JP2006047561A - Audio signal encoding device and audio signal decoding device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an audio signal encoding device capable of performing efficient encoding processing and an audio decoding device capable of decoding an encoded signal generated through such encoding processing. <P>SOLUTION: The audio decoding device is provided with an FM composition part 12 which predict spectrum data of a frequency range by using an FM composition system capable of representing a complicated waveform with less parameters than linear prediction, a residue signal calculation part 13 which finds a residue signal as the difference between the signal and the original signal, and an quantization/encoding part 14 which encodes the parameters and residue signal, so that encoding processing which is more efficient than processing using the linear prediction can be performed. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an audio signal encoding apparatus that encodes an audio signal, and an audio signal decoding apparatus that decodes an encoded bit stream.

  Conventionally, as a method for encoding an audio signal, there is an MPEG (Moving Picture Expert Group) Audio standard. There are a plurality of MPEG Audio standards. In the MPEG2 AAC and MPEG4 AAC standards, a TNS (Temporary Noise Shaping) technique can be used as a tool for increasing compression efficiency.

  FIG. 10 is a block diagram of an encoding apparatus compliant with the MPEG2 audio standard AAC (Advanced Audio Coding) standardized by ISO / IEC 13818-7, and FIG. 11 is a block of a TNS processing unit in the reference software of the MPEG2 audio standard AAC. FIG.

  In FIG. 10, an auditory psychological model unit 110 analyzes an input audio signal in accordance with human auditory characteristics and calculates a signal-to-mask ratio (SMR) value. The gain control unit 111 is used only for the SSR profile, divides the input signal into four equally-spaced bands, and controls gain for bands other than the lowest band. The MDCT unit 112 converts the input audio signal in the time domain into spectrum data in the frequency domain. The TNS processing unit 113 controls the temporal shape of the quantization noise. The intensity / coupling unit 114 and the M / S stereo unit 116 are one of modules that efficiently process stereo signals, and perform stereo correlation coding processing. The prediction unit 115 performs predictive coding. The normalization coefficient unit 117 calculates a normalization coefficient, and the quantization unit 118 nonlinearly quantizes the acoustic signal based on the normalization coefficient. Each quantized output is encoded by the noiseless encoding unit 119, and the multiplexer unit 120 forms a bit stream. Also, the spectrum processing unit 121 is grouped from the MDCT unit 112 that performs spectrum processing to the processing unit of the M / S stereo unit 116, and the processing unit of the noiseless coding unit 119 is grouped from the normalization coefficient unit 117 that performs quantization / coding. The quantization / encoding unit 122 is used.

  Next, the operation of the TNS processing unit 113 will be described with reference to FIG. In FIG. 11, the MDCT unit 131 converts a time domain input audio signal into frequency domain spectrum data. The linear prediction unit 132 performs linear prediction on spectrum data in the frequency domain. If the spectrum data can be predicted by linear prediction, the residual signal calculation unit 133 calculates a residual signal between the spectrum predicted by the linear prediction unit 132 and the spectrum data converted by the MDCT unit 131. . Further, the residual signal and the linear prediction coefficient are converted into an output bit stream by the quantization / encoding unit 134.

  In this way, the dispersion of MDCT coefficients is reduced by the TNS process, and the spectrum is flattened. Usually, the quantization noise is distributed on the entire time axis on the average, but the quantization noise is distributed a lot on the time axis where the waveform is large by the TNS processing. As a result, sound quality deterioration called pre-echo can be reduced.

  When performing the decoding process, a spectrum is calculated from the decoded residual signal and the linear prediction coefficient, and the audio signal is restored by performing inverse MDCT on the spectrum.

In addition, a speech coding / decoding device that compresses speech by reducing the increase in the amount of transmitted information using such linear prediction and the residual signal and increasing the pitch prediction accuracy has also been proposed (see, for example, Patent Document 1). ).
JP-A-9-081911

  However, in the TNS processing in the conventional audio signal encoding / decoding apparatus, prediction is effective for a simple waveform, but linear prediction cannot be performed for a complicated waveform, and the effect of the TNS processing can be sufficiently exhibited. There was a problem that I could not.

