WO2013118835A1 - Méthode d'encodage, dispositif d'encodage, méthode de décodage, dispositif de décodage, programme et support d'enregistrement - Google Patents

Méthode d'encodage, dispositif d'encodage, méthode de décodage, dispositif de décodage, programme et support d'enregistrement Download PDF

Info

Publication number
WO2013118835A1
WO2013118835A1 PCT/JP2013/052914 JP2013052914W WO2013118835A1 WO 2013118835 A1 WO2013118835 A1 WO 2013118835A1 JP 2013052914 W JP2013052914 W JP 2013052914W WO 2013118835 A1 WO2013118835 A1 WO 2013118835A1
Authority
WO
WIPO (PCT)
Prior art keywords
samples
signal sequence
range
normalized signal
predetermined value
Prior art date
Application number
PCT/JP2013/052914
Other languages
English (en)
Japanese (ja)
Inventor
勝宏 福井
祐介 日和▲崎▼
登 原田
守谷　健弘
優 鎌本
Original Assignee
日本電信電話株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 日本電信電話株式会社 filed Critical 日本電信電話株式会社
Priority to JP2013557574A priority Critical patent/JP5786044B2/ja
Publication of WO2013118835A1 publication Critical patent/WO2013118835A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
    • G10L19/02—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders
    • G10L19/032—Quantisation or dequantisation of spectral components

Definitions

  • the present invention relates to a technique for encoding a sound signal such as speech or music with a small amount of information, and more particularly, to an encoding technique for improving quantization accuracy.
  • the quantized value of the global gain (gain that affects the quantization accuracy of the normalized input signal sequence) is calculated in the time domain.
  • the energy of the signal in the time domain is equal to the energy of the signal in the frequency domain, even if the quantized value of the global gain is obtained in the frequency domain, this result is not different from that in the time domain. Therefore, here, a case where the quantized value of the global gain and the decoded value thereof are calculated in the frequency domain is illustrated.
  • the frequency domain transform unit 101 receives an input time domain signal sequence x F (t) in frame units composed of a plurality of consecutive samples included in the time domain input signal x (t).
  • the frequency domain transform unit 101 converts the frequency component at the L point (L is a positive integer, for example, 256) corresponding to the input time domain signal sequence x F (t) of one frame to the input frequency domain signal sequence X ( ⁇ ).
  • Output [ ⁇ ⁇ 0,..., L-1 ⁇ ].
  • t represents an index of discrete time
  • ⁇ represents an index of discrete frequency.
  • MDCT Modified Discrete Cosine Transform
  • DCT Discrete Cosine Transform
  • the normalization unit 102 includes an input frequency domain signal sequence X ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] and an input frequency domain signal sequence X ( ⁇ ) obtained by the gain control unit 104 described later.
  • a gain hereinafter referred to as a global gain
  • g that determines the quantization accuracy of each component of [ ⁇ ⁇ 0,..., L ⁇ 1 ⁇ ] is input.
  • the normalization unit 102 divides each component of the input frequency domain signal sequence X ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] by the global gain g, or the input frequency domain signal sequence X ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] by multiplying each component of [ ⁇ ⁇ 0,..., L-1 ⁇ ] by the reciprocal of the global gain g, respectively.
  • ⁇ ] Is normalized, and a normalized signal sequence X Q ( ⁇ ) [ ⁇ ⁇ 0,..., L ⁇ 1 ⁇ ] is output.
  • the quantizing unit 103 receives the normalized signal sequence X Q ( ⁇ ) [ ⁇ ⁇ 0,..., L ⁇ 1 ⁇ ].
  • the quantization unit 103 quantizes the normalized signal sequence X Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] by a predetermined method, and the normalized signal sequence X Q ( ⁇ ) Quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1] which is a series of quantized values of each component of [ ⁇ ⁇ 0,.
  • a normalized signal code that is a code corresponding to the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ], and the bits of the normalized signal code Number (hereinafter referred to as the number of consumed bits). Further, when receiving from the gain control unit 104 command information for outputting a quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] and a normalized signal code Then, the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] and the normalized signal code are output.
  • the gain control unit 104 receives the number of consumed bits.
  • the gain control unit 104 adjusts the global gain g so that the number of consumed bits approaches a maximum value that is less than or equal to the number of bits allocated in advance to the normalized signal code (hereinafter referred to as the specified number of bits).
  • the global gain g is output as a new global gain g.
  • a process of increasing the global gain g when the number of consumed bits is larger than the specified number of bits and decreasing the global gain g otherwise can be exemplified.
  • the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0, ..., L-1 ⁇ ] and the normalized signal code are Command information to be output is output to the quantization unit 103.
  • the global gain encoding unit 105 includes an input frequency domain signal sequence X ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] and a quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0. ,..., L-1 ⁇ ] is input.
  • the global gain encoding unit 105 includes an input frequency domain signal sequence X ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] and a quantum among a plurality of preset global gain quantization values.
  • the normalized signal code and the global gain code which are output codes of the encoding device, are transmitted to the decoding device and input to the decoding device.
  • a global gain code is input to the global gain decoding unit 106.
  • the global gain decoding unit 106 applies a decoding process corresponding to the encoding process performed by the global gain encoding unit 105 to decode the global gain code, and outputs a decoded global gain g ⁇ .
  • ⁇ Normalized signal decoding unit 107 A normalized signal code is input to the normalized signal decoding unit 107.
  • the normalized signal decoding unit 107 applies a decoding method corresponding to the encoding method performed by the quantization unit 103 of the encoding device, decodes the normalized signal code, and generates a decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] is output.
  • the decoded frequency component calculation unit 108 receives the decoded global gain g ⁇ and the decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ].
  • the decoded frequency component calculation unit 108 is obtained by multiplying each component of the decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] and the decoded global gain g ⁇ .
  • decoded frequency-domain signal sequence having sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ 0, ..., L-1 ⁇ ] is output as.
  • ⁇ Time domain conversion unit 109 Decoded frequency domain signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ 0,..., L ⁇ 1 ⁇ ] is input to time domain transform section 109.
  • the time domain transform unit 109 applies a frequency-time transform to the decoded frequency domain signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ], and outputs an output time domain signal sequence in units of frames.
  • the frequency-time conversion method is an inverse conversion corresponding to the time-frequency conversion method used in the frequency domain conversion unit 101.
