JP2009541790A - Adaptive high frequency domain encoding and decoding method and apparatus - Google Patents

Abstract

The present invention relates to a method and apparatus for encoding an audio signal, and adaptively encodes a signal provided in a region larger than a predetermined frequency in a time domain or a frequency domain using a signal provided in a region smaller than a predetermined frequency. Or by decoding, the audio signal is encoded using a small number of bits, or the sound quality of the signal corresponding to the high frequency region is not deteriorated even though the audio signal is decoded, so that the coding efficiency is maximized. It can be made.

Description

  The present invention relates to a method and an apparatus for encoding and decoding an audio signal such as a voice signal or a music signal, and more particularly, using a signal or spectrum provided in a low frequency region and using the signal or spectrum provided in the low frequency region. The present invention relates to a method and an apparatus for encoding and decoding a provided signal.

  In general, a signal corresponding to the high frequency region is less important for human recognition as speech than a signal corresponding to the low frequency region. Therefore, in encoding audio signals, if there is a restriction on available bits and coding efficiency must be increased, a signal corresponding to the low frequency region is allocated with many bits and encoded. In comparison, a signal corresponding to a high frequency region is encoded by assigning fewer bits.

  Therefore, there is a need for a method and apparatus that can improve the sound quality recognized by humans to the maximum even when using a small number of bits in encoding a signal corresponding to the high frequency region.

  A technical problem to be solved by the present invention is that a signal provided in a region larger than a preset frequency is adaptively applied in a time domain or a frequency domain using a signal provided in a region smaller than a preset frequency. It is to provide a method and apparatus for encoding and decoding.

  An adaptive high frequency domain encoding apparatus according to the present invention for achieving the above-described problem is obtained by classifying a high frequency signal, which is a signal provided in a region larger than a preset frequency, into a time domain or a frequency domain according to frequency bands. A domain conversion unit that selects and converts any one of them, and a time domain that encodes the frequency band converted into the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency An encoding unit and a frequency domain encoding unit that encodes the frequency band converted into the frequency domain by using an excitation spectrum provided in the low frequency region are provided.

  An adaptive high-frequency region encoding apparatus according to the present invention for achieving the above-described problem is provided in an excitation spectrum provided in a low-frequency region, which is a region smaller than a preset frequency, in a region larger than the preset frequency. A noise information encoding unit that selects a frequency band to be used for encoding a high frequency spectrum that is a provided spectrum, encodes information related to the selected frequency band, and an envelope of the high frequency spectrum And an envelope information encoding unit that extracts and encodes.

  An adaptive high-frequency domain encoding apparatus according to the present invention for achieving the above-described problem is provided in a time domain or a high-frequency signal, which is a signal provided in a region larger than a preset frequency, according to a preset criterion. A domain selector that selects one of the frequency domains, and if the time domain is selected by the domain selector, the high frequency signal is excited in a low frequency region that is smaller than a preset frequency. If the frequency domain is selected by the time domain encoding unit that encodes using a signal and the domain selection unit, the high frequency signal is generated by converting the high frequency signal into the frequency domain, and the high frequency spectrum is generated. A frequency domain encoding unit that encodes using an excitation spectrum provided in the low frequency region

  The adaptive high-frequency domain decoding apparatus according to the present invention for achieving the above-described problem is a domain for determining a domain in which each frequency band provided in a high-frequency region that is a region larger than a preset frequency is encoded. A determination unit, a time domain decoding unit that decodes a frequency band determined to be encoded in the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency, and the frequency domain A frequency domain decoding unit is provided that decodes a frequency band determined to be encoded using an excitation spectrum provided in a low frequency region.

  An adaptive high-frequency decoding apparatus according to the present invention for achieving the above-described problem is provided in a region larger than a preset frequency among excitation spectra provided in a low-frequency region that is smaller than a preset frequency. Using information related to a frequency band used to decode a certain high frequency region, a noise generating unit that generates noise in the high frequency region, and an envelope of a high frequency spectrum provided in the high frequency region An envelope adjustment unit that decodes and adjusts the envelope of the generated noise is provided.

  An adaptive high-frequency domain decoding apparatus according to the present invention for achieving the above-described problem includes a domain determination unit that determines a domain in which a high-frequency region that is a region larger than a preset frequency is encoded, and the domain determination unit If it is determined that the high frequency region is encoded in the time domain, a high frequency signal provided in the high frequency region is converted into an excitation signal provided in a low frequency region that is smaller than a preset frequency. If it is determined by the time domain decoding unit that uses and decodes the high frequency region in the frequency domain, the high frequency spectrum provided in the high frequency region is reduced to the low frequency region. A frequency domain decoding unit that performs decoding using an excitation spectrum provided in the frequency domain is provided.

  An adaptive high frequency domain coding method according to the present invention for achieving the above-described problem is achieved by dividing a high frequency signal, which is a signal provided in a region larger than a preset frequency, into a time domain or a frequency domain. Selecting one of them and converting it, encoding the frequency band converted to the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency, and The method includes encoding the frequency band converted into the frequency domain using an excitation spectrum provided in the low frequency region.

  The adaptive high frequency region encoding method according to the present invention for achieving the above-described problem is applied to a region larger than a preset frequency among excitation spectra provided in a low frequency region that is a region smaller than a preset frequency. Selecting a frequency band to be used for encoding a high frequency spectrum, which is a provided spectrum, encoding information relating to the selected frequency band, and extracting an envelope of the high frequency spectrum; The method includes a step of encoding.

  The adaptive high-frequency domain encoding method according to the present invention for achieving the above-described problem is based on a predetermined criterion for a high-frequency signal, which is a signal provided in a region larger than a predetermined frequency. If one of the frequency domains is selected, and if the time domain is selected in the selecting step, the excitation signal provided in the low frequency region, which is a region smaller than the preset frequency, is used as the high frequency signal. If the frequency domain is selected in the encoding step and the selecting step, the high frequency signal is converted into the frequency domain to generate a high frequency spectrum, and the high frequency spectrum is converted into the low frequency region. And encoding using the provided excitation spectrum.

  The adaptive high-frequency domain decoding method according to the present invention for achieving the above-described problem includes a step of determining a domain in which each frequency band provided in a high-frequency region that is a region larger than a preset frequency is encoded. Decoding the frequency band determined to be encoded in the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency, and determining that the frequency band is encoded in the frequency domain The method includes a step of decoding a frequency band using an excitation spectrum provided in a low frequency region.

  The adaptive high-frequency domain decoding method according to the present invention for achieving the above-described problem is provided in an excitation spectrum provided in a low-frequency region, which is a region smaller than a preset frequency, in a region greater than the preset frequency. Using information related to a frequency band used to decode a certain high frequency region, generating noise in the high frequency region, and decoding an envelope of a high frequency spectrum provided in the high frequency region Adjusting the envelope of the generated noise.

  The adaptive high-frequency domain decoding method according to the present invention for achieving the above-described problem includes a step of determining a domain in which a high-frequency region, which is a region larger than a preset frequency, is encoded, If it is determined that the high frequency region is encoded in the time domain, the high frequency signal provided in the high frequency region is used as the excitation signal provided in the low frequency region that is smaller than the preset frequency. If the high frequency region is determined to be encoded in the frequency domain in the decoding step and the determining step, the high frequency spectrum provided in the high frequency region is excited in the low frequency region. The method includes a step of decoding using a spectrum.

  The recording medium according to the present invention for achieving the above-mentioned problem is to select one of the time domain and the frequency domain for each high frequency signal, which is a signal provided in a region larger than a preset frequency, for each frequency band. Converting the frequency band into the time domain, encoding the frequency band converted into the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency, and the frequency converted into the frequency domain A computer-readable recording medium recording a program for causing a computer to execute a step of encoding a band using an excitation spectrum provided in the low frequency region.

  The recording medium according to the present invention for achieving the above-described problem is a high-frequency spectrum provided in a region higher than a preset frequency among excitation spectra provided in a low-frequency region that is a region smaller than a preset frequency. Selecting a frequency band to be used for encoding a frequency spectrum, encoding information related to the selected frequency band, and extracting and encoding an envelope of the high frequency spectrum by a computer; A computer-readable recording medium that records a program to be executed.

  The recording medium according to the present invention for achieving the above-described problem is either a time domain or a frequency domain depending on a preset reference for a high-frequency signal that is a signal provided in a region larger than a preset frequency. Selecting one, and if the time domain is selected in the selecting step, encoding the high frequency signal using an excitation signal provided in a low frequency region that is a region smaller than a preset frequency If the frequency domain is selected in the selecting step, the high frequency signal is converted into the frequency domain to generate a high frequency spectrum, and the high frequency spectrum is used in the low frequency region. Is a computer-readable recording medium on which a program for causing a computer to execute the encoding step is recorded

  The recording medium according to the present invention for accomplishing the above-described problem is a step of determining a domain in which each frequency band provided in a high frequency region, which is a region larger than a preset frequency, is encoded. Decoding the frequency band determined to be performed using an excitation signal provided in a low frequency region smaller than a preset frequency, and converting the frequency band determined to be encoded in the frequency domain to the low frequency region A computer-readable recording medium recording a program for causing a computer to execute a step of decoding using an excitation spectrum provided.

  The recording medium according to the present invention for accomplishing the above-described problem is a decoding medium that decodes a high-frequency region that is a region larger than a preset frequency among excitation spectra provided in a low-frequency region that is a region smaller than a preset frequency. Generating information in the high frequency region using information related to a frequency band to be used, and decoding an envelope of a high frequency spectrum provided in the high frequency region; A computer-readable recording medium having recorded thereon a program for causing a computer to execute the step of adjusting the envelope.

  The recording medium according to the present invention for achieving the above-described problem is a step of determining a domain in which a high frequency region, which is a region larger than a preset frequency, is encoded, and in the determining step, the high frequency region is a time domain. If determined to be encoded, the step of decoding the high frequency signal provided in the high frequency region using an excitation signal provided in a low frequency region that is a region smaller than a preset frequency, and If it is determined in the determining step that the high frequency region is encoded in the frequency domain, the high frequency spectrum provided in the high frequency region is decoded using the excitation spectrum provided in the low frequency region. It is a computer-readable recording medium having recorded thereon a program for causing a computer to execute the conversion step.

  Hereinafter, an adaptive time / frequency domain encoding method and apparatus according to the present invention will be described in detail with reference to the accompanying drawings.

  FIG. 1A is a block diagram showing an embodiment of an adaptive high frequency domain encoding apparatus according to the present invention. The adaptive high frequency domain encoding apparatus includes a first transform unit 100 and a domain selection unit 105. , Linear prediction unit 110, long interval prediction unit 115, excitation signal encoding unit 120, second transformation unit 125, quantization unit 130, inverse quantization unit 135, second inverse transformation unit 140, storage unit 145, excitation signal decoding The encoding unit 150, the excitation spectrum generation unit 155, the high frequency region encoding unit 160, and the multiplexing unit 165 are included.

  The first conversion unit 100 converts a signal input through the input terminal IN so as to be indicated in a time domain for each predetermined frequency band. As a method of conversion by the first conversion unit 100, there are a QMF (Quadrature Mirror Filter bank), a LOT (Lapped Orthogonal Transform), and the like.

  However, the first conversion unit 100 converts the signal input through the input terminal IN as shown in the time domain or the frequency domain, such as FV-MLT (Frequency Variable Modulated Transformed Transform), according to a predetermined frequency band. Even if the second conversion unit 125 is not provided, the first conversion unit 100 converts the domain selected by the domain selection unit 105 into the domain.

  The domain selection unit 105 encodes, in the time domain, a frequency band signal included in a region smaller than a predetermined frequency among the signals of each frequency band converted by the first conversion unit 100 according to a predetermined reference, Select whether to encode in the frequency domain. Further, the domain selection unit 105 encodes information about the domain in which each frequency band is encoded, and outputs the encoded information to the multiplexing unit 165.

  Here, examples of the predetermined reference include a linear prediction coding gain value, a spectral change between linear prediction filters of adjacent frames, a pitch delay, and a long interval prediction gain value.

  The linear prediction unit 110 performs LPC (Linear Predictive Coding) analysis on the signal provided in the frequency band determined to be encoded in the time domain by the domain selection unit 105, extracts LPC coefficients, and encodes the domain. The first excitation signal is extracted by removing the short-term correlation from the signal provided in the frequency band determined to be encoded in the time domain by the selection unit 105.

  The long interval prediction unit 115 performs long interval prediction (long-term prediction) on the first excitation signal extracted by the linear prediction unit 110, and extracts a second excitation signal. The long interval prediction unit 115 encodes the result of the long interval prediction and outputs the result to the multiplexing unit 165.