  The present invention has been made to solve such a conventional problem, and an audio signal encoding apparatus capable of performing efficient encoding processing and decoding such an encoded signal. An object of the present invention is to provide an audio signal decoding apparatus capable of performing the above.

  The audio signal encoding apparatus of the present invention includes a time-frequency conversion unit that converts a time-domain audio signal into a frequency domain, an FM synthesis unit that approximates the signal converted into the frequency domain by the time-frequency conversion unit, and the time A residual signal calculating means for calculating a difference between the signal converted into the frequency domain by the frequency converting means and the frequency domain signal approximated by the FM synthesizing means; and a frequency domain calculated by the residual signal calculating means. An encoding means for encoding the residual signal and the parameters used in the FM synthesizing means is provided.

  With this configuration, frequency domain spectrum data is predicted using an FM synthesis method that can represent a complex waveform with fewer parameters than linear prediction, and a residual signal that is a difference from this signal is obtained. Since the parameter and the residual signal are encoded, the encoding process can be performed more efficiently than the process using linear prediction.

  The audio signal encoding apparatus according to the present invention further includes an applied frequency calculating unit that calculates an application range of a frequency for obtaining a residual signal in the residual signal calculating unit, and the residual signal calculating unit includes the applied frequency calculating unit. The residual signal of the applied frequency band calculated by the means is calculated, and the encoding means has a configuration characterized in that the applied frequency band calculated by the applied frequency calculating means is encoded as a parameter. Yes.

  With this configuration, a frequency band in which a residual signal is calculated is designated, spectrum data in the frequency domain is predicted using an FM synthesis method capable of expressing a complex waveform with fewer parameters than linear prediction, and a signal obtained by FM synthesis. Since the residual signal is encoded only within the range where the difference between and is substantially affected, the residual signal is not encoded, and the encoding process is more efficient than the process using linear prediction. In addition, the encoding efficiency can be further increased without degrading the reproduction quality.

  Furthermore, the audio signal decoding apparatus of the present invention uses a decoding means for decoding the encoded frequency domain residual signal and the parameters used for FM synthesis, and the parameters decoded by the decoding means. FM combining means for performing FM combining, an addition signal calculating means for adding the frequency domain residual signal decoded by the decoding means and the frequency domain signal output by the FM combining means, and the addition A frequency time conversion means for converting a frequency domain signal generated by the signal calculation means into a time domain audio signal.

  With this configuration, it is possible to reproduce the audio signal by decoding the parameter of the FM synthesized signal and the residual signal of the audio signal and the FM synthesized signal, so that the compression rate is higher than that of the encoded signal by linear prediction. From the encoded signal, the audio signal can be decoded.

  Furthermore, the audio signal decoding apparatus of the present invention includes a decoding means for decoding the encoded frequency domain residual signal, the parameter used for FM synthesis, and the parameter of the frequency band in which the residual signal is calculated, FM synthesis means for performing FM synthesis using the FM synthesis parameters decoded by the decoding means, and a frequency domain decoded by the decoding means in an applied frequency band set by the frequency band parameters An addition signal calculation means for adding the residual signal of the signal and the frequency domain signal output by the FM synthesis means, and a frequency for converting the frequency domain signal generated by the addition signal calculation means into a time domain audio signal. And a time conversion means.

  With this configuration, the audio signal is reproduced by decoding the FM composite signal parameter, the residual signal of the audio signal and the FM composite signal, and the applied frequency information indicating the frequency to which the residual signal is applied. Since the residual signal is encoded only in a range in which the difference from the signal by FM synthesis substantially affects, the encoded signal having a higher compression rate than that of the encoded signal by linear prediction can be used. Can be decrypted.