  • the frequency-time conversion method here is IMDCT (Inverse Modified Discrete Cosine Transform) or IDCT (Inverse Discrete Cosine Transform).
  • the global gain is adjusted to appropriately control the coarseness of quantization of the normalized signal sequence, so that the number of bits consumed, which is the code amount of the normalized signal code, is the specified number of bits. Control is performed so that the following maximum value is obtained. For this reason, when the number of bits consumed is smaller than the specified number of bits, there is a problem in that the encoding process that makes full use of the number of bits allocated in advance for the normalized signal sequence cannot be performed.
  • an object of the present invention is to provide an encoding technique that improves the quantization accuracy of a normalized signal sequence with a small increase in code amount and a decoding technique thereof.
  • An encoding method is an encoding technique for encoding an input signal sequence in units of frames configured by a plurality of input signal samples, and each input signal sample included in the input signal sequence is normalized.
  • Normalized signal encoding processing for obtaining a normalized signal code obtained by encoding a sequence of the received signal and a quantized normalized signal sequence corresponding to the normalized signal code, and a gain corresponding to the input signal sequence
  • a global gain encoding process for obtaining a quantized global gain and a global gain code corresponding to the quantized global gain, and a quantized normalized signal sequence in N ranges (N is an integer of 2 or more) Segmentation processing to be segmented, correction gain obtained by correcting the quantized global gain with the gain correction amount for each segmented range, and each sample of the quantized normalized signal sequence Gain correction to obtain a gain correction amount code for specifying the gain correction amount for each divided range, in which the correlation between the signal sequence obtained by multiplying the signal value and the input signal sequence is
  • the number of samples whose sample energy is greater than or equal to a predetermined value among all the samples included in the first range to the nth range of the quantized normalized signal sequence is quantized.
  • the sample energy of all the samples included in the normalized signal sequence is the minimum number of samples that is greater than a predetermined value or equal to or greater than n / N of the number of samples greater than or equal to a predetermined value.
  • the number of samples whose absolute value of the sample is greater than or equal to the predetermined value is The absolute value of the sample among all the samples included in the normalized normalized signal sequence is a minimum number of samples that is greater than a predetermined value or equal to or greater than n / N of the number of samples greater than or equal to a predetermined value.
  • the number of samples whose sample energy is greater than or equal to a predetermined value among all samples included in the first range to the nth range of the quantized normalized signal sequence is quantized.
  • the maximum sample number is such that the energy of the sample is greater than a predetermined value or less than n / N of the number of samples greater than or equal to a predetermined value.
  • the number of samples whose absolute value of samples is greater than or equal to a predetermined value among all samples included in the first range to the nth range of the quantized normalized signal sequence is The absolute value of the sample among all the samples included in the normalized normalized signal sequence is a maximum number of samples that is greater than a predetermined value or equal to or less than n / N of the number of samples greater than or equal to a predetermined value. Seeking The quantized normalized signal sequence is determined by setting a range other than the first range to the (N ⁇ 1) th range in the quantized normalized signal sequence as the Nth range of the quantized normalized signal sequence. Is divided into N ranges.
  • the correction gain is, for example, (1) a value obtained by adding the quantized global gain and the gain correction amount, and (2) a square of the values of all the samples in the frame of the quantized global gain and the quantized normalized signal sequence.
  • the number of samples in which the energy of the samples in the frame of the quantized normalized signal sequence is larger than a predetermined value the energy of the samples in the divided range of the quantized normalized signal sequence is larger than the predetermined value.
  • the gain correction amount includes a pre-quantization width gain correction amount included in a gain correction amount code book in which candidates for pre-quantization width gain correction amounts are stored in advance, and a quantization step width corresponding to a quantized global gain. And a value obtained by multiplying.
  • a decoding method is a decoding technique for obtaining an output signal sequence by decoding a frame-by-frame code, and decoding a normalized signal code included in the code to obtain a decoded normalized signal sequence
  • Signal processing segmentation processing that divides the decoded normalized signal sequence into N ranges (N is an integer of 2 or more), and gain corresponding to each range by decoding the gain correction amount code included in the code
  • N is an integer of 2 or more
  • a gain correction amount decoding process for obtaining a correction amount, a global gain decoding process for obtaining a decoded global gain by decoding a global gain code included in the code, and a decoding global gain is corrected by a gain correction amount for each divided range.
  • nth range (n is an integer from 1 to N-1) of the decoded normalized signal sequence, (a) Among all samples included in the first range to the nth range of the decoded normalized signal sequence, the number of samples whose sample energy is greater than or equal to a predetermined value and the decoding normalization N of N samples of the number of samples whose sample energy is greater than or equal to or greater than a predetermined value among all the samples included in the completed signal sequence is closest.
  • the number of samples whose absolute value of samples is greater than or equal to a predetermined value among all samples included in the first range to the nth range of the decoded normalized signal sequence is the decoding normal So that the absolute value of the samples among all the samples included in the digitized signal sequence is a minimum number of samples that is greater than a predetermined value or equal to or greater than n / N of the number of samples greater than or equal to a predetermined value.
  • the number of samples whose sample energy is greater than or equal to a predetermined value among all samples included in the first range to the nth range of the decoded normalized signal sequence is decoded normalization Among all the samples included in the completed signal series, the sample energy is greater than a predetermined value or the maximum number of samples that is n or less of N / N of the number of samples greater than or equal to a predetermined value.
  • the number of samples whose absolute value of the sample is greater than or equal to the predetermined value is the decoding normal
  • the absolute value of the sample among all the samples included in the digitized signal sequence is greater than the predetermined value or the maximum number of samples that is n or less of N / N of the number of samples greater than or equal to the predetermined value, Seeking
  • N decoded normalized signal sequences are obtained. This is done by dividing the range.
  • the correction gain is, for example, (1) a value obtained by adding the decoding global gain and the gain correction amount, and (2) decoding the sum of squares of the values of the decoding global gain and all samples in the frame of the decoded normalized signal sequence.
  • the gain correction amount includes a gain correction amount before quantization width included in a gain correction amount codebook in which candidates for a gain correction amount before quantization width multiplication are stored in advance, a quantization step width corresponding to a decoded global gain, It is good also as a value obtained by multiplying.
  • the acoustic signal handled in each embodiment is a signal such as a sound, a sound such as a musical sound, and a video.