  Examples of long interval prediction performed by the long interval prediction unit 115 include a continuity of periodicity, a frequency spectral tilt, and a frame energy. Here, the continuity of the periodicity may be a level in which a change in pitch lag is small and a frame having a high pitch correlation is continuously maintained for a certain period or more. Further, the continuity of the periodicity may be such that the first formant frequency (1st format frequency) is very low and a frame having a high degree of pitch correlation is continuously maintained for a certain period or more.

  The excitation signal encoding unit 120 encodes the second excitation signal extracted by the long interval prediction unit 115.

  The second conversion unit 125 generates a spectrum by converting the signal provided in the frequency band determined to be encoded in the frequency domain by the domain selection unit 105 from the time domain to the frequency domain.

  The quantization unit 130 quantizes the spectrum generated by the second conversion unit 125. The spectrum quantized by the quantization unit 130 is output to the multiplexing unit 165.

  The inverse quantization unit 135 performs inverse quantization on the spectrum quantized by the quantization unit 130.

  The second inverse transformation unit 140 is an inverse process of the transformation performed by the second transformation unit 125, and inversely transforms the spectrum inversely quantized by the inverse quantization unit 135 from the frequency domain to the time domain.

  The storage unit 145 stores the signal inversely converted by the second inverse conversion unit 140. The reason why the signal inversely transformed by the storage unit 145 is stored is that the long section prediction unit 115 performs the long section prediction on the frequency band signal encoded in the time domain from the next frame.

  The excitation signal decoding unit 150 decodes the second excitation signal encoded by the excitation signal encoding unit 120.

  The excitation spectrum generation unit 155 generates an excitation spectrum by performing whitening processing on the spectrum inversely quantized by the inverse quantization unit 135.

  The high frequency region encoding unit 160 uses a signal provided in a frequency band corresponding to a region smaller than a predetermined frequency to generate a signal in a frequency band corresponding to a region larger than the predetermined frequency in the time domain or the frequency domain. Encode adaptively. The signal used by the high frequency domain encoding unit 160 is the second excitation signal decoded by the excitation signal decoding unit 150 when encoded in the time domain, and when encoded in the frequency domain, the excitation spectrum. It is an excitation spectrum generated by the generation unit 155.

  The multiplexing unit 165 is information about the domain in which each frequency band encoded by the domain selection unit 105 is encoded, the LPC coefficient encoded by the linear prediction unit 110, and encoded by the long interval prediction unit 115 As a result of the long interval prediction, the second excitation signal encoded by the excitation signal encoding unit 120, the spectrum quantized by the quantization unit 130, and the result encoded by the encoding unit 160 in the high frequency region Is multiplexed, and a bit stream is generated and output through the output terminal OUT.

  FIG. 1B is a block diagram illustrating an embodiment of a high frequency domain encoder 160 included in the adaptive high frequency domain encoder according to the present invention. , A domain selection unit 170, a linear prediction unit 175, a multiplication unit 180, a gain value encoding unit 185, a noise information encoding unit 190, and an envelope information encoding unit 195.

  The domain selection unit 170 selects whether a signal provided in each frequency band corresponding to a region larger than a predetermined frequency is encoded in the time domain or in the frequency domain.

  In selecting a domain to be encoded by the domain selection unit 170, a frequency band corresponding to a region smaller than a predetermined frequency used for encoding a frequency band corresponding to a region larger than a predetermined frequency is encoded in the time domain. The domain can be selected based on whether it is encoded in the frequency domain. If a predetermined frequency band corresponding to a region smaller than a predetermined frequency used for encoding a predetermined frequency band corresponding to a region larger than the predetermined frequency is encoded in the time domain, the corresponding predetermined frequency is The frequency band corresponding to the large region is selected to be encoded in the time domain, and if the frequency band corresponding to the region smaller than the default frequency used for encoding the frequency band corresponding to the region larger than the predetermined frequency is the frequency When encoded in the domain, a frequency band corresponding to a region larger than the corresponding predetermined frequency is selected to be encoded in the frequency domain.

  The linear prediction unit 175 performs LPC (Linear Predictive Coding) analysis on the signal provided in the frequency band selected by the domain selection unit 170 in the time domain, and extracts LPC coefficients. The LPC coefficients extracted by the linear prediction unit 175 are encoded and output to the multiplexing unit 165 of FIG. 1A through the output terminal OUT1, and used by the decoder to restore the envelope as in the example shown in FIG. 7A. It is done.

  The multiplication unit 180 multiplies the second excitation signal input from the excitation signal decoding unit 150 in FIG. 1A input through the input terminal IN1 and the envelope of the LPC coefficient extracted by the linear prediction unit 175. As an example of the signal multiplied by the multiplication unit 180, there is a signal corresponding to the member number 710 shown in FIG. 7B.

  The gain value encoding unit 185 calculates a gain value that compensates for a mismatch that exists in a low frequency signal that is a signal provided in a region where the signal multiplied by the multiplication unit 180 is smaller than a predetermined frequency, and performs encoding. Turn into. The signal corresponding to the member number 720 of the low frequency signal and the signal corresponding to the member number 710 of the signal multiplied by the multiplying unit 180 as shown in FIG. 7B according to the gain value calculated by the gain value encoding unit 185 Mismatches existing at the boundaries of can be compensated by the decoder as shown in FIG. 7C. Further, the gain value encoded by the gain value encoding unit 185 is output to the multiplexing unit 165 of FIG. 1A through the output terminal OUT2.

  The noise information encoding unit 190 selects the frequency band of the excitation spectrum generated by the excitation spectrum generation unit 155 used to generate noise in the frequency band selected when the domain selection unit 170 encodes in the frequency domain. To encode information about the corresponding frequency band. Information about the frequency band encoded by the noise information encoding unit 190 is output to the multiplexing unit 165 of FIG. 1A through the output terminal OUT3.

  The envelope information encoding unit 195 performs the envelope information of the spectrum provided in the frequency band selected by encoding in the frequency domain by the domain selection unit 170 among the frequency bands provided in a region larger than the predetermined frequency. Extract and encode. The envelope information of the spectrum encoded by the envelope information encoding unit 195 is output to the multiplexing unit 165 of FIG. 1A through the output terminal OUT4.

  As in the embodiment of FIGS. 1A and 1B, a domain to be encoded is first determined, and the present invention is not limited to an open-loop method in which encoding is performed in the determined domain. After encoding both the frequency domain and the time domain, a closed-loop method may be performed in which the encoded results are compared and a better domain is selected.

  FIG. 2A is a block diagram illustrating an embodiment of an adaptive high-frequency domain encoding apparatus according to the present invention. The adaptive high-frequency domain encoding apparatus includes an area dividing unit 200, a linear prediction unit 205, Transformer 210, quantization unit 215, inverse quantization unit 220, inverse transform unit 225, storage unit 230, signal analysis unit 235, long interval prediction unit 240 and switching unit 245, high frequency domain encoding unit 250 and multiplexing Part 255.

  The area dividing unit 200 divides a signal input through the input terminal IN into a low frequency signal provided in an area smaller than a predetermined frequency and a high frequency signal provided in an area larger than the predetermined frequency.

  The linear prediction unit 205 performs LPC (Linear Predictive Coding) analysis on the low frequency signal divided by the region dividing unit 200 to extract LPC coefficients, and the short interval from the low frequency signal divided by the region dividing unit 200 The correlation is removed and the first excitation signal is extracted. Further, the linear prediction unit 205 encodes the extracted LPC coefficient and outputs the encoded LPC coefficient to the multiplexing unit 255.

  The conversion unit 210 converts the first excitation signal extracted by the linear prediction unit 205 from the time domain to the frequency domain to generate an excitation spectrum.

  The quantization unit 215 quantizes the excitation spectrum generated by the conversion unit 210. The excitation spectrum quantized by the quantization unit 215 is output to the multiplexing unit 255.

  The inverse quantization unit 220 inversely quantizes the excitation spectrum quantized by the quantization unit 215.

  The inverse transformation unit 225 is an inverse process of the transformation performed by the transformation unit 210, and inversely transforms the excitation spectrum inversely quantized by the inverse quantization unit 220 from the frequency domain to the time domain to generate a second excitation signal. .

  The storage unit 230 stores the second excitation signal inversely converted by the inverse conversion unit 225. The reason for storing the second excitation signal in the storage unit 230 is to use it for long section prediction in the long section prediction unit 240 for the frequency band signal encoded in the time domain from the next frame.

  The signal analysis unit 235 analyzes the first excitation signal extracted by the linear prediction unit 205 and determines whether or not the long interval prediction unit 240 performs long interval prediction according to the characteristics of the low frequency signal. Here, as an example for the characteristics of a low frequency signal that is a reference for determining whether to perform long section prediction in the signal analysis unit 235, linear prediction coding gain value, spectrum change between linear prediction filters of adjacent frames, pitch delay And a long interval prediction gain value.

  The long interval prediction unit 240 performs long interval prediction (long-term prediction) on the first excitation signal extracted by the linear prediction unit 205 when the signal analysis unit 235 determines to perform long interval prediction. A third excitation signal is extracted.

  Examples of long section prediction performed by the long section prediction unit 240 include the degree of continuity of periodicity, the degree of gradient of the frequency spectrum, and the degree of frame energy. Here, the continuity of the periodicity can be such a degree that a change in pitch lag is small and a frame having a high degree of pitch correlation is continuously maintained for a certain period or more. Further, the continuity of periodicity may be such that the first formant frequency is very low and a frame having a high degree of pitch correlation is continuously maintained for a certain period or longer.

  The switching unit 245 switches the second excitation signal extracted by the long interval prediction unit 240 according to whether the long interval prediction determined by the signal analysis unit 235 is performed.

  The high frequency domain encoding unit 160 encodes a high frequency signal in the frequency domain using an excitation spectrum provided in a region smaller than a predetermined frequency that is inversely quantized by the inverse quantization unit 220.

  The multiplexing unit 255 includes the LPC coefficients encoded by the linear prediction unit 205, the excitation spectrum quantized by the quantization unit 215, the result of the long interval prediction performed by the long interval prediction unit 240, and the code in the high frequency region. The bit stream is generated by including the result encoded by the encoding unit 250, and multiplexed to be output through the output terminal OUT.

  FIG. 2B is a block diagram illustrating an example of a high frequency domain encoder 250 included in the adaptive high frequency domain encoder according to the present invention. The high frequency domain encoder 250 includes: , A noise information encoding unit 260 and an envelope information encoding unit 265.

  The noise information encoding unit 260 is a frequency band used for encoding a high frequency spectrum larger than a predetermined frequency among the excitation spectra inversely quantized by the inverse quantization unit 220 of FIG. 2A input at the input terminal IN1. The information about is encoded. Information about the frequency band used for encoding the high frequency spectrum encoded by the noise information encoding unit 260 is output to the multiplexing unit 255 of FIG. 2A through the output terminal OUT1.

  The envelope information encoding unit 265 receives the high frequency spectrum through the input terminal IN2, extracts the envelope for the high frequency spectrum, and encodes the extracted envelope information. As an example of the envelope information, there is an energy value calculated in a predetermined frequency band unit. The envelope information encoded by the envelope information encoding unit 265 is output to the multiplexing unit 255 of FIG. 2B through the output terminal OUT2.

  FIG. 3A is a block diagram illustrating an embodiment of an adaptive high-frequency domain encoding apparatus according to the present invention. The adaptive high-frequency domain encoding apparatus includes an area dividing unit 300, a linear prediction unit 305, Domain selection unit 310, long interval prediction unit 315, excitation signal encoding unit 320, transformation unit 325, quantization unit 330, inverse quantization unit 335, inverse transformation unit 340, storage unit 345, excitation signal decoding unit 350, high A frequency domain encoding unit 360 and a multiplexing unit 365 are included.

  The area dividing unit 300 converts a signal input through the input terminal IN into a low frequency signal that is a signal provided in an area smaller than a predetermined frequency and a high frequency signal that is a signal provided in an area larger than the predetermined frequency. To divide.

  The linear prediction unit 305 performs LPC analysis on the low frequency signal divided by the region dividing unit 300 to extract LPC coefficients, and removes the short interval correlation from the low frequency signal divided by the region dividing unit 300. A first excitation signal is extracted. The LPC coefficients extracted by the linear prediction unit 305 are encoded and output to the multiplexing unit 365.

  The domain selection unit 310 selects whether to encode the first excitation signal extracted by the linear prediction unit 305 in the time domain or the frequency domain according to a predetermined criterion. Here, examples of the predetermined reference include a linear prediction coding gain value, a spectral change between linear prediction filters of adjacent frames, a pitch delay, and a long interval prediction gain value.