  The present invention uses an FM synthesis method capable of expressing a complex waveform with fewer parameters than linear prediction, and FM synthesis means for predicting frequency domain spectrum data, and a residual difference between this signal and the original signal. By providing a residual signal calculating means for obtaining a difference signal and an encoding means for encoding the parameter and the residual signal, the encoding process can be performed more efficiently than the process using linear prediction. Therefore, it is possible to provide an audio signal encoding / decoding device having the effect of being able to perform the above.

  Hereinafter, an audio signal encoding device and an audio signal decoding device according to embodiments of the present invention will be described with reference to the drawings.

(First embodiment)
A block diagram of an audio signal encoding apparatus according to the first embodiment of the present invention will be described with reference to FIG.

  As shown in FIG. 1, the audio signal encoding apparatus 10 can synthesize a high-quality sound with a small number of parameters and a time-frequency conversion unit 11 that converts an input audio signal in the time domain into a spectrum signal in the frequency domain. The FM synthesis unit 12, the residual signal calculation unit 13 that calculates the difference between the spectrum signal in the frequency domain and the spectrum signal in the frequency domain calculated by the FM synthesis unit 12, and the residual signal calculation unit 13 The configuration includes a residual signal and a quantization / encoding unit 14 that quantizes / encodes parameters used in the FM synthesis unit 12.

  The operation of the audio signal encoding apparatus 10 configured as described above will be described.

  First, the time domain input audio signal is converted into a frequency domain spectrum signal by the time frequency converter 11. The time frequency conversion unit 11 can use FFT, MDCT, or the like. Next, the FM synthesizing unit 12 approximates the spectrum signal in the frequency domain converted by the time-frequency converting unit 11 using the FM synthesizing method. The FM synthesis method is a voice synthesis method widely used as a musical instrument synthesizer and a sound source of a mobile phone, and can output a complex waveform with a small number of parameters. In the FM synthesis method, various sounds are created by deforming a waveform by FM (frequency modulation).

  Here, the FM synthesizing unit 12 is composed of a plurality of transmitters, and creates a voice by modulating another transmitter with the output of the transmitter. The number of the transmitters can be changed according to the circuit scale and computing capacity. Two to four sound sources are used for mobile phones, and four to eight are used for musical instruments. The simplest configuration diagram of the FM synthesis unit is shown in FIG.

  As shown in FIG. 2, the FM synthesizing unit includes two transmitters 21 and 22, and the transmitter 21 is connected to the transmitter 22. The transmitter 22 modulates and outputs the waveform output from the transmitter 21. The method of connecting the transmitters is called an algorithm, and can have a plurality of algorithms. For example, there are those in which two transmitters are connected in series as shown in FIG. 2, and those in which two transmitters 31 and 32 are connected in parallel as shown in FIG.

  Next, a configuration method of the FM synthesis unit 12 will be described with reference to FIG. In the FM synthesizing unit 12, the fundamental frequency and the like of the spectrum signal in the frequency domain are analyzed by the analyzing unit 46, an FM synthesis algorithm and parameters are set based on the analysis result, and the time domain synthesized waveform is output by the FM synthesizing unit 45. To do. The combined waveform is converted into a spectrum signal in the frequency domain by the time frequency converter 49.

  Further, if the circuit has a sufficient amount of calculation, the configuration as shown in FIG. 5 can be adopted. Here, the signal converted into the spectrum signal in the frequency domain by the time-frequency converter 51 is compared with the signal converted from the FM synthesized waveform generated by the FM synthesizer 55 into the frequency domain by the time-frequency converter 59. The parameters are compared by the unit 57 and the difference between the two spectra is minimized. The FM synthesizing unit 12 outputs a signal obtained by converting the FM synthesized waveform that minimizes the error into a spectrum signal in the frequency domain.