  • the acoustic signal is a time domain signal.
  • the time domain signal may be converted into a frequency domain signal or a frequency domain signal may be converted into a time domain signal by a known technique as necessary. You can also. Therefore, the signal to be encoded may be a time domain signal or a frequency domain signal (in the following description, the frequency domain signal is treated for the sake of concrete explanation).
  • the signal input as the target of the encoding process is a sequence (sample sequence) composed of a plurality of samples, and the encoding process is normally executed in units of frames. I will call it.
  • each component included in the input signal sequence X ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ], the quantized global gain g ⁇ , and the quantized normalized signal The relationship between the components included in the sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] can be expressed by Expression (1).
  • e g is the quantization error between the global gain g and the quantized global gain g ⁇
  • e XQ is the normalized input signal sequence X Q ( ⁇ ) [ ⁇ ⁇ 0, ..., L-1 ⁇ ].
  • Quantization normalized signal sequence X ⁇ Q ( ⁇ ) represents a quantization error between corresponding components (components having the same value of ⁇ ) included in [ ⁇ ⁇ 0, ..., L-1 ⁇ ] .
  • the number of bits consumed by a normalized signal code that is a code corresponding to a quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] is Depending on the input signal sequence, a part of the predetermined number of bits predetermined for the normalized signal code often remains as unused bits. Therefore, the excess one or more bits (hereinafter, referred to as surplus bits) utilizing the reduction of the quantization error e g and e XQ. Furthermore, not only the surplus bits, but also one or a plurality of bits prepared in advance for reducing the quantization error may be used.
  • the entire sequence of the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ 0,..., L-1 ⁇ ] is divided into a plurality of ranges.
  • An easily conceivable method for dividing the same signal sequence B into N ranges by the encoding device and the decoding device is to specify a range such as the boundary position of adjacent ranges and the number of components included in each range.
  • information is output from the encoding device.
  • a large number of bits are required to output information specifying the range.
  • the coding apparatus and the decoding apparatus perform classification according to the same standard without using the information specifying the range as the output of the coding apparatus, that is, without consuming bits.
  • the gain correction bits that is, the amount of information for correcting the quantized global gain, are given to each range as evenly as possible, the components of the quantized normalized signal sequence included in each range It is desirable that the amount of information be as uniform as possible.
  • the “criteria for classifying so that the number of significant samples included in each range is as equal as possible” is adopted as a criterion for series division.
  • “significant” can be defined as, for example, “the sample amplitude is not zero” or “the absolute value of the sample amplitude is greater than or greater than a predetermined value”.
  • Quantization of a normalized signal sequence often employs a quantization method that assigns codes only to some of the samples included in the normalized signal sequence. Assuming that the average value of the amplitude of the samples included in the quantized normalized signal sequence is almost the same at all discrete frequencies, the amount of information in each range of the quantized normalized signal sequence is the amplitude included in each range. Can be approximated by the number of non-zero samples. Therefore, if the “criteria for classifying so that the number of significant samples included in each range is as equal as possible” is adopted, the information amount of the components of the quantized normalized signal sequence included in each range is approximately calculated. It is possible to make it as uniform as possible. A specific classification method based on this criterion will be described in detail later. For example, an algebraic codebook adopted in G.729 of the ITU-T standard standard is used as a quantization method in which the amplitudes of the samples included in the quantized normalized signal sequence are all the same. Quantization methods are mentioned.
  • the encoding apparatus 1 (see FIG. 2) of the first embodiment includes a normalized signal encoding unit 120, a global gain encoding unit 105, a gain correction amount encoding unit 140, and a sorting unit 150.
  • the classification unit 150 is illustrated as a component of the gain correction amount encoding unit 140.
  • the classification unit 150 includes the gain correction amount. It may be a component different from the encoding unit 140.
  • the encoding device 1 may include a frequency domain transform unit 101 and a synthesis unit 160 as necessary.
  • the input signal sequence of the encoding device 1 is an input signal sequence X ( ⁇ that is a frequency component of L points (L is a positive integer, for example, 256) corresponding to the acoustic signal x (t) in units of frames. ) [ ⁇ ⁇ L min ,..., L max ⁇ ]
  • t is an index of discrete time
  • ⁇ is an index of discrete frequency
  • L min is an index of minimum discrete frequency among frequency components at L point
  • L max is a maximum discrete frequency among frequency components at L point. Represents the index.
  • the frame-wise acoustic signal x (t) itself may be used as the input signal sequence of the encoding device 1, or a residual signal obtained by performing linear prediction analysis on the frame-wise acoustic signal x (t) is encoded. 1 may be used as the input signal sequence, or a frequency component at L point (L is a positive integer, for example, 256) corresponding to the residual signal may be used as the input signal sequence.
  • the encoding device 1 may include a frequency domain transform unit 101 as a preprocessing unit of the encoding device 1 or in the encoding device 1.
  • the frequency domain transform unit 101 generates frequency components at L points (L is a positive integer, for example, 256) corresponding to the time domain acoustic signal x (t) in units of frames, and the input signal sequence X ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ]
  • MDCT Modified Discrete Cosine Transform
  • DCT Discrete Cosine Transform
  • a residual signal obtained by linear prediction analysis of the time domain acoustic signals in units of frames may be set as x (t).
  • the normalized signal encoding unit 120 encodes a sequence based on a signal obtained by normalizing each component of the input signal sequence X ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] in units of frames.
  • the quantized signal code and the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] corresponding to the normalized signal code are output (step S1e).
  • the normalization signal encoding unit 120 is realized by, for example, the normalization unit 102, the quantization unit 103, and the gain control unit 104 in FIG. Each of the normalization unit 102, the quantization unit 103, and the gain control unit 104 operates as described in the [Background Art] column.
  • the global gain encoding unit 105 supports a quantized global gain g ⁇ that is a gain corresponding to the input signal sequence X ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] and a quantized global gain g ⁇ .
  • the global gain code to be obtained is obtained (step S2e).
  • the global gain encoding unit 105 also obtains a quantization step width corresponding to the quantized global gain g ⁇ as necessary.
  • the global gain encoding unit 105 operates, for example, as described in the “Background art” column.
  • the global gain encoding unit 105 includes a table storing a plurality of sets of quantized global gain candidates and global gain codes corresponding to the candidates, and the global gain obtained by the normalized signal encoding unit 120
  • the candidate for the quantized global gain closest to the gain g may be set as the quantized global gain g ⁇ , and the global gain code corresponding to the candidate may be output.