  If the domain selection unit 310 selects to encode the first excitation signal in the time domain, the long interval prediction unit 315 performs long interval prediction on the first excitation signal extracted by the linear prediction unit 305. To extract the second excitation signal.

  Examples of long section prediction performed by the long section prediction unit 315 include a degree of periodic continuity, a degree of gradient of a frequency spectrum, and a degree of frame energy. Here, the continuity of the periodicity can be such a degree that a change in pitch lag is small and a frame having a high degree of pitch correlation is continuously maintained for a certain period or more. Further, the continuity of periodicity may be such that the first formant frequency is very low and a frame having a high degree of pitch correlation is continuously maintained for a certain period or longer.

  The excitation signal encoding unit 320 encodes the second excitation signal extracted by the long interval prediction unit 315.

  If the domain selection unit 310 selects to encode the first excitation signal in the frequency domain, the conversion unit 325 converts the first excitation signal extracted by the linear prediction unit 305 from the time domain to the frequency domain. Generate an excitation spectrum.

  The quantization unit 330 quantizes the excitation spectrum generated by the conversion unit 325. The excitation spectrum quantized by the quantization unit 330 is output to the multiplexing unit 365.

  The inverse quantization unit 335 performs inverse quantization on the excitation spectrum quantized by the quantization unit 330.

  The inverse transform unit 340 is an inverse process of the transform performed by the transform unit 325, and inverse transforms the excitation spectrum inversely quantized by the inverse quantization unit 335 from the frequency domain to the time domain.

  The storage unit 345 stores the third excitation signal inversely converted by the inverse conversion unit 340. The reason for storing the third excitation signal in the storage unit 345 is that the signal provided in the frequency band to be encoded in the time domain at the time of encoding the next frame is used for the long interval prediction in the long interval prediction unit 315. Because.

  The excitation signal decoding unit 350 decodes the second excitation signal encoded by the excitation signal encoding unit 320.

  The high frequency domain encoding unit 360 adaptively encodes a signal corresponding to a region larger than the predetermined frequency in the time domain or the frequency domain using a signal or spectrum corresponding to a region smaller than the predetermined frequency. The signal used by the high frequency domain encoding unit 360 is an excitation signal decoded by the excitation signal decoding unit 350 when being encoded in the time domain, and is a spectrum used by the high frequency domain encoding unit 360. Is an excitation spectrum inversely quantized by the inverse quantization unit 335 when encoding in the frequency domain.

  The multiplexing unit 365 is information about the LPC coefficients extracted by the linear prediction unit 305 and the domain obtained by encoding the low frequency signal selected by the domain selection unit 310 as a result of the long interval prediction performed by the long interval prediction unit 315. By multiplexing including the excitation signal encoded by the excitation signal encoding unit 320, the excitation spectrum quantized by the quantization unit 330 and the result encoded by the encoding unit 360 in the high frequency region, A bit stream is generated and output through the output terminal OUT.

  FIG. 3B is a block diagram illustrating an embodiment of a high frequency domain encoder 360 included in an adaptive high frequency domain encoder according to the present invention, wherein the high frequency domain encoder 360 includes: , A domain selection unit 370, a linear prediction unit 375, a multiplication unit 380, a gain value encoding unit 385, a noise information encoding unit 390, and an envelope information encoding unit 395.

  The domain selection unit 370 inputs the low frequency signal, which is a signal provided in a region smaller than the predetermined frequency selected by the domain selection unit 310 of FIG. 3A input through the input terminal IN1, from a predetermined frequency. A domain for encoding a high-frequency signal that is a signal provided in a large region is selected. If the domain selection unit 310 of FIG. 3A selects to encode the low frequency signal in the frequency domain, the domain selection unit 370 also selects to encode the high frequency signal in the frequency domain. If the domain selection unit 310 in FIG. 3A selects to encode the low frequency signal in the time domain, the domain selection unit 370 also selects to encode the high frequency signal in the time domain.

  If the domain selection unit 370 selects that the high-frequency signal is encoded in the time domain, the linear prediction unit 375 performs LPC analysis on the high-frequency signal input through the input terminal IN2 and extracts LPC coefficients. The LPC coefficients extracted by the linear prediction unit 375 are encoded and output to the multiplexing unit 365 of FIG. 3A through the output terminal OUT1, and are used by the decoder to restore the envelope shown in FIG. 7A.

  The multiplication unit 380 multiplies the excitation signal decoded by the excitation signal decoding unit 350 of FIG. 3A input through the input terminal IN3 and the envelope of the high frequency signal by the LPC coefficient extracted by the linear prediction unit 375. An example of the signal multiplied by the multiplication unit 380 is a signal corresponding to the member number 710 illustrated in FIG. 7B.

  The gain value encoding unit 385 calculates and encodes a gain value that compensates for mismatching in which the signal multiplied by the multiplication unit 380 is present in the low frequency signal. As shown in FIG. 7B, the mismatching present at the boundary between the signal corresponding to the member number 720 of the low frequency signal and the signal corresponding to the member number 710 of the signal multiplied by the multiplier 380 is shown in FIG. 7C. To compensate. Then, the gain value encoded by the gain value encoding unit 385 is output to the multiplexing unit 365 of FIG. 3A through the output terminal OUT2.

  The noise information encoding unit 390 selects a frequency band to be used for decoding the high frequency spectrum by the decoder from the excitation spectrum inversely quantized by the inverse quantization unit 335, and the selected frequency band. Is encoded. Information about the frequency band encoded by the noise information encoding unit 390 is output through the output terminal OUT3.

  The envelope information encoding unit 395 extracts and encodes the envelope information of the high frequency spectrum. Here, as an example of the envelope information, there is an energy value calculated in a predetermined frequency band unit. The envelope information of the high frequency spectrum encoded by the envelope information encoding unit 395 is output to the multiplexing unit 365 of FIG. 3A through the output terminal OUT4.

  As in the embodiment of FIGS. 3A and 3B, the domain to be encoded is further determined, and the present invention is not limited to the open loop method of encoding in the determined domain. It is also possible to implement a closed loop method in which encoding is performed in both the frequency domain and the time domain, and then the encoded results are compared to select a better domain.

  FIG. 4A is a block diagram illustrating an embodiment of an adaptive high frequency domain decoding device according to the present invention, which includes a demultiplexing unit 400, a domain determination unit 405, an excitation. Signal decoding unit 410, long interval synthesis unit 415, linear synthesis unit 420, inverse quantization unit 430, second inverse transformation unit 433, excitation spectrum generation unit 435, high frequency domain decoding unit 440, and first inverse transformation unit 445 Comprising.

  The demultiplexer 400 demultiplexes the bitstream input from the encoding end through the input terminal IN. In the demultiplexer 400, information about the domain in which each frequency band encoded by the encoder is encoded, the LPC coefficient encoded by the encoder, and the long interval prediction encoded by the encoder As a result, the excitation signal encoded by the encoder, the spectrum quantized by the encoder, and the signal or spectrum provided in the low frequency region can be used to decode the high frequency signal. Including and demultiplexing.

  The domain determining unit 405 receives information about a domain in which a frequency band provided in a low frequency region smaller than a predetermined frequency is encoded by the encoder from the demultiplexing unit 400, and each frequency band is encoded. Determine the domain that has been converted.

  The excitation signal decoding unit 410 receives, from the demultiplexing unit 400, the excitation signal encoded by the encoder provided in the frequency band determined by the domain determination unit 405 as the frequency band encoded in the time domain. It is input and decrypted.

  The long interval synthesizer 415 receives from the demultiplexer 400 the result of the long segment prediction performed by the encoder for the frequency band determined by the domain determiner 405 as the frequency band encoded in the time domain. The result of the long-term prediction performed on the excitation signal input and decoded and decoded by the excitation signal decoding unit 410 is synthesized.

  The linear synthesizing unit 420 receives the LPC coefficient of the frequency band determined to be encoded in the time domain by the domain determining unit 405 from the demultiplexing unit 400 and decodes it, and the signal synthesized by the long interval synthesizing unit 415 Are combined with LPC coefficients.

  The inverse quantization unit 430 receives the spectrum provided in the frequency band determined to be encoded in the frequency domain by the domain determination unit 405 from the demultiplexing unit 400 and performs inverse quantization.

  The second inverse transform unit 433 inversely transforms the spectrum dequantized by the inverse quantization unit 430 from the frequency domain to the time domain as an inverse process of the transform performed by the second transform unit 125 of FIG. 1A.

  The excitation spectrum generation unit 435 generates an excitation spectrum by performing whitening processing on the spectrum inversely quantized by the inverse quantization unit 430.

  The high frequency domain decoding unit 440 includes the excitation signal provided in the predetermined frequency band decoded by the excitation signal decoding unit 410 or the excitation spectrum provided in the predetermined frequency band decoded by the excitation spectrum generation unit 435. Is used to decode a high-frequency signal that is a signal provided in a region larger than a predetermined frequency.

  The first inverse transform unit 445 is a reverse process of the conversion performed by the first transform unit 100 in FIG. 1A, and is a signal provided in a predetermined frequency band synthesized by the signal synthesis unit 425 or a second inverse transform unit 433. The inversely transformed signal is synthesized by combining the inversely transformed signal provided in the predetermined frequency band and the high frequency signal decoded by the high frequency domain decoding unit 440 into the time domain, and then output through the output terminal OUT. . There are QMF (Quadrature Mirror Filterbank), LOT (Lapped Orthogonal Transform) and the like as a method of conversion by the first inverse conversion unit 445.

  However, the first inverse transform unit 445 combines the signals shown in the time domain or the frequency domain for each predetermined frequency band, such as FV-MLT (Frequency Varying Modulated Laminated Transform), into the time domain, and performs high frequency domain decoding. This may be implemented without the conversion unit 440 having the separate conversion unit 480 for converting the frequency domain into the time domain.

  FIG. 4B is a block diagram illustrating an embodiment of a high frequency domain decoding unit 440 included in an adaptive high frequency domain decoding apparatus according to the present invention. The adaptive high frequency domain decoding unit 440 includes: , A domain determination unit 450, a linear synthesis unit 455, a multiplication unit 460, a gain value application unit 465, a noise information decoding unit 470, an envelope adjustment unit 475, and an inverse conversion unit 480.

  The domain determination unit 450 determines whether a signal provided in a frequency band included in a high frequency region that is a region larger than a predetermined frequency is encoded in the time domain or in the frequency domain. The domain in which each frequency band is encoded by the domain determining unit 450 is transmitted by transmitting information related to the domain from the encoding end and is input from the demultiplexing unit 400 of FIG. 4A or included in the high frequency region. In addition, a domain obtained by decoding a frequency band included in a low frequency region that is smaller than a predetermined frequency to be used in each frequency band is input from the domain determination unit 405 in FIG. 4A.

  The linear synthesizing unit 455 receives the LPC coefficient for the frequency band encoded in the time domain among the frequency bands included in the high frequency region, and inputs the LPC coefficient from the demultiplexing unit 400 of FIG. 4A through the input terminal IN1 and decodes it. . An envelope like the graph shown in FIG. 7A can be restored by the LPC coefficients decoded by the linear synthesis unit 455.

  The multiplication unit 460 multiplies the excitation signal decoded by the excitation signal decoding unit 150 of FIG. 4A input through the input terminal IN2 and the envelope of the LPC coefficient decoded by the linear synthesis unit 455. An example of the signal multiplied by the multiplication unit 460 is a signal corresponding to the member number 710 shown in FIG. 7B.

  The gain value application unit 465 decodes the gain value input through the input terminal IN3 and applies it to the signal multiplied by the multiplication unit 460. By applying the gain value in the gain value application unit 465, mismatching existing between the decoded low frequency signal and the decoded high frequency signal can be compensated. For example, the high frequency signal multiplied by the multiplication unit 460 causes mismatching at the boundary surface with the low frequency signal as shown in FIG. 7B, but the gain value application unit 465 applies the gain value. As shown in FIG. 7C, the occurrence of mismatch between the low frequency signal and the high frequency signal is prevented. The signal to which the gain value is applied by the gain value application unit 465 is output to the first inverse conversion unit 445 through the output terminal OUT1.

  The noise decoding unit 470 transmits information about the frequency band used for decoding the frequency band spectrum provided in the high frequency region of the excitation spectrum generated by the excitation spectrum generation unit 435 of FIG. 4A through the input terminal IN4. Input from the demultiplexing unit 400 of FIG. 4A and decode. The noise decoding unit 470 uses the decoded information to patch the excitation spectrum provided in the corresponding frequency band to the frequency band determined to be encoded in the frequency domain by the domain determination unit 450 or symmetrically. Noise is generated by folding. For example, the noise decoding unit 535 patches the excitation spectrum provided in the low frequency region illustrated in FIG. 8A to the high frequency region as illustrated in FIG. 8B.