  Next, the residual signal calculation unit 13 calculates a residual signal between the frequency domain spectral signal output by the FM synthesis unit 12 and the frequency domain spectral signal converted by the time frequency conversion unit 11. Next, the quantization / encoding unit 14 quantizes / encodes the residual signal and parameters such as the algorithm and frequency used for FM synthesis, and outputs a bit stream.

  It is also possible to apply the M / S stereo part and the psychoacoustic model part, which are tools for increasing the compression efficiency in the MPEG2 audio standard AAC, to this encoding.

  According to the audio signal encoding apparatus of the first embodiment of the present invention as described above, the spectrum data in the frequency domain is predicted using the FM synthesis method that can express a complex waveform with fewer parameters than the linear prediction. By doing so, it is possible to perform more efficient encoding processing than processing using linear prediction.

(Second Embodiment)
Next, a block diagram of an audio signal encoding apparatus according to the second embodiment of the present invention will be described with reference to FIG.

  As shown in FIG. 6, the audio signal encoding device 60 can synthesize a high-quality sound with a small number of parameters and a time-frequency conversion unit 61 that converts a time-domain input audio signal into a frequency-domain spectral signal. An FM synthesizing unit 62, an applied frequency calculating unit 68 that determines in which frequency band the difference between the spectrum signal in the frequency domain and the spectrum signal in the frequency domain calculated by the FM synthesizing unit 62 is calculated, and the frequency domain In the residual signal calculation unit 63 that calculates the difference between the spectrum signal of the frequency domain and the spectrum signal in the frequency domain calculated by the FM synthesis unit 62, the residual signal calculated by the residual signal calculation unit 63 and the FM synthesis unit 62 The configuration includes a quantization / encoding unit 64 that quantizes / encodes used parameters and the like.

  The operation of the audio signal encoding device 60 configured as described above will be described.

  First, the time domain input audio signal is converted into a frequency domain spectrum signal by the time frequency converter 61. The time frequency conversion unit 61 can use FFT, MDCT, or the like. Next, the FM synthesizing unit 62 approximates the frequency domain spectrum signal converted by the time-frequency converting unit 61 using the technique described in the first embodiment using the FM synthesizing method.

  Next, the applied frequency calculation unit 68 determines in which frequency band the difference is to be performed between the spectrum signal in the frequency domain converted by the time frequency conversion unit 61 and the signal calculated by the FM synthesis unit 62. For example, as shown in FIG. 7, the difference is obtained at a certain frequency or higher, and is input to the quantization / encoding unit 64 without obtaining the difference at a certain frequency or lower. Further, a difference may be obtained for each band obtained by band division called a scale factor band set in the MPEG2 audio standard AAC.

  Next, the residual signal calculation unit 63 calculates an application frequency of a residual signal between the frequency domain spectrum signal output by the FM synthesis unit 62 and the frequency domain spectrum signal converted by the time frequency conversion unit 61. Calculation is performed in the frequency band calculated by the unit 68. Next, the quantization / encoding unit 64 quantizes / encodes the residual signal, the frequency band information for which the residual signal has been obtained, the parameters such as the algorithm and frequency used for the FM synthesis, and the bit stream Output.

  It is also possible to apply the M / S stereo part and the psychoacoustic model part, which are tools for increasing the compression efficiency in the MPEG2 audio standard AAC, to this encoding.

  As described above, according to the audio signal encoding device in the second embodiment of the present invention, the frequency band in which the residual signal is calculated can be designated, and a complex waveform can be expressed with fewer parameters than linear prediction. Using the synthesis method, spectrum data in the frequency domain is predicted, and the residual signal is encoded only in a range where the difference from the signal by the FM synthesis becomes a problem, so that the unnecessary residual signal is not encoded. In addition to performing more efficient encoding processing than processing using linear prediction, it is possible to further increase the encoding efficiency without degrading the quality during reproduction.

(Third embodiment)
Next, a block diagram of an audio signal decoding apparatus according to the third embodiment of the present invention will be described with reference to FIG.