  • the global gain coding unit 105 multiplies each component of the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] and the gain and obtains a signal.
  • Quantized global gain g ⁇ obtained on the basis of the maximum correlation or minimum error between the sequence and the input signal sequence X ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] and this quantization
  • a global gain code corresponding to the global gain may be obtained and output.
  • the quantization step width corresponding to the quantized global gain ⁇ is also the gain correction amount code. Is output to the conversion unit 140.
  • the sorting unit 150 determines that the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] is “as large as possible in the number of significant samples included in each range. According to the “criterion for classification”, the range is divided into N ranges (where N is a predetermined integer of 2 or more) (step S3e).
  • a set of discrete frequency indices of the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L max ⁇ ] is represented by ⁇ L min , ..., L max ⁇
  • the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] corresponds to the entire sequence B
  • division information Information relating to the division into N ranges obtained by this division processing (hereinafter referred to as division information) is output from the division unit 150 and provided to the gain correction amount encoding unit 140.
  • the gain correction amount encoding unit 140 includes an input signal sequence X ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ], a quantized global gain g ⁇ , and a quantized normalized signal sequence X ⁇ Q. ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] and classification information are input.
  • the gain correction amount encoding unit 140 uses a gain correction amount code book stored in a storage unit (not shown) to correct the quantization global gain for each range divided by the gain correction amount, Quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] multiplied by the value of each sample and the input signal sequence X ( ⁇ ) [ ⁇
  • a gain correction amount code which is a code for specifying the gain correction amount for each divided range, in which the correlation with ⁇ ⁇ L min ,..., L max ⁇ ] is maximum or the error is minimum is output (step S4e). ).
  • the synthesis unit 160 outputs a bit stream in which the normalized signal code, the gain correction amount code, and the global gain code are collected.
  • the bit stream is transmitted to the decoding device 2.
  • the classification process based on the “criteria for classifying so that the number of significant samples included in each range is as equal as possible” is performed by, for example, the nth range of a quantized normalized signal sequence (where n is 1 to N ⁇ 1).
  • the number of samples whose absolute value of samples is greater than or equal to a predetermined value among all samples included in the first range to the n-th range of the quantized normalized signal sequence is set to a minimum number of samples that is n or more of N / N of the number of samples that are greater than or equal to a predetermined value.
  • the number of samples whose sample energy is greater than or equal to a predetermined value among all the samples included in the first range to the nth range of the quantized normalized signal sequence is quantum So that the maximum number of samples becomes n or less of N / N of the number of samples whose sample energy is greater than a predetermined value or greater than or equal to a predetermined value among all samples included in the normalized normalized signal sequence, Or (f) The number of samples whose absolute value of samples is greater than or equal to a predetermined value among all samples included in the first range to the n-th range of the quantized normalized signal sequence, The absolute value of the sample among all the samples included in the quantized normalized signal sequence is set to a maximum number of samples that is n or less of N / N of the number of samples that are greater than or equal to the predetermined value. , Seeking The quantized normalized signal sequence is determined by setting a range other than the first range to the (N ⁇ 1) th range in the quantized normalized signal sequence as the
  • the classification processing exemplified above realizes classification based on “a criterion for classifying so that the number of significant samples included in each range is as equal as possible” by a method of sequentially determining each range. . According to the classification process exemplified above, it is possible to realize classification according to “a criterion for classifying so that the number of significant samples included in each range is as equal as possible” with a small amount of calculation processing.
  • FIG. 1 A first example of the sorting process will be described with reference to FIGS. 4, 5, and 6.
  • FIG. The sorting process of the first example corresponds to the above (a).
  • the division processing of the first example is performed by changing the nth range (n is an integer from 1 to N ⁇ 1) of the quantized normalized signal sequence to the nth range from the first range of the quantized normalized signal sequence.
  • the number of samples whose sample energy is greater than or equal to or greater than or equal to a predetermined value among all the samples included up to the range of, and the sample energy of all samples included in the quantized normalized signal sequence The number n of samples that are greater than or equal to a predetermined value and n of the number of samples are determined to be closest to each other, and the first normalized range of the quantized normalized signal sequence is other than the N ⁇ 1th range. Is defined as the Nth range of the quantized normalized signal sequence, thereby dividing the quantized normalized signal sequence into N ranges.
  • Quantization normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L max ⁇ ] to be classified is changed to X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L mid -1 ⁇ . ]
  • X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L mid ,..., L max ⁇ ] are divided into two ranges, specifically, the first range is the low range and the second range.
  • L mid which is the sample number on the lowest frequency side of the second range, is determined as information representing the boundary with the high frequency range.
  • f count ( ⁇ ) is determined by equation (2).
  • the f count ( ⁇ ) for each index ⁇ includes the energy of the sample corresponding to the index ⁇ of the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ]
  • 2 is set to 1 for information indicating that “sample energy
  • 0 is set as information indicating that “the sample energy
  • the predetermined value is arbitrarily set to a minute amount ⁇ ( ⁇ is a value of 0 or more).
  • quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min , ..., L max ⁇ ] first range outside the range of, namely, X ⁇ Q [ ⁇ ⁇ L mid, ..., L max ⁇ ] Is the second range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into two ranges.
  • Sorting information identifying circuit 150 outputs may be the L mid, may be a value obtained by calculating the predetermined value in the L mid, sample number of the first range L mid -1-L It may be min +1, may be the number of samples in the second range L max -L mid +1, or anything insofar as the information can identify the first range and the second range. Good.
  • FIG. 5 shows an example of dividing the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] to be divided into four ranges, specifically, the first L (1) , which is the lowest sample number of the second range, is determined as information indicating the boundary between the second range and the second range, and the boundary between the second range and the third range is determined.
  • L (2) which is the sample number on the lowest side of the third range, is determined as information to be expressed, and the lowest range of the fourth range is set as information indicating the boundary between the third range and the fourth range. This is an example of determining L (3) which is the sample number on the side.
  • the number of samples f count (where the sample energy is greater than a predetermined value among all the samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] included in the quantized normalized signal sequence f count ( L min ) +... + F count (L max ) (ie, half) and all of the quantized normalized signal sequences included in the first range to the second range.
  • L (2) be the sample number on the lowest side of the third range.