  The envelope adjustment unit 475 receives and decodes the envelope information of the high frequency spectrum encoded by the encoder from the demultiplexing unit 400 of FIG. 4B through the input terminal IN5. Using the envelope information of the high frequency spectrum decoded by the envelope adjustment unit 475, the noise envelope generated by the noise decoding unit 470 is adjusted. For example, the envelope adjustment unit 475 adjusts the envelope generated by the noise decoding unit 470 shown in FIG. 8B using the envelope information of the high frequency spectrum as shown in FIG. 8C. .

  The inverse transformation unit 480 is a reverse process of the transformation performed by the second transformation unit 125 of FIG. 1A, and inversely transforms noise whose envelope has been adjusted by the envelope adjustment unit 475 from the frequency domain to the time domain, thereby increasing the frequency. Generate a signal.

  FIG. 5A is a block diagram illustrating an embodiment of an adaptive high frequency domain decoding device according to the present invention, in which the adaptive high frequency domain decoding device includes a demultiplexing unit 500, an inverse quantization unit 505, It includes an inverse transform unit 510, a long interval synthesis unit 515, a linear synthesis unit 520, and a high frequency domain decoding unit 525.

  The demultiplexer 500 demultiplexes the bitstream input from the encoding end through the input terminal IN. In the demultiplexing unit 500, the LPC coefficient encoded by the encoder, the excitation spectrum quantized by the encoder, the result of the long interval prediction performed by the encoder, and the low frequency of the region smaller than the predetermined frequency Information that can decode a high-frequency signal provided in a region larger than a predetermined frequency is demultiplexed using an excitation spectrum provided in the frequency region.

  The inverse quantization unit 505 receives the excitation spectrum provided in the low frequency region quantized by the encoder from the inverse multiplexing unit 500 and performs inverse quantization.

  The inverse transform unit 510 is an inverse process of the transform performed by the transform unit 210 in FIG. 2A, and generates an excitation signal by inverse transforming the excitation spectrum inversely quantized by the inverse quantization unit 505 from the frequency domain to the time domain. To do.

  The long interval synthesizer 515 receives the result of the long interval prediction performed on the excitation signal provided in the low frequency region by the encoder, is decoded from the demultiplexer 500, and is generated by the inverse transformer 510. The result of the long-term prediction performed on the excited excitation signal is selectively synthesized. When combining the result of the long section prediction performed by the long section combining unit 515, the result of the long section prediction performed by the encoder is combined only with the transmitted excitation signal.

  The linear synthesis unit 520 receives the LPC coefficients from the demultiplexing unit 500 and decodes them. After decoding the LPC coefficients, the linear synthesis unit 520 synthesizes the LPC coefficients with the excitation signal generated by the inverse transform unit 510 when the result of the long interval prediction performed by the long interval synthesis unit 515 is not synthesized. When the result of the long section prediction performed by the long section combining unit 515 is combined, the LPC coefficient is combined with the signal combined by the long section combining unit 515. The signal synthesized by the linear synthesizing unit 520 is obtained by restoring the low frequency signal that is finally provided in the low frequency region.

  The high frequency domain decoding unit 525 decodes the high frequency signal using the excitation spectrum provided in the low frequency domain inversely quantized by the inverse quantization unit 505.

  The region synthesizing unit 530 synthesizes the low frequency signal restored by the linear synthesizing unit 520 and the high frequency signal decoded by the high frequency region decoding unit 525, and outputs the synthesized signal through the output terminal OUT.

  FIG. 5B is a block diagram illustrating an example of a high frequency domain decoding unit 525 included in an adaptive high frequency domain decoding device according to the present invention, in which the high frequency domain decoding unit 525 is configured to perform noise decoding. A conversion unit 535, an envelope adjustment unit 540, and an inverse conversion unit 545.

  The noise decoding unit 535 demultiplexes information on a frequency band used for encoding a high frequency spectrum larger than a predetermined frequency among excitation spectra provided in a low frequency region which is a region smaller than the predetermined frequency. The data is input from the unit 500 through the input terminal IN1 and decoded. The noise decoding unit 535 selects a predetermined excitation spectrum to be used from among the excitation spectra inversely quantized by the inverse quantization unit 505 through the input terminal IN1 ′ based on the decoded information, and sets the corresponding excitation spectrum as a default. Noise is generated by patching in a region larger than the frequency of or by folding symmetrically. For example, the noise decoding unit 535 patches the excitation spectrum provided in the low frequency region shown in FIG. 8A to the high frequency region as shown in FIG. 8B.

  The envelope adjustment unit 540 receives and decodes the envelope information of the high frequency spectrum encoded by the encoder from the demultiplexing unit 500 of FIG. 5A through the input terminal IN2. The envelope adjustment unit 540 adjusts the noise envelope generated by the noise decoding unit 535 using the decoded envelope information of the high frequency spectrum. For example, the envelope adjustment unit 540 adjusts the envelope generated by the noise decoding unit 535 shown in FIG. 8B using the envelope information of the high frequency spectrum as shown in FIG. 8C. .

  The inverse transformation unit 545 is an inverse process of the transformation performed by the transformation unit 210 in FIG. 2A, and inversely transforms the noise whose envelope has been adjusted by the envelope adjustment unit 540 from the frequency domain to the time domain to convert a high frequency signal. Generate. The high frequency signal generated by the inverse conversion unit 545 is output to the region synthesis unit 530 of FIG. 2B through the output terminal OUT1.

  FIG. 6A is a block diagram illustrating an embodiment of a high frequency domain decoding apparatus according to the present invention, which includes a demultiplexing unit 600, a domain determination unit 605, and an excitation signal decoding unit. 610, a long interval synthesis unit 615, an inverse quantization unit 620, an inverse transformation unit 625, a linear synthesis unit 630, a high frequency domain decoding unit 635, and a domain synthesis unit 640.

  The demultiplexing unit 600 demultiplexes the bitstream input from the encoding end through the input terminal IN. In the demultiplexing unit 600, information on the domain in which the low frequency signal selected by the encoder is encoded, the LPC coefficient encoded by the encoder, the result of the long interval prediction performed by the encoder, the code Information that decodes the high frequency signal using the excitation signal encoded by the encoder, the excitation spectrum quantized by the encoder, and the low frequency signal or the low frequency spectrum provided in a region smaller than the predetermined frequency Is demultiplexed.

  The domain determination unit 605 receives and decodes domain information in which a low frequency region corresponding to a region smaller than a predetermined frequency is encoded by the encoder from the demultiplexing unit 600 and decodes the low frequency region in the time domain. It is determined whether it has been encoded or in the frequency domain.

  If the domain determination unit 605 determines that the low frequency region is encoded in the time domain, the excitation signal decoding unit 610 transmits the low frequency region excitation signal encoded by the encoder from the demultiplexing unit 400. It is input and decrypted.

  The long interval synthesizer 615 receives a result of the long interval prediction performed by the encoder from the low frequency signal, which is a signal provided in a region smaller than the predetermined frequency, from the demultiplexer 600 and decodes the result. The excitation signal decoded by the excitation signal decoding unit 610 is combined with the result of the long interval prediction.

  If the domain determining unit 605 determines that the low frequency region is encoded in the frequency domain, the inverse quantization unit 620 receives the excitation spectrum quantized by the encoder from the demultiplexing unit 600 and performs inverse quantization. Turn into.

  The inverse transformation unit 625 is an inverse process of the transformation performed by the transformation unit 325 in FIG. 3A, and generates an excitation signal by inversely transforming the excitation spectrum inversely quantized by the inverse quantization unit 620 from the frequency domain to the time domain. To do.

  The linear synthesis unit 630 receives the LPC coefficients of the low frequency signal from the demultiplexing unit 600 and decodes them, and decodes them into the excitation signal synthesized by the long interval synthesis unit 615 or the excitation signal generated by the inverse transformation unit 625. Synthesized LPC coefficients. The signal synthesized by the linear synthesizing unit 630 is obtained by restoring the low-frequency signal that is finally provided in the low-frequency region.

  The high frequency domain decoding unit 635 decodes the high frequency signal using the excitation spectrum inversely quantized by the inverse quantization unit 620 or the excitation signal decoded by the excitation signal decoding unit 610. If the low frequency region is encoded in the time domain, the high frequency region decoding unit 635 uses the excitation spectrum inversely quantized by the inverse quantization unit 620 to decode the high frequency signal. If the low frequency region is encoded in the frequency domain, the high frequency region decoding unit 635 uses the excitation signal decoded by the excitation signal double flower unit 610 to decode the high frequency signal.

  The region synthesizer 640 synthesizes the low frequency signal restored by the linear synthesizer 630 and the high frequency signal decoded by the high frequency region decoder 635 and outputs the synthesized signal through the output terminal OUT.

  FIG. 6B is a block diagram illustrating an embodiment of a high frequency domain decoding unit 635 included in the high frequency domain decoding device according to the present invention. The high frequency domain decoding unit 635 includes a domain determination unit 645. , A linear synthesis unit 650, a multiplication unit 655, a noise decoding unit 665, an envelope adjustment unit 670, and an inverse conversion unit 675.

  The domain determination unit 645 may decode a high frequency region corresponding to a region larger than the predetermined frequency in the time domain relative to a domain in which a low frequency region corresponding to a region smaller than the predetermined frequency is encoded, To determine whether to decrypt.

  If the domain determining unit 645 determines that the high frequency region is to be decoded in the time domain, the linear synthesizing unit 650 demultiplexes the LPC coefficients for the high frequency signal provided in the region larger than the predetermined frequency in FIG. 6A. The data is input from the conversion unit 600 through the input terminal IN1 and decoded. An envelope as shown in the graph of FIG. 7A can be restored by the LPC coefficients decoded by the linear synthesis unit 650.

  The multiplication unit 655 multiplies the excitation signal decoded by the excitation signal decoding unit 610 of FIG. 6A input through the input terminal IN2 by the envelope of the LPC coefficient decoded by the linear synthesis unit 650. As an example of the signal multiplied by the multiplier 655, there is a signal corresponding to the member number 710 shown in FIG. 7B.

  The gain value applying unit 660 decodes the gain value input through the input terminal IN3 from the demultiplexing unit 600 of FIG. 6A and applies the decoded signal to the signal multiplied by the multiplying unit 655. By applying the gain value in the gain value application unit 660, mismatching existing between the low frequency signal restored by the linear synthesis unit 630 in FIG. 6A and the decoded high frequency signal can be compensated. For example, the high frequency signal multiplied by the multiplication unit 655 causes mismatching at the boundary surface with the low frequency signal as shown in FIG. 7B, but the gain value application unit 660 applies the gain value. As shown in FIG. 7C, occurrence of mismatch between the low frequency signal and the high frequency signal is prevented. The signal to which the gain value is applied by the gain value application unit 660 is output to the region synthesis unit 640 of FIG. 6A through the output terminal OUT1.

  If the domain determination unit 645 determines to decode the high frequency region in the frequency domain, the noise decoding unit 665 transmits the excitation spectrum inversely quantized by the inverse quantization unit 620 of FIG. 6A through the input terminal IN4. A spectrum is generated by being input and patched in a high frequency region or by being folded symmetrically. For example, the excitation spectrum shown in FIG. 8A is patched to the high frequency region as shown in FIG. 8B.

  The envelope adjustment unit 670 receives and decodes the envelope information of the high frequency spectrum encoded by the encoder from the demultiplexing unit 600 of FIG. 6A through the input terminal IN5. The envelope adjustment unit 670 adjusts the envelope of the spectrum generated by the noise decoding unit 665 using the decoded envelope information of the high frequency spectrum. For example, the envelope adjustment unit 670 uses the envelope information of the high frequency spectrum for the spectrum generated by the noise decoding unit 665 shown in FIG. 8B to change the envelope as shown in FIG. 8C. Adjust.

  The inverse transformation unit 675 is a reverse process of the transformation performed by the transformation unit 325 of FIG. 3A, and inversely transforms the spectrum whose envelope has been adjusted by the envelope adjustment unit 670 from the frequency domain to the time domain to convert a high frequency signal. Generate.

  FIG. 9A is a flowchart illustrating an embodiment of an adaptive high frequency domain encoding method according to the present invention.

  First, the input signal is converted according to a predetermined frequency band as shown in the time domain (operation 900). There are QMF, LOT, etc. as a method of conversion in the 900th stage.