  As shown in FIG. 8, the audio signal decoding apparatus 80 performs FM synthesis using a decoding unit 81 that decodes a compressed bitstream, and parameters such as an algorithm and frequency used in FM synthesis at the time of encoding. , An FM synthesizing unit 82 that outputs an FM synthesized signal as a frequency domain signal, an addition signal calculating unit 83 that adds the FM synthesized frequency domain signal and the residual signal, and a frequency domain signal as a time domain The frequency time conversion unit 84 converts the audio signal into an audio signal.

  The operation of the audio signal decoding apparatus 80 configured as described above will be described.

  First, the compressed input bit stream is decoded by a decoding unit 81 into parameters such as an algorithm and frequency used in FM synthesis at the time of encoding, and a residual signal. Next, the FM synthesizing unit 82 performs FM synthesis using parameters such as the algorithm and frequency used in the FM synthesis decoded by the decoding unit 81, and the synthesized waveform is subjected to time-frequency conversion in the frequency domain. Convert to signal.

  Next, the addition signal calculation unit 83 adds the residual signal decoded by the decoding unit 81 and the signal output by the FM synthesis unit 82. The frequency domain signal thus added is converted into a time domain audio signal by the frequency time converter 84.

  Thus, according to the audio signal decoding device in the third embodiment of the present invention, the audio signal is decoded by decoding the parameter of the FM synthesized signal and the residual signal of the audio signal and the FM synthesized signal. Therefore, the audio signal can be decoded from the encoded signal having a higher compression rate than the encoded signal based on the linear prediction.

(Fourth embodiment)
Next, a block diagram of an audio signal decoding apparatus according to the fourth embodiment of the present invention will be described with reference to FIG.

  As shown in FIG. 9, the audio signal decoding device 90 performs FM synthesis using a decoding unit 91 that decodes a compressed bitstream, and parameters such as an algorithm and frequency used in FM synthesis at the time of encoding. , An FM synthesizing unit 92 for outputting the FM synthesized signal as a frequency domain signal, and an addition signal calculating unit for adding the FM synthesized frequency domain signal and the residual signal in a frequency band determined by the applied frequency information 93 and a frequency time conversion unit 94 that converts a frequency domain signal into a time domain audio signal.

  The operation of the audio signal decoding apparatus 90 configured as described above will be described.

  First, the compressed input bit stream is decoded by the decoding unit 91 using parameters such as an algorithm and frequency used in FM synthesis at the time of encoding, a residual signal, and an applied frequency indicating a frequency to which the residual signal is applied. Decoded into information. Next, the FM synthesizing unit 92 performs FM synthesis using parameters such as the algorithm and frequency used in the FM synthesis decoded by the decoding unit 91, and the synthesized waveform is subjected to time-frequency conversion in the frequency domain. Convert to signal.

  Next, the addition signal calculation unit 93 decodes the residual signal decoded by the decoding unit 91 and the signal output by the FM synthesizing unit 92 by the decoding unit 91 and is indicated by the applied frequency information. Add in the selected frequency band. The added frequency domain signal is converted into a time domain audio signal by the frequency time converter 94.

  Thus, according to the audio signal decoding device in the fourth embodiment of the present invention, the FM composite signal parameter, the residual signal between the audio signal and the FM composite signal, and the residual signal are applied. Since the audio signal can be reproduced by decoding the applied frequency information indicating the generated frequency, the residual signal is encoded only to the extent that the difference from the signal by the FM synthesis substantially affects, An audio signal can be decoded from an encoded signal having a compression rate that is particularly higher than that of an encoded signal based on linear prediction.

  As described above, the audio signal encoding / decoding device according to the present invention predicts frequency domain spectrum data using an FM synthesis method capable of expressing a complex waveform with fewer parameters than linear prediction, and Since a residual signal that is a difference from this signal is obtained and the parameter and the residual signal are encoded, an effect that the encoding process can be performed more efficiently than the process using linear prediction is achieved. And is useful as an audio signal encoding / decoding device or the like for encoding an audio signal.