  • X ⁇ Q [ ⁇ ⁇ L (1), ..., L (2) -1 ⁇ ] is determined as the second range.
  • the number of samples f count (where the sample energy is greater than a predetermined value among all the samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] included in the quantized normalized signal sequence f count ( L min ) +... + F count (L max ) and the energy of the sample is predetermined among all samples included in the quantized normalized signal sequence from the first range to the third range.
  • the number of samples larger than the value f count (L min ) + ... + f count (L (3) -1) and L (3) obtained so as to minimize the difference value (absolute value of the difference ) The sample number on the lowest side of the range of.
  • X ⁇ Q [ ⁇ ⁇ L (2), ..., L (3) -1 ⁇ ] is determined as the third range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is in a range other than the first to third ranges, that is, X ⁇ Q [ ⁇ ⁇ L ( 3) ,..., L max ⁇ ] is the fourth range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into four ranges.
  • Sorting information identifying circuit 150 outputs may be a L (1) and L (2) and L (3), predetermined to each of the L (1) and L (2) and L (3)
  • the calculated value may be the number of samples in each range, or anything insofar as it is information that can identify all four ranges.
  • FIG. 6 shows an example of dividing the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] to be divided into N ranges, specifically, the nth
  • L (n) which is a sample number on the lowest side of the (n + 1) th range, is determined as information indicating the boundary between the (n + 1) th range and the (n + 1) th range.
  • L min is assumed to be L (0) .
  • L (n) obtained so that the difference value (absolute value) from + f count (L (n) -1) is minimized is set to the sample number on the lowest side of the ( n + 1 ) th range. .
  • X ⁇ Q [ ⁇ ⁇ L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ] is determined as the nth range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is in a range other than the first range to the (N ⁇ 1) th range, that is, X ⁇ Q [ ⁇ ⁇ L (N ⁇ 1) ,..., L max ⁇ ] is the Nth range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into N ranges.
  • Sorting information identifying circuit 150 is output, L (n) (n is the integer from 1 to N-1) may be a, L (n) (each n is an integer from 1 to N-1 ) May be a value obtained by calculating a predetermined value, may be the number of samples in each range, or may be anything as long as it is information that can specify all N ranges.
  • the second example of the sorting process corresponds to the above (b).
  • 2 ” in the classification process of the first example is replaced with “absolute value of sample
  • This is the same method as the sorting process in the first example.
  • it is possible to perform the sorting process with a smaller amount of calculation processing than the sorting process of the first example because the square calculation performed in the sorting process of the first example can be omitted.
  • FIG. 7 A third example of the sorting process will be described with reference to FIGS. 7, 8, and 9.
  • FIG. The classification process of the third example corresponds to the above (c).
  • the segmenting process of the third example includes the nth range (n is an integer from 1 to N ⁇ 1) of the quantized normalized signal sequence, and the nth range from the first range of the quantized normalized signal sequence.
  • the number of samples whose sample energy is greater than or equal to or greater than or equal to a predetermined value among all samples included up to the range of is the sample energy of all samples included in the quantized normalized signal sequence.
  • the minimum number of samples that is greater than or equal to n / N of the number of samples that is greater than or equal to a predetermined value is obtained, and from the first range of the quantized normalized signal sequence to the N ⁇ th.
  • Quantization normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L max ⁇ ] to be classified is changed to X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L mid -1 ⁇ . ]
  • X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L mid ,..., L max ⁇ ] are divided into two ranges, specifically, the first range is the low range and the second range.
  • L mid which is the sample number on the lowest frequency side of the second range, is determined as information representing the boundary with the high frequency range.
  • the number k of the discrete frequency index ⁇ is increased from L min in order, and the sample energy is greater than a predetermined value among all samples included in the quantized normalized signal sequence from L min to the index k.
  • number f count (L min) + ... + f count (k) is (f count (L min) + ... + f count (L max)) / 2 greater than or equal whether whether it is determined, for the first time f count ( L min ) + ... + f count (k) is defined as the first range up to a discrete frequency index k where (f count (L min ) + ...
  • the first range is determined as X ⁇ Q [ ⁇ ⁇ L min ,..., L mid ⁇ 1 ⁇ ].
  • quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min , ..., L max ⁇ ] first range outside the range of, namely, X ⁇ Q [ ⁇ ⁇ L mid, ..., L max ⁇ ] Is the second range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into two ranges.
  • Sorting information identifying circuit 150 outputs may be the L mid, may be a value obtained by calculating the predetermined value in the L mid, number of samples L mid -L min of the first range Alternatively, the number of samples in the second range may be L max ⁇ L mid +1. In short, any information that can identify the first range and the second range may be used.
  • FIG. 8 shows an example in which the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] to be classified is divided into four ranges, specifically, the first L (1) , which is the lowest sample number of the second range, is determined as information indicating the boundary between the second range and the second range, and the boundary between the second range and the third range is determined.
  • L (2) which is the sample number on the lowest side of the third range, is determined as information to be expressed, and the lowest range of the fourth range is set as information indicating the boundary between the third range and the fourth range. This is an example of determining L (3) which is the sample number on the side.
  • the energy of the sample among all the samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L (1) -1 ⁇ ] included in the first range of the quantized normalized signal sequence is The number of samples greater than a predetermined value f count (L min ) +... + F count (L (1) ⁇ 1) is equal to all samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min, ..., L max ⁇ number f count (L min) of the sample energy is greater than a predetermined value of a sample of the + ...
  • f count ( ⁇ ) is determined by equation (2). Then, let f count (L min ) +... + F count (L max ) be F.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is in a range other than the first to third ranges, that is, X ⁇ Q [ ⁇ ⁇ L ( 3) ,..., L max ⁇ ] is the fourth range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into four ranges.
  • Sorting information identifying circuit 150 outputs may be a L (1) and L (2) and L (3), predetermined to each of the L (1) and L (2) and L (3)
  • the calculated value may be the number of samples in each range, or anything insofar as it is information that can identify all four ranges.
  • FIG. 9 shows an example of dividing the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] to be divided into N ranges, specifically, the nth
  • L (n) which is a sample number on the lowest side of the (n + 1) th range, is determined as information indicating the boundary between the (n + 1) th range and the (n + 1) th range.
  • This process can be specifically realized by, for example, the following.