  However, in step 900, an input signal such as FV-MLT is converted in the time domain or frequency domain according to a predetermined frequency band, so that it is selected in step 905 without performing step 925. In step 900, the converted domain may be implemented.

  Select whether to encode in the time domain or in the frequency domain a frequency band signal included in a region smaller than a predetermined frequency among the signals in each frequency band converted in step 900. (Step 905). Here, examples of the predetermined reference include a linear prediction coding gain value, a spectral change between linear prediction filters of adjacent frames, a pitch delay, and a long interval prediction gain value.

  In step 905, LPC analysis is performed on the signal provided in the frequency band determined to be encoded in the time domain to extract and encode LPC coefficients, and in step 905, it is determined to encode in the time domain. The first excitation signal is extracted by removing the short-term correlation from the signal provided in the frequency band (operation 910).

  A long interval prediction is performed on the first excitation signal extracted in operation 910, and a second excitation signal is extracted (operation 915).

  Examples of long interval prediction performed in step 915 include the degree of continuity of periodicity, the degree of gradient of the frequency spectrum, and the degree of frame energy. Here, the continuity of the periodicity can be such a degree that a change in pitch lag is small and a frame having a high degree of pitch correlation is continuously maintained for a certain period or more. Further, the continuity of periodicity may be such that the first formant frequency is very low and a frame having a high degree of pitch correlation is continuously maintained for a certain period or longer.

  The second excitation signal extracted in operation 915 is encoded (operation 920). In operation 905, a signal provided in a frequency band determined to be encoded in the frequency domain is converted from the time domain to the frequency domain to generate a spectrum (operation 925). The spectrum generated in operation 925 is quantized (operation 930). The spectrum quantized in operation 930 is inversely quantized (operation 935). As a reverse process of the conversion performed in operation 925, the spectrum inversely quantized in operation 935 is inversely converted from the frequency domain to the time domain (operation 940).

  The signal inversely transformed by the second inverse transform unit 140 is stored (operation 945). The reason why the signal inversely transformed in operation 945 is stored is that the signal provided in the frequency band to be encoded in the time domain from the next frame is used for the long interval prediction in operation 915.

  The second excitation signal encoded in operation 920 is decoded (operation 950). The spectrum dequantized in operation 935 is whitened to generate an excitation spectrum (operation 955).

  A signal in a frequency band corresponding to a region larger than the predetermined frequency is adaptively encoded in a time domain or a frequency domain using a signal provided in a frequency band corresponding to a region smaller than the predetermined frequency (operation 960). ). The signal used in operation 960 is the second excitation signal decoded in operation 950 when encoded in the time domain, and the excitation spectrum generated in operation 955 when encoded in the frequency domain. is there.

  Information about the domain in which each frequency band encoded in operation 905 is encoded, the LPC coefficient encoded in operation 910, and the long interval prediction encoded in operation 915 are performed. A bitstream is generated by multiplexing the second excitation signal encoded in operation 920, the spectrum quantized in operation 930, and the result encoded in operation 960 (operation 965). .

  FIG. 9B is a flowchart illustrating an example of operation 960 included in the adaptive high frequency domain encoding method according to the present invention.

  First, it is selected whether to encode a signal provided in each frequency band corresponding to a region larger than a predetermined frequency in the time domain or in the frequency domain (operation 970).

  In selecting a domain to be encoded in operation 970, a frequency band corresponding to a region smaller than a predetermined frequency used for encoding a frequency band corresponding to a region larger than a predetermined frequency is encoded in the time domain. A domain can be selected based on whether it is encoded in the frequency domain. If a predetermined frequency band corresponding to a region smaller than a predetermined frequency used for encoding a predetermined frequency band corresponding to a region larger than the predetermined frequency is encoded in the time domain, the corresponding predetermined frequency is The frequency band corresponding to the large region is selected to be encoded in the time domain, and if the frequency band corresponding to the region smaller than the default frequency used for encoding the frequency band corresponding to the region larger than the predetermined frequency is selected. When encoded in the frequency domain, a frequency band corresponding to a region larger than the corresponding predetermined frequency is selected to be encoded in the frequency domain.

  In operation 970, an LPC coefficient is extracted by performing LPC analysis on a signal included in a frequency band selected when encoding in the time domain (operation 975). The LPC coefficients extracted in operation 975 are used by the decoder to restore the envelope as in the example shown in FIG. 7A.

  The multiplier 180 multiplies the second excitation signal decoded in operation 950 of FIG. 9A by the envelope of the LPC coefficient extracted in operation 975 (operation 980). An example of the signal multiplied in operation 980 is a signal corresponding to the member number 710 illustrated in FIG. 7B.

  A gain value that compensates for mismatching existing in a low-frequency signal that is a signal provided in a region where the signal multiplied in operation 980 is smaller than a predetermined frequency is calculated and encoded (operation 985). The boundary between the signal corresponding to the member number 720 of the low frequency signal and the signal corresponding to the member number 710 of the signal multiplied in step 980, as shown in FIG. 7B, according to the gain value calculated in step 985. Can be compensated with a decoder as shown in FIG. 7C.

  In step 970, when the frequency domain is encoded, the frequency band of the excitation spectrum generated in step 955 is used to generate noise in the selected frequency band, and information on the corresponding frequency band is encoded. (Step 990).

  Out of frequency bands provided in a region larger than a predetermined frequency, if encoding is performed in the frequency domain in operation 970, the envelope information of the spectrum provided in the selected frequency band is extracted and encoded (operation 995). ).

  The domain to be encoded as in the embodiment of FIGS. 9A and 9B is not limited to the open loop method in which the domain to be encoded is determined in advance and encoded in the determined domain. It is also possible to implement a closed loop method in which encoding is performed in both the frequency domain and the time domain, and then the encoded results are compared to select a better domain.

  FIG. 10A is a flowchart illustrating an embodiment of an adaptive high frequency domain encoding method according to the present invention.

  First, the input signal is divided into a low frequency signal provided in a region smaller than a predetermined frequency and a high frequency signal provided in a region larger than the predetermined frequency (operation 1000).

  LPC analysis is performed on the low frequency signal divided in step 1000 to extract LPC coefficients, and the first excitation signal is extracted by removing short-term correlation from the low frequency signal divided in step 1000 ( (Step 1005).

  The first excitation signal extracted in operation 1005 is converted from the time domain to the frequency domain to generate an excitation spectrum (operation 1010).

  The excitation spectrum generated in operation 1010 is quantized (operation 1015). The excitation spectrum quantized in operation 1015 is inversely quantized (operation 1020). In the reverse process of the conversion performed in operation 1010, the excitation spectrum inversely quantized in operation 1020 is inversely converted from the frequency domain to the time domain to generate a second excitation signal (operation 1025).

  The second excitation signal inversely transformed in operation 1025 is stored (operation 1030). The reason why the second excitation signal is stored in operation 1030 is that the signal provided in the frequency band to be encoded in the time domain is used for long section prediction in operation 1040 when encoding the next frame. is there.

  The first excitation signal extracted in operation 1005 is analyzed to determine whether the long interval prediction unit 240 performs long interval prediction according to the characteristics of the low frequency signal (operation 1035). Examples of the characteristics of a low frequency signal that is a criterion for determining whether or not to perform long interval prediction in operation 1035 include linear prediction coding gain values, spectral changes between linear prediction filters of adjacent frames, pitch delay, and length. There is an interval prediction gain value and the like.

  If it is determined in step 1035 to perform long section prediction, long section prediction is performed on the first excitation signal extracted in step 1005 to extract a third excitation signal (step 1040).

  Examples of long interval prediction performed in step 1040 include the degree of continuity of periodicity, the degree of gradient of the frequency spectrum, and the degree of frame energy. Here, the continuity of the periodicity can be such a degree that a change in pitch lag is small and a frame having a high degree of pitch correlation is continuously maintained for a certain period or more. Further, the continuity of periodicity may be such that the first formant frequency is very low and a frame having a high degree of pitch correlation is continuously maintained for a certain period or longer.

  A high frequency signal is encoded in the frequency domain using an excitation spectrum provided in a region smaller than the predetermined frequency inversely quantized in operation 1020 (operation 1050).

  A bitstream including the LPC coefficients encoded in operation 1005, the excitation spectrum quantized in operation 1015, the results of long-term prediction performed in operation 1040, and the results encoded in operation 1050 It multiplexes by producing | generating (step 1055).

  FIG. 10B is a flowchart illustrating an embodiment for operation 1050 included in the adaptive high frequency domain encoding method according to the present invention.

  First, information about a frequency band used for encoding a high frequency spectrum larger than a predetermined frequency among the excitation spectrum inversely quantized in step 1020 of FIG. 10A is encoded (step 1060).

  After operation 1060, a high frequency spectrum is input to extract an envelope for the high frequency spectrum, and the extracted envelope information is encoded (operation 1065). As an example of the envelope information, there is an energy value calculated in a predetermined frequency band unit.

  FIG. 11A is a flowchart illustrating an embodiment of an adaptive high frequency domain encoding method according to the present invention.

  First, the input signal is divided into a low frequency signal, which is a signal provided in a region smaller than a predetermined frequency, and a high frequency signal, which is a signal provided in a region higher than a predetermined frequency (step 1100).

  LPC analysis is performed on the low frequency signal divided in step 1100 to extract LPC coefficients, and short interval correlation is removed from the low frequency signal divided in step 1100 to extract a first excitation signal ( Step 1105).

  The first excitation signal extracted in operation 1105 is selected to be encoded in the time domain or in the frequency domain according to a predetermined criterion (operation 1110). Here, examples of the predetermined reference include a linear prediction coding gain value, a spectral change between linear prediction filters of adjacent frames, a pitch delay, and a long interval prediction gain value.

  If it is selected in step 1110 to encode the first excitation signal in the time domain, the first excitation signal extracted in step 1105 is subjected to long interval prediction to extract a second excitation signal ( Step 1115).

  Examples of long interval prediction performed in step 1115 include the degree of periodic continuity, the degree of gradient of the frequency spectrum, and the degree of frame energy. Here, the continuity of the periodicity can be such a degree that a change in pitch lag is small and a frame having a high degree of pitch correlation is continuously maintained for a certain period or more. In addition, the continuity of the periodicity may be such that a frame having a very low first formant frequency and a high degree of pitch correlation is continuously maintained for a certain period or more.

  The second excitation signal extracted in operation 1115 is encoded (operation 1120).

  If it is selected in step 1110 to encode the first excitation signal in the frequency domain, the first excitation signal extracted in step 1105 is converted from the time domain to the frequency domain to generate an excitation spectrum (step 1). 1125).

  The excitation spectrum generated in operation 1125 is quantized (operation 1130). The excitation spectrum quantized in operation 1130 is inversely quantized (operation 1135). In the reverse process of the transformation performed in operation 1125, the excitation spectrum inversely quantized in operation 1135 is inversely transformed from the frequency domain to the time domain (operation 1140).

  The third excitation signal inversely transformed in operation 1140 is stored (operation 1145). The reason for storing the third excitation signal in operation 1145 is that the third excitation signal is used in the long-term prediction performed in operation 1115 for the frequency band signal encoded in the time domain from the next frame. It is. The second excitation signal encoded in operation 1120 is decoded (operation 1150).

  A signal corresponding to a region larger than a predetermined frequency is adaptively encoded in a time domain or a frequency domain using a signal or spectrum corresponding to a region smaller than a predetermined frequency (operation 1160). The signal used in operation 1160 is the excitation signal decoded in operation 1150 when encoded in the time domain, and the spectrum used in operation 1160 is processed in operation 1135 when encoded in the frequency domain. It is an excitation spectrum inversely quantized with.

  LPC coefficients extracted in step 1105, information about the domain in which the low-frequency signal selected in step 1110 is encoded, results of long interval prediction performed in step 1115, and excitation encoded in step 1120 A bit stream is generated by multiplexing the signal, the excitation spectrum quantized in operation 1130, and the result encoded in operation 1160 (operation 1165).

  FIG. 11B is a flowchart illustrating an example of operation 1160 included in the adaptive high frequency domain encoding method according to the present invention.

  First, a high-frequency signal that is a signal provided in a region larger than the predetermined frequency by a domain that encodes a low-frequency signal that is a signal provided in a region smaller than the predetermined frequency selected in step 1110 of FIG. 11A. A domain to be encoded is selected (operation 1170). If the low frequency signal is selected to be encoded in the frequency domain in step 1110 of FIG. 11A, the high frequency signal is selected to be encoded in the frequency domain even in step 1170. If the low frequency signal is selected to be encoded in the time domain in step 1110 of FIG. 11A, the high frequency signal is selected to be encoded in the time domain even in step 1170.

