity near the boundary between the intensity stereo processing are performed. A stereo audio signal encoding method, comprising a step of replacing a stereo processing frequency spectrum with a frequency spectrum interpolated between a non-intensity stereo processing frequency spectrum and an intensity stereo processing frequency spectrum. 2. A method for converting a stereo audio signal into a frequency spectrum for each frame and performing high-efficiency coding using intensity stereo coding, wherein a correlation between the frequency spectra of the stereo audio signal for each frame is provided. And setting phase information indicating a phase relationship of a frequency spectrum of a stereo audio signal at the time of intensity stereo decoding according to the correlation between a current frame and a previous frame. A stereo audio signal encoding method. 3. A method for converting a stereo audio signal into a frequency spectrum, grouping the frequency spectrum in band units, and performing high-efficiency encoding using intensity stereo encoding, wherein the stereo audio signal is divided into bands. Determining the correlation between the frequency spectra of the current band and the phase information representing the phase relationship of the frequency spectrum of the stereo audio signal at the time of intensity stereo decoding according to the correlation between the current band and the adjacent band. A stereo audio signal encoding method. 4. The stereo audio signal encoding method according to claim 2, wherein the correlation is based on a cross-correlation coefficient. 5. An apparatus comprising: means for converting a stereo audio signal into a frequency spectrum; and means for performing high-efficiency coding using intensity stereo coding, wherein the non-intensity stereo processing and the intensity stereo processing are performed. A stereo audio signal encoding apparatus characterized by comprising means for replacing a non-intensity stereo processing frequency spectrum near a boundary with a frequency spectrum interpolated by a non-intensity stereo processing frequency spectrum and an intensity stereo processing frequency spectrum. 6. An apparatus comprising: means for converting a stereo audio signal into a frequency spectrum for each frame; and means for performing high-efficiency coding using intensity stereo coding, wherein the stereo audio signal is provided for each frame. Means for determining the correlation between the frequency spectra of
Means for setting phase information indicating a phase relationship of a frequency spectrum of a stereo audio signal at the time of intensity stereo decoding, according to the correlation between a current frame and a previous frame. Signal encoding device. 7. An apparatus comprising: means for converting a stereo audio signal into a frequency spectrum, grouping the frequency spectrum in band units, and means for performing high-efficiency coding using intensity stereo coding. ,
Means for determining a correlation between frequency spectra of the stereo audio signal for each band, and a phase of a frequency spectrum of the stereo audio signal at the time of intensity stereo decoding according to the correlation between a current band and an adjacent band. Means for setting phase information indicating the relationship. 8. The stereo audio signal encoding apparatus according to claim 6, wherein the correlation is based on a cross-correlation coefficient.