  • f count ( ⁇ ) is determined by equation (2). Then, let f count (L min ) +... + F count (L max ) be F.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is in a range other than the first range to the (N ⁇ 1) th range, that is, X ⁇ Q [ ⁇ ⁇ L (n) ,..., L max ⁇ ] is the Nth range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into N ranges.
  • Sorting information identifying circuit 150 is output, L (n) (n is the integer from 1 to N-1) may be a, L (n) (each n is an integer from 1 to N-1 ) May be a value obtained by calculating a predetermined value, may be the number of samples in each range, or may be anything as long as it is information that can specify all N ranges.
  • the fourth example of the sorting process corresponds to the above (d).
  • 2 ” in the classification process of the third example is replaced with “absolute value of sample
  • a fifth example of the sorting process will be described with reference to FIG. 10, FIG. 11, and FIG.
  • the classification process of the fifth example corresponds to the above (e).
  • the partition processing of the fifth example is performed by changing the nth range (n is an integer from 1 to N ⁇ 1) of the quantized normalized signal sequence to the nth range from the first range of the quantized normalized signal sequence.
  • the number of samples whose sample energy is greater than or equal to or greater than or equal to a predetermined value among all samples included up to the range of is the sample energy of all samples included in the quantized normalized signal sequence.
  • Quantization normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L max ⁇ ] to be classified is changed to X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L mid -1 ⁇ . ]
  • X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L mid ,..., L max ⁇ ] are divided into two ranges, specifically, the first range is the low range and the second range.
  • L mid which is the sample number on the lowest frequency side of the second range, is determined as information representing the boundary with the high frequency range.
  • the number k of the discrete frequency index ⁇ is increased from L min in order, and the sample energy is greater than a predetermined value among all samples included in the quantized normalized signal sequence from L min to the index k.
  • number f count (L min) + ... + f count (k) is (f count (L min) + ... + f count (L max)) / 2 than it is determined whether or not large, for the first time f count (L min ) + ... + f count (k) is k-1 less than the index k of the discrete frequency where (f count (L min ) + ...
  • the first range is determined as X ⁇ Q [ ⁇ ⁇ L min ,..., L mid ⁇ 1 ⁇ ].
  • quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min , ..., L max ⁇ ] first range outside the range of, namely, X ⁇ Q [ ⁇ ⁇ L mid, ..., L max ⁇ ] Is the second range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into two ranges.
  • Sorting information identifying circuit 150 outputs may be the L mid, may be a value obtained by calculating the predetermined value in the L mid, number of samples L mid -L min of the first range Alternatively, the number of samples in the second range may be L max ⁇ L mid +1. In short, any information that can identify the first range and the second range may be used.
  • FIG. 11 shows an example of dividing the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] to be divided into four ranges, specifically, the first L (1) , which is the lowest sample number of the second range, is determined as information indicating the boundary between the second range and the second range, and the boundary between the second range and the third range is determined.
  • L (2) which is the sample number on the lowest side of the third range, is determined as information to be expressed, and the lowest range of the fourth range is set as information indicating the boundary between the third range and the fourth range. This is an example of determining L (3) which is the sample number on the side.
  • This process can be specifically realized by, for example, the following.
  • f count ( ⁇ ) is determined by equation (2). Then, let f count (L min ) +... + F count (L max ) be F.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is in a range other than the first to third ranges, that is, X ⁇ Q [ ⁇ ⁇ L ( 3) ,..., L max ⁇ ] is the fourth range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into four ranges.
  • Sorting information identifying circuit 150 outputs may be a L (1) and L (2) and L (3), predetermined to each of the L (1) and L (2) and L (3)
  • the calculated value may be the number of samples in each range, or anything insofar as it is information that can identify all four ranges.
  • FIG. 12 shows an example of dividing the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] to be divided into N ranges, specifically, the nth
  • L (n) which is a sample number on the lowest side of the (n + 1) th range, is determined as information indicating the boundary between the (n + 1) th range and the (n + 1) th range.
  • + f count (L (n) ) is all samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L max ⁇ ] included in the quantized normalized signal sequence the first range of n content of greater than n, and quantized normalized signal sequence number energy of the sample is larger samples than the predetermined value f count (L min) + ... + f count (L max) of Out of all the samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L (n) -1 ⁇ ] included in the n-th range, the sample energy is larger than a predetermined value The number of samples f count (L min ) + ...
  • + f count (L (n) ⁇ 1) is equal to all samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L max ⁇ ] energy of the sample is less than or equal to n divided by n of the number of larger samples than the predetermined value f count (L min) + ... + f count (L max) of the, L (n) of the Obtained as the sample number at the lowest side of the range of n + 1.
  • This process can be specifically realized by, for example, the following.
  • f count ( ⁇ ) is determined by equation (2). Then, let f count (L min ) +... + F count (L max ) be F.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is in a range other than the first range to the (N ⁇ 1) th range, that is, X ⁇ Q [ ⁇ ⁇ L (N ⁇ 1) ,..., L max ⁇ ] is the Nth range.
  • the quantized normalized signal sequence X ⁇ Q [ ⁇ ⁇ L min ,..., L max ⁇ ] is divided into N ranges.
  • Sorting information identifying circuit 150 is output, L (n) (n is the integer from 1 to N-1) may be a, L (n) (each n is an integer from 1 to N-1 ) May be a value obtained by calculating a predetermined value, may be the number of samples in each range, or may be anything as long as it is information that can specify all N ranges.
  • the sixth example of the sorting process corresponds to the above (f).
  • 2 ” in the classification process of the fifth example is replaced with “absolute value of sample
  • This is the same method as the sorting process in the fifth example.
  • it is possible to perform the sorting process with a smaller calculation processing amount than the sorting process of the fifth example, because the square calculation performed in the sorting process of the fifth example can be omitted.
  • the gain correction amount encoding unit 140 uses a gain correction amount codebook stored in a storage unit (not shown) to correct the quantized global gain g ⁇ for each divided range with a gain correction amount.
  • the signal sequence obtained by multiplying the gain and the value of each sample of the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] and the input signal sequence X ( ⁇ )
  • a gain correction amount code which is a code for specifying a gain correction amount for each divided range, in which the correlation with [ ⁇ ⁇ L min ,..., L max ⁇ ] is maximum or the error is minimum is output.
  • the first example of the gain correction amount encoding process is an example in which the gain obtained by adding the quantized global gain g ⁇ and the gain correction amount is used as the correction gain.