  If the high frequency signal is selected to be encoded in the time domain in operation 1170, LPC analysis is performed on the high frequency signal to extract LPC coefficients (operation 1175). The LPC coefficients extracted in operation 1175 are used to restore the envelope shown in FIG. 7A by the decoder.

  The second excitation signal decoded in operation 1150 of FIG. 11A is multiplied by the envelope of the high frequency signal based on the LPC coefficients extracted in operation 1175 (operation 1180). An example of the signal multiplied in operation 1180 is a signal corresponding to the member number 710 shown in FIG. 7B.

  In operation 1185, a gain value that compensates for mismatching in which the signal multiplied in operation 1180 is present in the low frequency signal is calculated. As shown in FIG. 7B, FIG. 7C shows the mismatching existing at the boundary between the signal corresponding to the member number 720 of the low frequency signal and the signal corresponding to the member number 710 of the signal multiplied in the 1180 stage. Compensate as was.

  Of the excitation spectrum inversely quantized in operation 1135, the decoder selects a frequency band to be used for decoding a high frequency spectrum and encodes information about the selected frequency band (1190). Stage).

  After step 1190, envelope information of the high frequency spectrum is extracted and encoded (step 1195). Here, as an example of the envelope information, there is an energy value calculated in a predetermined frequency band unit.

  As in the embodiment of FIGS. 11A and 11B, the domain to be encoded is determined first, and the present invention is not limited to the open loop method in which encoding is performed with the determined domain. It is also possible to implement a closed loop method in which encoding is performed in both the frequency domain and the time domain, and then the encoded results are compared to select a better domain.

  FIG. 12A is a flowchart illustrating an embodiment of an adaptive high frequency domain decoding method according to the present invention.

  First, the bitstream input from the encoding end is demultiplexed (operation 1200). In operation 1200, information on the domain in which each frequency band encoded by the encoder is encoded, LPC coefficients encoded by the encoder, and long-interval prediction encoded by the encoder are performed. As a result, it includes an excitation signal encoded by the encoder, a spectrum quantized by the encoder, and a signal or spectrum provided in the low frequency region, so that a high frequency signal can be decoded. Demultiplex.

  In step 1205, information about a domain in which a frequency band provided in a low frequency region smaller than a predetermined frequency is encoded by an encoder is input to determine a domain in which each frequency band is encoded (operation 1205). ).

  In operation 1205, an excitation signal encoded by an encoder provided in a frequency band determined to be a frequency band encoded in the time domain is decoded (operation 1210).

  In step 1205, the result of the long-term prediction performed by the encoder for the frequency band determined to be a frequency band encoded in the time domain is decoded, and the excitation decoded in step 1210 is decoded. The result of the long interval prediction is combined with the signal (operation 1215).

  In operation 1205, the LPC coefficient of the frequency band determined to be encoded in the time domain is decoded, and the LPC coefficient is combined with the signal combined in operation 1215 (operation 1220).

  A spectrum provided in a frequency band determined to be encoded in the frequency domain in operation 1205 is inversely quantized (operation 1230).

  In FIG. 9A, the inverse process of the transform performed at step 1225, the spectrum inversely quantized at step 1230 is inverse transformed from the frequency domain to the time domain (step 1233).

  The spectrum inversely quantized in operation 1230 is whitened to generate an excitation spectrum (operation 1235).

  A signal provided in a region larger than a predetermined frequency using the excitation signal provided in the predetermined frequency band decoded in operation 1210 or the excitation spectrum provided in the predetermined frequency band decoded in operation 1235. Are decoded (operation 1240).

  9A is a reverse process of the conversion performed in operation 900 of FIG. 9A, the signal provided in the predetermined frequency band synthesized in operation 1225 or the signal provided in the predetermined frequency band inversely converted in operation 1233 and The high frequency signal decoded in operation 1240 is combined into a time domain signal and inversely converted (operation 1245). There are QMF, LOT, etc. as a method of conversion in the 1245th stage.

  However, in operation 1245, a signal shown in the time domain or frequency domain for each predetermined frequency band, such as FV-MLT, is synthesized into a time domain signal, and in operation 1240, the frequency domain is further converted from the time domain to the time domain. This can be done without performing step 1280.

  FIG. 12B is a flowchart illustrating an example of operation 1240 included in the adaptive high frequency domain decoding method according to the present invention.

  First, it is determined whether a signal provided in a frequency band included in a higher frequency region than a predetermined frequency is encoded in the time domain or in the frequency domain (operation 1250). In operation 1250, the domain in which each frequency band is encoded is known by transmitting information related to the domain from the encoding end, or from a predetermined frequency to be used in each frequency band included in the high frequency region. A domain where a frequency band included in a low frequency region of a small region is decoded can be seen from step 1205 in FIG. 12A.

  Among the frequency bands included in the high frequency region, LPC coefficients for the frequency band encoded in the time domain are decoded (operation 1255). The envelope as shown in the graph of FIG. 7A may be restored using the LPC coefficients decoded in operation 1255.

  The excitation signal decoded in operation 1250 of FIG. 12A is multiplied by the envelope of the LPC coefficients decoded in operation 1255 (operation 1260). An example of the signal multiplied in operation 1260 is a signal corresponding to the member number 710 shown in FIG. 7B.

  After operation 1260, the gain value is decoded and applied to the signal multiplied in operation 1260 (operation 1265). In operation 1265, a gain value may be applied to compensate for a mismatch that exists between the decoded low frequency signal and the decoded high frequency signal. For example, as shown in FIG. 7B, the high frequency signal multiplied in operation 1260 may be mismatched at the interface with the low frequency signal, but by applying a gain value in operation 1265, As shown in FIG. 7C, the occurrence of mismatch between the low frequency signal and the high frequency signal is prevented.

  Of the excitation spectrum generated in operation 1235 of FIG. 12A, information on the frequency band used to decode the spectrum of the frequency band provided in the high frequency region is decoded (operation 1270). In operation 1270, the decoded information is used to patch the excitation spectrum provided in the corresponding frequency band to the frequency band determined to be encoded in the frequency domain in operation 1250, or symmetrically. Noise is generated by folding. For example, in operation 1270, the excitation spectrum provided in the low frequency region shown in FIG. 8A is patched in the high frequency region as shown in FIG. 8B.

  The envelope information of the high frequency spectrum encoded by the encoder is decoded (operation 1275). The envelope of the noise generated in operation 1270 is adjusted using the envelope information of the high frequency spectrum decoded in operation 1275. For example, in operation 1275, the envelope is adjusted as shown in FIG. 8C using the envelope information of the high frequency spectrum with respect to the noise generated in operation 1270 shown in FIG. 8B.

  9A is a reverse process of the conversion performed in operation 925 of FIG. 9A, and the high-frequency signal is generated by performing inverse conversion of the noise whose envelope has been adjusted in operation 1275 from the frequency domain to the time domain (operation 1280).

  FIG. 13A is a flowchart illustrating an embodiment of an adaptive high frequency domain decoding method according to the present invention.

  First, the bitstream input from the encoding end is demultiplexed (operation 1300). In step 1300, the LPC coefficients encoded by the encoder, the excitation spectrum quantized by the encoder, the result of the long-term prediction performed by the encoder, and the low frequency region smaller than the predetermined frequency are set. Information that can be decoded from a high frequency signal provided in a region larger than a predetermined frequency is demultiplexed using the provided excitation spectrum.

  The excitation spectrum provided in the low frequency region quantized by the encoder is inversely quantized (step 1305).

  In the reverse process of the transformation performed in step 1010 of FIG. 10A, the excitation spectrum inversely quantized in step 1305 is inversely transformed from the frequency domain to the time domain to generate an excitation signal (step 1310).

  The encoder decodes the result of the long interval prediction performed on the excitation signal provided in the low frequency region, and selectively selects the result of the long interval prediction performed on the excitation signal generated in operation 1310. Synthesize (step 1315). When combining the result of the long interval prediction performed in operation 1315, the result of the long interval prediction performed by the encoder is combined only with the excitation signal transmitted.

  The LPC coefficient is decoded (operation 1320). In operation 1320, after decoding the LPC coefficients, if the result of the long interval prediction in operation 1315 is not combined, the LPC coefficients are combined with the excitation signal generated in operation 1310. When combining the result of the long interval prediction, the LPC coefficient is combined with the signal combined in operation 1315. The signal synthesized in operation 1320 is obtained by restoring the low-frequency signal that is finally provided in the low-frequency region.

  The high frequency signal is decoded using the excitation spectrum provided in the low frequency region inversely quantized in operation 1305 (operation 1325). The low-frequency signal restored in operation 1320 and the high-frequency signal decoded in operation 1325 are combined (operation 1330).

  FIG. 13B is a flowchart illustrating an example of operation 1325 included in the adaptive high frequency domain decoding method according to the present invention.

  First, information about a frequency band used for encoding a high frequency spectrum larger than a predetermined frequency among the excitation spectra provided in a low frequency region which is a region smaller than the predetermined frequency is decoded (operation 1335). . In operation 1335, a predetermined excitation spectrum to be used is selected from the excitation spectrum inversely quantized in operation 1305 according to the decoded information, and the corresponding excitation spectrum is patched to a region larger than a predetermined frequency. Noise is generated by symmetrical folding. For example, in operation 1335, the excitation spectrum provided in the low frequency region illustrated in FIG. 8A is patched in the high frequency region as illustrated in FIG. 8B.

  The envelope information of the high frequency spectrum encoded by the encoder is decoded (operation 1340). In operation 1340, the envelope of the noise generated in operation 1335 is adjusted using the decoded envelope information of the high frequency spectrum. For example, in operation 1340, the noise generated in operation 1335 shown in FIG. 8B is adjusted using the envelope information of the high frequency spectrum as shown in FIG. 8C.

  In FIG. 11A, a reverse process of the transformation performed in operation 1110, the noise whose envelope is adjusted in operation 1340 is inversely transformed from the frequency domain to the time domain to generate a high frequency signal (operation 1345).

  FIG. 14A is a flowchart illustrating an embodiment of an adaptive high frequency domain decoding method according to the present invention.

  First, the bitstream input from the encoding end is demultiplexed (operation 1400). In operation 1400, information about the domain in which the low-frequency signal selected by the encoder is encoded, the LPC coefficient encoded by the encoder, the result of long-term prediction performed by the encoder, the encoder Reverses the information encoded in the code, the excitation spectrum quantized by the encoder, and the low-frequency signal or the low-frequency spectrum provided in a region smaller than the predetermined frequency to decode the high-frequency signal. Multiplex.

  Decodes information in the domain where the low frequency region corresponding to the region smaller than the predetermined frequency was encoded by the encoder to determine whether the low frequency region was encoded in the time domain or in the frequency domain (Step 1405).

  If it is determined in step 1405 that the low frequency region is encoded in the time domain, the low frequency region excitation signal encoded by the encoder is decoded (step 1410).

  A result obtained by performing the long section prediction on the low frequency signal, which is a signal provided in a region smaller than the predetermined frequency, is decoded, and the long section prediction is applied to the excitation signal decoded in operation 1410. The results obtained are synthesized (step 1415).

  If it is determined in operation 1405 that the low frequency region is encoded in the frequency domain, the excitation spectrum quantized by the encoder is inversely quantized (operation 1420).

  In the inverse process of the transformation performed in step 1125 of FIG. 11A, the excitation spectrum inversely quantized in step 1420 is inversely transformed from the frequency domain to the time domain to generate an excitation signal (step 1425).

  The LPC coefficients of the low frequency signal are decoded, and the decoded LPC coefficients are synthesized with the excitation signal synthesized in operation 1415 or the excitation signal generated in operation 1425 (operation 1430). The signal synthesized in operation 1430 is obtained by restoring the low-frequency signal that is finally provided in the low-frequency region.

  The high frequency signal is decoded using the excitation spectrum inversely quantized in operation 1420 or the excitation signal decoded in operation 1410 (operation 1435). If the low frequency region is encoded in the time domain, step 1435 decodes the high frequency signal using the excitation spectrum inversely quantized in step 1420, and if the low frequency region is If encoded in the domain, operation 1435 decodes the high frequency signal using the excitation signal decoded in operation 1410.

  The low-frequency signal restored in operation 1430 is combined with the high-frequency signal decoded in operation 1435 (operation 1440).

  FIG. 14B is a flowchart illustrating an example of operation 1435 included in the adaptive high frequency domain decoding method according to the present invention.

  First, whether the domain in which the low frequency region corresponding to the region smaller than the predetermined frequency is encoded is decoded into the high frequency region corresponding to the region larger than the predetermined frequency in the time domain or in the frequency domain. (Step 1445).