1 is a block diagram of an audio signal encoding device according to a first embodiment of the present invention. The block diagram of the FM synthetic | combination part in the 1st Embodiment of this invention The block diagram of the FM synthetic | combination part in the 1st Embodiment of this invention The figure which shows the calculation method of the FM synthetic | combination parameter in the 1st Embodiment of this invention The figure which shows the calculation method of the FM synthetic | combination parameter in the 1st Embodiment of this invention Block diagram of an audio signal encoding apparatus according to the second embodiment of the present invention The figure which shows an example of the applied frequency area | region of the applied frequency calculation part in the 2nd Embodiment of this invention. The block diagram of the audio signal decoding apparatus in the 3rd Embodiment of this invention The block diagram of the audio signal decoding apparatus in the 4th Embodiment of this invention Block diagram of a conventional audio signal encoding device Block diagram of a TNS processing unit in a conventional audio signal encoding device

Explanation of symbols

10, 60 Audio signal encoding device 11, 41, 49, 51, 59, 61 Time frequency conversion unit 12, 62, 82, 92 FM synthesis unit
13, 63 Residual signal calculation unit 14, 64 Quantization / coding unit 21, 22, 31, 32 Transmitter 45, 55 FM synthesis unit 46 Analysis unit 57 Comparison unit 68 Applicable frequency calculation unit 80, 90 Audio signal decoding Device 81, 91 Decoding unit 83, 93 Addition signal calculation unit 84, 94 Frequency time conversion unit 110 Auditory psychological model unit 111 Gain control unit 112 MDCT unit 113 TNS processing unit 114 Intensity / coupling unit 115 Prediction unit 116 M / S stereo section 117 normalization coefficient section 118 quantization section 119 noiseless coding section 120 multiplexer section 121 spectrum processing section 122 quantization / coding section 131 MDCT section 132 linear prediction section 133 residual signal calculation section 134 quantization / coding Part

Claims (4)

A time-frequency conversion means for converting a time-domain audio signal into a frequency domain;
FM synthesis means for approximating the signal transformed into the frequency domain by the time-frequency transformation means;
Residual signal calculation means for calculating a difference between the signal converted into the frequency domain by the time frequency conversion means and the frequency domain signal approximated by the FM synthesis means;
An audio signal encoding apparatus comprising: encoding means for encoding the residual signal in the frequency domain calculated by the residual signal calculating means and the parameter used in the FM synthesizing means. An application frequency calculating means for calculating an application range of a frequency for obtaining a residual signal in the residual signal calculating means;
The residual signal calculating means calculates the residual signal of the applied frequency band calculated by the applied frequency calculating means;
The audio signal encoding apparatus according to claim 1, wherein the encoding means encodes the applied frequency band calculated by the applied frequency calculating means as a parameter. Decoding means for decoding the encoded frequency domain residual signal and the parameters used for FM synthesis;
FM synthesis means for performing FM synthesis using the parameters decoded by the decoding means;
An addition signal calculation means for adding the frequency domain residual signal decoded by the decoding means and the frequency domain signal output by the FM synthesis means;
An audio signal decoding apparatus comprising: frequency time conversion means for converting a frequency domain signal generated by the addition signal calculation means into a time domain audio signal. Decoding means for decoding the encoded frequency domain residual signal, parameters used for FM synthesis, and parameters of the frequency band from which the residual signal was calculated;
FM synthesis means for performing FM synthesis using the parameters of FM synthesis decoded by the decoding means;
Addition signal calculation for adding the frequency domain residual signal decoded by the decoding unit and the frequency domain signal output by the FM synthesizing unit in the applied frequency band set by the frequency band parameter Means,
An audio signal decoding apparatus comprising: frequency time conversion means for converting a frequency domain signal generated by the addition signal calculation means into a time domain audio signal.

Images