  • a gain correction amount candidate ⁇ low (m) in the first range, a gain correction amount candidate ⁇ high (m) in the second range, and codes for specifying these gain correction amount candidates M sets of idx (m) are stored (M is a predetermined integer of 2 or more). Specifically, a set of ⁇ low (1), ⁇ high (1) and idx (1), a set of ⁇ low (2), ⁇ high (2) and idx (2), ..., ⁇ A set of low (M), ⁇ high (M), and idx (M) is stored in the storage unit as a gain correction amount codebook.
  • the gain correction amount encoding unit 140 adds a quantized global gain g ⁇ and a gain correction amount candidate ⁇ low (m) in the first range for each m from 1 to M.
  • the first range of quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L mid -1 ⁇ ]
  • the input signal series X ( ⁇ ) in the range of 1 [ ⁇ ⁇ L min ,..., L mid ⁇ 1 ⁇ ] and the sum of squares of the difference between corresponding samples, the quantized global gain g ⁇ and the second range of the gain correction amount candidate ⁇ high (m) and a value obtained by adding the second range of quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L mid, ..., L max ⁇ ] multiplied by the value of each sample and the corresponding samples of the second range of input signal sequences X ( ⁇ ) [ ⁇ ⁇ L mid ,..., L max ⁇ ]
  • a code idx (m) corresponding to m that minimizes the added value is output as a gain correction amount code idx. That is, the gain correction amount code idx is obtained by the equation (13).
  • equation (13) corresponds to vector quantization based on the criterion that minimizes the error, but vector quantization based on the criterion that maximizes the correlation, and the criterion based on the criterion that minimizes the error or maximizes the correlation.
  • a technique such as scalar quantization may be applied.
  • a gain correction amount candidate ⁇ 1 (m) in the first range, a gain correction amount candidate ⁇ 2 (m) in the second range, and a gain correction amount candidate ⁇ in the third range are stored.
  • ⁇ 1 (1), ⁇ 2 (1), ⁇ 3 (1), ⁇ 4 (1), and idx (1) pairs, ⁇ 1 (2), ⁇ 2 (2), and ⁇ 3 (2), ⁇ 4 (2), and idx (2), ..., ⁇ 1 (M), ⁇ 2 (M), ⁇ 3 (M), ⁇ 4 (M), and idx (M ) Is stored in the storage unit as a gain correction amount code book.
  • the gain correction amount coding unit first adds a value obtained by adding the quantized global gain g ⁇ and the gain correction amount candidate ⁇ 1 (m) in the first range for each of 1 to M.
  • a signal sequence obtained by multiplying the value of each sample of the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L (1) ⁇ 1 ⁇ ] in the first range; The sum of squares of the difference between the corresponding samples of the input signal sequence X ( ⁇ ) [ ⁇ ⁇ L min ,..., L (1) ⁇ 1 ⁇ ] in the first range and the quantized global gain g ⁇
  • a code idx (m) corresponding to m that minimizes the added value is output as a gain correction amount code idx. That is, the gain correction amount code idx is obtained by the equation (15).
  • Equation (15) corresponds to the vector quantization based on the criterion that minimizes the error, but the vector quantization based on the criterion that maximizes the correlation and the criterion based on the criterion that minimizes the error or maximizes the correlation.
  • a technique such as scalar quantization may be applied.
  • gain correction amount candidates ⁇ 1 (m),..., ⁇ N (m) in the first to Nth ranges and codes idx () for specifying these gain correction amount candidates M sets (m) are stored (M is a predetermined integer of 2 or more). Specifically, a set of ⁇ 1 (1), ..., ⁇ N (1) and idx (1), a set of ⁇ 1 (2), ..., ⁇ N (2) and idx (2), ..., ⁇ 1 (M),..., ⁇ N (M) and idx (M) are stored in the storage unit as a gain correction amount code book.
  • the gain correction amount encoding unit firstly, for each m of 1 to M, the quantized global gain g ⁇ for each of the first range to the Nth range and the gain correction amount candidate ⁇ for the nth range.
  • the added value is obtained by equation (16).
  • a code idx (m) corresponding to m that minimizes the added value is output as a gain correction amount code idx. That is, the gain correction amount code idx is obtained by the equation (17).
  • equation (17) corresponds to vector quantization based on the criterion that minimizes the error.
  • the vector quantization based on the criterion that maximizes the correlation and the criterion based on the criterion that minimizes the error or maximizes the correlation.
  • a technique such as scalar quantization may be applied.
  • the quantized global gain g ⁇ and the square sum of the values of all samples in the frame of the quantized normalized signal sequence are divided into quantized normalized signal sequences.
  • a value obtained by multiplying the gain correction amount by a value obtained by dividing the value of all the samples within the range by the sum of squares is used as the correction gain.
  • the storage unit (not shown) stores the gain correction amount candidates ⁇ 1 (m),..., ⁇ N (m) of the first to Nth ranges and the gain correction amounts.
  • M sets (M is a predetermined integer equal to or greater than 2) are stored with a set of codes idx (m) for specifying candidates.
  • a set of ⁇ 1 (1), ..., ⁇ N (1) and idx (1), a set of ⁇ 1 (2), ..., ⁇ N (2) and idx (2), ..., ⁇ 1 (M),..., ⁇ N (M) and idx (M) are stored in the storage unit as a gain correction amount code book.
  • the gain correction amount encoding unit For each of the first range to the Nth range, all samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (N) ⁇ 1 ⁇ in the frame of the quantized normalized signal sequence ] All the samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ within the partitioned range of the quantized normalized signal sequence ], When the value obtained by dividing the sum of squares by s (n) is defined as s (n), the gain correction amount encoding unit firstly, for each m of 1 to M, for each of the first range to the Nth range.
  • a code idx (m) corresponding to m that minimizes the added value is output as a gain correction amount code idx. That is, the gain correction amount code idx can be obtained from the equation (20).
  • equation (20) corresponds to vector quantization based on the criterion that minimizes the error, but vector quantization based on the criterion that maximizes the correlation, and the criterion based on the criterion that minimizes the error or maximizes the correlation.
  • a technique such as scalar quantization may be applied.
  • the quantized global gain g ⁇ and the number of samples in which the energy of the sample in the frame of the quantized normalized signal sequence is larger than a predetermined value are represented by the quantized normalized signal.