  If it is determined in step 1445 that the high frequency region is decoded in the time domain, the LPC coefficient for the high frequency signal provided in the region larger than the predetermined frequency is decoded (step 1450). An envelope as shown in the graph of FIG. 7A may be restored using the LPC coefficients decoded in operation 1450.

  The excitation signal decoded in operation 1410 of FIG. 12A is multiplied by the envelope of the LPC coefficients decoded in operation 1450 (operation 1455). An example of the signal multiplied in operation 1455 is a signal corresponding to the member number 710 shown in FIG. 7B.

  The gain value encoded by the encoder is decoded and applied to the signal multiplied in operation 1455 (operation 1460). By applying the gain value in operation 1460, mismatching existing between the low frequency signal restored in operation 1430 of FIG. 14A and the decoded high frequency signal may be compensated. For example, as shown in FIG. 7B, the high-frequency signal multiplied in operation 1455 causes mismatching at the interface with the low-frequency signal. However, by applying the gain value in operation 1460, As shown in 7C, the occurrence of mismatch between the low frequency signal and the high frequency signal is prevented.

  If it is determined in step 1445 that the high frequency region is decoded in the frequency domain, the excitation spectrum dequantized in step 1420 of FIG. 14A is patched to the high frequency region or folded symmetrically. To generate a spectrum (step 1465). For example, step 1465 patches the excitation spectrum shown in FIG. 8A to the high frequency region as shown in FIG. 8B.

  The high frequency spectrum envelope information encoded by the encoder is input and decoded (operation 1470). In operation 1470, the envelope information of the spectrum generated in operation 1465 is adjusted using the decoded envelope information of the high frequency spectrum. For example, in operation 1470, the envelope generated in operation 1465 shown in FIG. 8B is adjusted using the envelope information of the high frequency spectrum as shown in FIG. 8C.

  In FIG. 11A, a reverse process of the transformation performed in operation 1125, the spectrum whose envelope is adjusted in operation 1470 is inversely transformed from the frequency domain to the time domain to generate a high frequency signal (operation 1475).

  The present invention can be embodied as a computer readable code on a computer readable recording medium (including any apparatus having an information processing function). Computer-readable recording media include all types of recording devices that can store data that can be read by a computer system. Examples of a computer-readable recording device include a ROM (Read Only Memory), a RAM (Random Access Memory), a CD-ROM, a magnetic tape, a floppy (registered trademark) disk, and an optical data storage device.

  Although the present invention has been described with reference to the embodiments illustrated in the drawings to assist the understanding of the present invention, this is merely exemplary, and those skilled in the art will recognize that various modifications and equivalent other implementations are possible. It will be understood that examples are possible. Accordingly, the true technical protection scope of the present invention should be determined by the appended claims.

  According to the adaptive high frequency domain encoding method and apparatus of the present invention, a signal provided in a region larger than a predetermined frequency is converted into a time domain or a frequency domain using a signal provided in a region smaller than the predetermined frequency. To adaptively encode or decode.

  Thereby, although the audio signal is encoded or decoded using a small number of bits, the coding efficiency can be maximized in order not to deteriorate the sound quality of the signal corresponding to the high frequency region.

1 is a block diagram illustrating an embodiment of an adaptive high frequency domain encoding apparatus according to the present invention. FIG. FIG. 6 is a block diagram illustrating an example of a high frequency domain encoder 160 included in an adaptive high frequency domain encoder according to the present invention. 1 is a block diagram illustrating an embodiment of an adaptive high frequency domain encoding apparatus according to the present invention. FIG. FIG. 5 is a block diagram illustrating an example of a high frequency domain encoding unit 250 included in an adaptive high frequency domain encoding apparatus according to the present invention. 1 is a block diagram illustrating an embodiment of an adaptive high frequency domain encoding apparatus according to the present invention. FIG. FIG. 10 is a block diagram illustrating an example of a high frequency domain encoding unit 360 included in an adaptive high frequency domain encoding apparatus according to the present invention. 1 is a block diagram illustrating an embodiment of an adaptive high frequency domain decoding device according to the present invention. FIG. FIG. 10 is a block diagram illustrating an example of a high frequency domain decoding unit 440 included in an adaptive high frequency domain decoding device according to the present invention. 1 is a block diagram illustrating an embodiment of an adaptive high frequency domain decoding device according to the present invention. FIG. FIG. 10 is a block diagram illustrating an embodiment of a high frequency domain decoding unit 525 included in an adaptive high frequency domain decoding device according to the present invention. 1 is a block diagram illustrating an embodiment of an adaptive high frequency domain decoding device according to the present invention. FIG. 1 is a block diagram illustrating an embodiment of an adaptive high frequency domain decoding device according to the present invention. FIG. It is a graph which shows one Example of the envelope restored | reconstructed with the LPC coefficient. It is a graph which shows one Example which multiplied the excitation signal of the envelope which was decompress | restored by the low frequency signal and the LPC coefficient. It is a graph which shows one Example which compensated the mismatching which exists between a low frequency signal and a high frequency signal. It is a graph which shows one Example of the excitation spectrum with which the low frequency area | region was equipped. It is a graph which shows one Example which patched the excitation spectrum with which the low frequency area | region was equipped to the high frequency area | region. It is a graph which shows one Example which adjusted the envelope of the high frequency spectrum. 3 is a flowchart illustrating an embodiment of an adaptive high frequency domain encoding method according to the present invention. 10 is a flowchart illustrating an exemplary embodiment for operation 960 included in an adaptive high frequency domain encoding method according to the present invention. 3 is a flowchart illustrating an embodiment of an adaptive high frequency domain encoding method according to the present invention. FIG. 10 is a flowchart illustrating an example of operation 1050 included in the adaptive high frequency domain encoding method according to the present invention. 3 is a flowchart illustrating an embodiment of an adaptive high frequency domain encoding method according to the present invention. 6 is a flowchart illustrating an example of operation 1160 included in an adaptive high frequency domain encoding method according to the present invention. 6 is a flowchart illustrating an embodiment of an adaptive high frequency domain decoding method according to the present invention. 12 is a flowchart illustrating an embodiment for operation 1240 included in an adaptive high frequency domain decoding method according to the present invention. 6 is a flowchart illustrating an embodiment of an adaptive high frequency domain decoding method according to the present invention. 12 is a flowchart illustrating an example of operation 1325 included in an adaptive high frequency domain decoding method according to the present invention. 6 is a flowchart illustrating an embodiment of an adaptive high frequency domain decoding method according to the present invention. FIG. 40 is a flowchart illustrating an embodiment for operation 1435 included in the adaptive high frequency domain decoding method according to the present invention.

Claims (54)

A domain converter that selects a high-frequency signal, which is a signal provided in a region larger than a preset frequency, for each frequency band, and selects either the time domain or the frequency domain and converts it,
A time domain encoding unit that encodes the frequency band converted to the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency;
An adaptive high frequency domain encoding device comprising: a frequency domain encoding unit that encodes the frequency band converted to the frequency domain using an excitation spectrum provided in the low frequency region . The time domain encoding unit includes:
A linear prediction unit that performs linear prediction on a signal provided in a frequency band converted to the time domain;
A multiplier for multiplying the envelope generated by the performed linear prediction by the excitation signal;
A gain value encoding unit that calculates and encodes a gain value that matches a boundary between a low frequency signal of the signal provided in the low frequency region and an envelope multiplied by the excitation signal; The adaptive high frequency domain encoding apparatus according to claim 1. The frequency domain encoding unit includes:
Noise information encoding that selects a frequency band to be used for encoding a spectrum provided in a frequency band converted to the frequency domain from the excitation spectrum, and encodes information related to the selected frequency band And
The adaptive high frequency domain according to claim 1, further comprising: an envelope information encoding unit that extracts and encodes an envelope of a spectrum provided in a frequency band converted to the frequency domain. Encoding device. The excitation signal is
Among the low frequency signals provided in the low frequency region, the signal is extracted by performing linear prediction or long interval prediction on a signal provided in a frequency band indicated in a time domain. The adaptive high frequency domain encoding apparatus according to claim 1. The excitation spectrum is
2. The adaptive spectrum according to claim 1, wherein the spectrum is a whitening-processed spectrum with respect to a spectrum provided in a frequency band indicated in a frequency domain among low-frequency spectra provided in the low-frequency region. High frequency domain encoding device. Selects the frequency band used to encode the high frequency spectrum, which is the spectrum provided in the region higher than the preset frequency, among the excitation spectrum provided in the low frequency region, which is the region smaller than the preset frequency. A noise information encoding unit that encodes information related to the selected frequency band;
An adaptive high-frequency domain encoding apparatus comprising: an envelope information encoding unit that extracts and encodes an envelope of the high-frequency spectrum. The excitation spectrum is
7. The adaptive high frequency region encoding apparatus according to claim 6, wherein the low frequency spectrum provided in the low frequency region is a spectrum subjected to whitening processing. A domain selection unit that selects any one of the time domain and the frequency domain according to a preset reference with respect to a high-frequency signal that is a signal provided in a region larger than the preset frequency;
If a time domain is selected by the domain selection unit, a time domain encoding unit that encodes the high frequency signal using an excitation signal provided in a low frequency region that is a region smaller than a preset frequency;
If a frequency domain is selected by the domain selection unit, the high frequency signal is converted into a frequency domain to generate a high frequency spectrum, and the high frequency spectrum is generated using an excitation spectrum provided in the low frequency region. An adaptive high frequency domain encoding apparatus comprising: a frequency domain encoding unit for encoding. The time domain encoding unit includes:
If a time domain is selected by the domain selection unit, a linear prediction unit that performs linear prediction on the high-frequency signal;
A multiplier for multiplying the envelope generated by the performed linear prediction by the excitation signal;
A gain value encoding unit that calculates and encodes a gain value that matches a boundary between a low frequency signal that is a signal provided in the low frequency region and an envelope that is multiplied by the excitation signal; The adaptive high frequency domain encoding apparatus according to claim 8. The frequency domain encoding unit includes:
When the time domain is selected by the domain selection unit, the high frequency signal is converted into a frequency domain, and the conversion unit generates the high frequency spectrum;
A noise information encoding unit that selects a frequency band to be used for encoding the high frequency spectrum from the excitation spectrum, and encodes information related to the selected frequency band;
9. The adaptive high frequency region encoding apparatus according to claim 8, further comprising an envelope information encoding unit that extracts and encodes an envelope of the high frequency spectrum. The excitation signal is
9. The adaptive high frequency region encoding apparatus according to claim 8, wherein the low frequency signal is a signal extracted by performing linear prediction or long interval prediction on a low frequency signal provided in the low frequency region. The excitation spectrum is
A spectrum obtained by converting a signal extracted by performing linear prediction or long-term prediction to a low frequency signal provided in the low frequency region into a frequency domain, or a low frequency spectrum that is a spectrum provided in the low frequency region 9. The adaptive high-frequency region encoding apparatus according to claim 8, wherein the spectrum is a whitening-processed spectrum. A domain determination unit that determines a domain in which each frequency band provided in a high frequency region, which is a region larger than a preset frequency, is encoded;
A time domain decoding unit that decodes a frequency band determined to be encoded in the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency;
A frequency domain decoding unit that decodes a frequency band determined to be encoded in the frequency domain by using an excitation spectrum provided in a low frequency region, and performing adaptive high frequency region decoding Device. The time domain decoding unit includes:
A linear synthesis unit that decodes LPC (Linear Predictive Coding) coefficients for a signal provided in the frequency band converted to the time domain;
A multiplier for multiplying the envelope generated by the decoded LPC coefficients by the excitation signal;
A gain value that matches a boundary between a low-frequency signal that is a signal provided in the low-frequency region and a high-frequency signal provided in the high-frequency region is decoded and applied to an envelope that is multiplied by the excitation signal. 14. The adaptive high frequency domain decoding apparatus according to claim 13, further comprising a gain value application unit. The frequency domain decoding unit includes:
A noise generation unit that generates information in a frequency band determined to be encoded in the frequency domain, using information on a frequency band used to decode the high frequency region of the excitation spectrum;
The envelope adjustment unit for decoding an envelope of a spectrum provided in a frequency band converted into the frequency domain and adjusting an envelope of the generated noise. Adaptive high frequency domain decoder. The excitation signal is
14. The adaptive high frequency domain decoding apparatus according to claim 13, wherein the apparatus is a signal extracted by performing linear prediction or long interval prediction on a low frequency signal provided in the low frequency domain. The excitation spectrum is
14. The adaptive high frequency domain decoding apparatus according to claim 13, wherein the low frequency spectrum provided in the low frequency domain is a spectrum subjected to whitening processing. Of the excitation spectrum provided in the low frequency region that is a region smaller than the preset frequency, using information related to the frequency band used to decode the high frequency region that is a region greater than the preset frequency, A noise generator for generating noise in the high frequency region;
An adaptive high frequency domain decoding device comprising: an envelope adjustment unit that decodes an envelope of a high frequency spectrum provided in the high frequency region and adjusts an envelope of the generated noise . The excitation spectrum is
19. The adaptive high frequency domain decoding apparatus according to claim 18, wherein the low frequency spectrum provided in the low frequency domain is a spectrum subjected to whitening processing. A domain determination unit for determining a domain in which a high frequency region, which is a region larger than a preset frequency, is encoded;
If the domain determination unit determines that the high frequency region is encoded in the time domain, the high frequency signal provided in the high frequency region is provided in a low frequency region that is smaller than a preset frequency. A time domain decoding unit for decoding using the excited excitation signal;
If the domain determination unit determines that the high frequency region is encoded in the frequency domain, the high frequency spectrum provided in the high frequency region is decoded using the excitation spectrum provided in the low frequency region. An adaptive high frequency domain decoding apparatus comprising: a frequency domain decoding unit configured to convert to a high frequency domain. The time domain decoding unit includes:
A linear synthesis unit for decoding LPC coefficients for the high-frequency signal;
A multiplier for multiplying the envelope generated by the decoded LPC coefficients by the excitation signal;
A gain value application unit that decodes a gain value that matches a boundary between a low-frequency signal that is a signal provided in the low-frequency region and the high-frequency signal, and applies the decoded value to an envelope that is multiplied by the excitation signal 21. The adaptive high frequency decoding apparatus according to claim 20, wherein: The frequency domain decoding unit includes:
A noise generation unit that generates information in the high frequency region using information related to a frequency band used for decoding of the excitation spectrum;
21. The adaptive high frequency domain decoding apparatus according to claim 20, further comprising: an envelope adjustment unit that decodes an envelope of the high frequency spectrum and adjusts an envelope of the generated noise. The excitation signal is
21. The adaptive high frequency domain decoding apparatus according to claim 20, wherein the low frequency signal is a signal extracted by performing linear prediction or long interval prediction on a low frequency signal provided in the low frequency domain. The excitation spectrum is
A spectrum obtained by converting a signal extracted by performing linear prediction or long interval prediction on a low frequency signal that is a signal provided in the low frequency region into a frequency domain, or a spectrum provided in the low frequency region. 21. The adaptive high-frequency decoding apparatus according to claim 20, wherein the low-frequency spectrum is a spectrum subjected to whitening processing. Selecting a high frequency signal, which is a signal provided in a region larger than a preset frequency, for each frequency band, selecting either the time domain or the frequency domain, and converting it,
Encoding the frequency band converted to the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency;
And encoding the frequency band converted into the frequency domain using an excitation spectrum provided in the low frequency region. The step of encoding using the excitation signal includes:
Performing linear prediction on a signal provided in a frequency band converted to the time domain;
Multiplying the envelope generated by the performed linear prediction by the excitation signal;
And calculating and encoding a gain value for matching a boundary between a low frequency signal, which is a signal provided in the low frequency region, and an envelope multiplied by the excitation signal. 25. The adaptive high frequency domain encoding method according to 25. The encoding using the excitation spectrum includes:
Selecting a frequency band to be used for encoding a spectrum provided in a frequency band converted into the frequency domain from the excitation spectrum, and encoding information relating to the selected frequency band;
26. The method of claim 25, further comprising: extracting and encoding a spectrum envelope provided in a frequency band converted to the frequency domain. The excitation signal is
Among the low frequency signals provided in the low frequency region, the signal is extracted by performing linear prediction or long interval prediction on a signal provided in a frequency band indicated in a time domain. The adaptive high frequency domain encoding method according to claim 25. The excitation spectrum is
26. The adaptive spectrum according to claim 25, wherein the spectrum is a whitening-processed spectrum with respect to a spectrum provided in a frequency band indicated in a frequency domain among low frequency spectra provided in the low frequency region. High frequency domain encoding method. Selects the frequency band used to encode the high frequency spectrum, which is the spectrum provided in the region higher than the preset frequency, among the excitation spectrum provided in the low frequency region, which is the region smaller than the preset frequency. And encoding information related to the selected frequency band;
An adaptive high frequency domain encoding method comprising: extracting and encoding an envelope of the high frequency spectrum. The excitation spectrum is
31. The adaptive high frequency region encoding method according to claim 30, wherein the low frequency spectrum provided in the low frequency region is a spectrum subjected to whitening processing. Selecting either one of the time domain and the frequency domain according to a preset criterion for a high-frequency signal that is a signal provided in a region larger than the preset frequency;
If the time domain is selected in the selecting step, the high frequency signal is encoded using an excitation signal provided in a low frequency region that is a region smaller than a preset frequency; and
If a frequency domain is selected in the selecting step, the high frequency signal is converted into a frequency domain to generate a high frequency spectrum, and the high frequency spectrum is generated using an excitation spectrum provided in the low frequency region. An adaptive high frequency domain encoding method. The step of encoding using the excitation signal includes:
If a time domain is selected by the domain selection unit, performing linear prediction on the high frequency signal;
Multiplying the envelope generated by the performed linear prediction by the excitation signal;
And calculating and encoding a gain value for matching a boundary between a low frequency signal, which is a signal provided in the low frequency region, and an envelope multiplied by the excitation signal. 33. The adaptive high frequency domain encoding method according to 32. The encoding using the excitation spectrum includes:
If the time domain is selected in the selecting step, converting the high frequency signal to the frequency domain and generating the high frequency spectrum;
Selecting a frequency band to be used for encoding the high frequency spectrum from the excitation spectrum, and encoding information related to the selected frequency band;
The adaptive high frequency domain encoding method of claim 32, further comprising: extracting and encoding an envelope of the high frequency spectrum. The excitation signal is
The adaptive high frequency domain encoding method according to claim 32, wherein the low frequency signal provided in the low frequency domain is a signal extracted by performing linear prediction or long interval prediction. The excitation spectrum is
A spectrum obtained by converting a signal extracted by performing linear prediction or long-term prediction to a low frequency signal provided in the low frequency region into a frequency domain, or a low frequency spectrum that is a spectrum provided in the low frequency region 33. The adaptive high frequency domain encoding method according to claim 32, wherein the spectrum is a whitened spectrum. Determining a domain in which each frequency band provided in a high frequency region, which is a region larger than a preset frequency, is encoded;
Decoding the frequency band determined to be encoded in the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency;
And decoding the frequency band determined to be encoded in the frequency domain using an excitation spectrum provided in a low frequency region. Decoding using the excitation signal comprises:
Decoding LPC coefficients for a signal provided in a frequency band converted to the time domain;
Multiplying the envelope generated by the decoded LPC coefficients by the excitation signal;
A gain value that matches a boundary between a low-frequency signal that is a signal provided in the low-frequency region and a high-frequency signal provided in the high-frequency region is decoded and applied to an envelope that is multiplied by the excitation signal. 38. The adaptive high frequency domain decoding method of claim 37, further comprising: Decoding using the excitation spectrum comprises:
Generating information in a frequency band determined to be encoded in the frequency domain using information related to a frequency band used to decode the high frequency region of the excitation spectrum; and
38. The method of claim 37, further comprising: decoding a spectrum envelope provided in a frequency band transformed to the frequency domain and adjusting the generated noise envelope. Frequency domain decoding method. The excitation signal is
38. The adaptive high frequency domain decoding method according to claim 37, wherein the adaptive high frequency domain decoding method is a signal extracted by performing linear prediction or long interval prediction on a low frequency signal provided in the low frequency region. The excitation spectrum is
The adaptive high frequency domain decoding method according to claim 37, wherein the low frequency spectrum provided in the low frequency domain is a spectrum subjected to whitening processing. Of the excitation spectrum provided in the low frequency region that is a region smaller than the preset frequency, using information related to the frequency band used to decode the high frequency region that is a region greater than the preset frequency, Generating noise in the high frequency region;
And decoding an envelope of a high frequency spectrum provided in the high frequency region and adjusting an envelope of the generated noise. The excitation spectrum is
43. The adaptive high frequency domain decoding method according to claim 42, wherein the low frequency spectrum provided in the low frequency domain is a spectrum subjected to whitening processing. Determining a domain in which a high frequency region, which is a region larger than a preset frequency, is encoded;
If it is determined that the high frequency region is encoded in the time domain in the determining step, the high frequency signal provided in the high frequency region is provided in a low frequency region that is smaller than a preset frequency. Decoding using the excited signal;
If it is determined in the determining step that the high frequency region is encoded in the frequency domain, the high frequency spectrum provided in the high frequency region is decoded using the excitation spectrum provided in the low frequency region. And an adaptive high frequency domain decoding method. Decoding using the excitation signal comprises:
Decoding LPC coefficients for the high frequency signal;
Multiplying the envelope generated by the decoded LPC coefficients by the excitation signal;
Decoding a gain value that matches a boundary between a low frequency signal that is a signal provided in the low frequency region and the high frequency signal, and applying the decoded value to an envelope that is multiplied by the excitation signal. 45. The adaptive high frequency domain decoding method according to claim 44. Decoding using the excitation spectrum comprises:
Using information related to a frequency band used for decoding of the excitation spectrum, and generating noise in the high frequency region;
45. The method of claim 44, further comprising: decoding an envelope of the high frequency spectrum and adjusting an envelope of the generated noise. The excitation signal is
45. The adaptive high frequency domain decoding method according to claim 44, wherein the low frequency signal provided in the low frequency domain is a signal extracted by performing linear prediction or long interval prediction. The excitation spectrum is
A spectrum obtained by converting a signal extracted by performing linear prediction or long interval prediction on a low frequency signal that is a signal provided in the low frequency region into a frequency domain, or a spectrum provided in the low frequency region. 45. The adaptive high frequency domain decoding method according to claim 44, wherein the low frequency spectrum is a spectrum subjected to whitening processing. Selecting a high frequency signal, which is a signal provided in a region larger than a preset frequency, for each frequency band, selecting either the time domain or the frequency domain, and converting it,
Encoding the frequency band converted to the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency;
The computer-readable recording medium which recorded the program for performing the step which encodes the frequency band converted into the said frequency domain using the excitation spectrum provided in the said low frequency area | region by a computer. Selects the frequency band used to encode the high frequency spectrum, which is the spectrum provided in the region higher than the preset frequency, among the excitation spectrum provided in the low frequency region, which is the region smaller than the preset frequency. And encoding information related to the selected frequency band;
A computer-readable recording medium storing a program for causing a computer to extract and encode the high frequency spectrum envelope. Selecting either one of the time domain and the frequency domain according to a preset criterion for a high-frequency signal that is a signal provided in a region larger than the preset frequency;
If the time domain is selected in the selecting step, the high frequency signal is encoded using an excitation signal provided in a low frequency region that is a region smaller than a preset frequency; and
If a frequency domain is selected in the selecting step, the high frequency signal is converted into a frequency domain to generate a high frequency spectrum, and the high frequency spectrum is generated using an excitation spectrum provided in the low frequency region. A computer-readable recording medium storing a program for causing a computer to execute the encoding step. Determining a domain in which each frequency band provided in a high frequency region, which is a region larger than a preset frequency, is encoded;
Decoding the frequency band determined to be encoded in the time domain using an excitation signal provided in a low frequency region smaller than a preset frequency;
A computer-readable recording medium storing a program for causing a computer to execute a step of decoding a frequency band determined to be encoded in the frequency domain using an excitation spectrum provided in a low frequency region . Of the excitation spectrum provided in the low frequency region that is a region smaller than the preset frequency, using information related to the frequency band used to decode the high frequency region that is a region greater than the preset frequency, Generating noise in the high frequency region;
A computer-readable recording medium storing a program for causing a computer to execute a step of decoding an envelope of a high frequency spectrum provided in the high frequency region and adjusting the envelope of the generated noise. Determining a domain in which a high frequency region, which is a region larger than a preset frequency, is encoded;
If it is determined that the high frequency region is encoded in the time domain in the determining step, the high frequency signal provided in the high frequency region is provided in a low frequency region that is smaller than a preset frequency. Decoding using the excited signal;
If it is determined in the determining step that the high frequency region is encoded in the frequency domain, the high frequency spectrum provided in the high frequency region is decoded using the excitation spectrum provided in the low frequency region. A computer-readable recording medium having recorded thereon a program for causing the computer to execute the step of converting.

Images