  • a correction gain is obtained by adding a value obtained by multiplying a gain correction amount by a value obtained by dividing the energy of a sample within a series-divided range by the number of samples greater than a predetermined value.
  • the storage unit (not shown) stores the gain correction amount candidates ⁇ 1 (m),..., ⁇ N (m) of the first to Nth ranges and the gain correction amounts.
  • M sets (M is a predetermined integer equal to or greater than 2) are stored with a set of codes idx (m) for specifying candidates.
  • a set of ⁇ 1 (1), ..., ⁇ N (1) and idx (1), a set of ⁇ 1 (2), ..., ⁇ N (2) and idx (2), ..., ⁇ 1 (M),..., ⁇ N (M) and idx (M) are stored in the storage unit as a gain correction amount code book.
  • the predetermined value may be 0 or a value greater than or equal to 0, or may be a value obtained by multiplying the quantized global gain g ⁇ by a predetermined value ⁇ .
  • the gain correction amount encoding unit 140 first, for each m of 1 to M, the quantized global gain g ⁇ for each of the first range to the Nth range and the gain correction amount for the nth range. And a value obtained by adding the product of the candidates ⁇ n (m) and s (n), and the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L ( n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ] multiplied by the value of each sample and the input signal sequence X ( ⁇ ) [ ⁇ ⁇ L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ], the sum of squares of the difference between the values of the corresponding samples is obtained for the first range to the Nth range.
  • the added value is obtained by equation (22).
  • a code idx (m) corresponding to m that minimizes the added value is output as a gain correction amount code idx. That is, the gain correction amount code idx is obtained by the equation (23).
  • equation (23) corresponds to vector quantization based on the criterion that minimizes the error, but vector quantization based on the criterion that maximizes the correlation, and the criterion based on the criterion that minimizes the error or maximizes the correlation.
  • a technique such as scalar quantization may be applied.
  • the pre-quantization width multiplication gain candidates ⁇ 1 (m),..., ⁇ N (m) of the first range to the Nth range and their pre-quantization width multiplication gains are stored.
  • M sets (M is a predetermined integer equal to or greater than 2) are stored as M sets of codes idx (m) that specify correction amount candidates.
  • a set of ⁇ 1 (1), ..., ⁇ N (1) and idx (1), a set of ⁇ 1 (2), ..., ⁇ N (2) and idx (2), ..., ⁇ 1 (M),..., ⁇ N (M) and idx (M) are stored in the storage unit as a gain correction amount code book.
  • the gain correction amount encoding unit 140 uses the quantization step width step corresponding to the quantized global gain g ⁇ to first, for each m from 1 to M, for each of the first range to the Nth range.
  • a code idx (m) corresponding to m that minimizes the added value is output as a gain correction amount code idx. That is, the gain correction amount code idx is obtained by the equation (25).
  • equation (25) corresponds to vector quantization based on the criterion that minimizes the error, but vector quantization based on the criterion that maximizes the correlation, and the criterion based on the criterion that minimizes the error or maximizes the correlation.
  • a technique such as scalar quantization may be applied.
  • the quantization step width step corresponding to the quantized global gain g ⁇ is a difference value between adjacent candidates of the quantized global gain in the global gain encoder 105.
  • the quantization step width step and the product of the quantization width pre-multiplication gain correction amount candidate ⁇ n (m) whose absolute value is less than 1 are added to the quantized global gain g ⁇ .
  • the pre-quantization width multiplication gain candidate ⁇ n (m) included in the gain correction amount codebook stored in a storage unit may be generated by learning. In this case, but it may also include those that are not less than 1 candidate before quantization width multiplied gain correction amount ⁇ n (m). However, even contain even those that are not less than 1 to the quantization width multiplied before gain correction amount candidate delta n (m), the candidate quantization step width step and the quantization width multiplied before gain correction amount delta n ( The distance between the quantized global gain g ⁇ and the adjacent quantized global gain candidates is corrected by correcting the quantized global gain so that the product of m) is added to the quantized global gain g ⁇ . That is, correction depending on the quantization step width can be performed on the quantization global gain.
  • the modified example of the fourth example obtains a gain correction amount depending on the quantization step width corresponding to the quantized global gain g ⁇ , and the quantized global gain g ⁇ and all the quantized normalized signal sequences in the frame. Corrected by adding the gain correction amount multiplied by the value obtained by dividing the sum of squares of the sample values by the sum of squares of the values of all the samples in the divided range of the quantized normalized signal sequence. Either the gain or the quantized global gain g ⁇ and the number of samples in which the energy of the sample in the frame of the quantized normalized signal sequence is greater than a predetermined value is divided into the quantized normalized signal sequence. In this example, the gain obtained by adding the gain correction amount to the value obtained by dividing the energy of the sample within the range by the number of samples larger than a predetermined value is used as the correction gain.
  • the absolute values of all the values of the gain correction amount candidates ⁇ n (m) before quantization width multiplication included in the gain correction amount codebook stored in a storage unit (not shown) are set to 1 / N Keep it below. That is, the gain correction amount codebook stores candidates for gain correction amount before quantization width multiplication as ⁇ n (m).
  • the addition value is obtained by the expression (26) using s (n) of the second example or the third example, and the expression is replaced by the expression (25) of the fourth example.
  • the gain correction amount code idx is obtained from (27).
  • equation (27) corresponds to vector quantization based on the criterion that minimizes the error, but vector quantization based on the criterion that maximizes the correlation, and the criterion based on the criterion that minimizes the error or maximizes the correlation.
  • a technique such as scalar quantization may be applied.
  • the average value of s (n) in the second example or the third example is N. Therefore, for example, the product of the quantization step width step and the candidate ⁇ n (m) of the gain correction amount before the quantization width whose absolute value is less than 1 / N and s (n) whose average value is N is quantized.
  • the gain obtained by correcting the quantized global gain g ⁇ with the gain correction amount is the quantized global gain g ⁇ . It is possible to be between quantized global gain candidates adjacent to.
  • the pre-quantization width multiplication gain candidate ⁇ n (m) included in the gain correction amount codebook stored in a storage unit may be generated by learning.
  • the gain correction amount candidate before quantization width multiplication ⁇ n (m) is not less than 1 / N.
  • the quantization step width step and the gain correction amount candidate ⁇ before quantization width multiplication ⁇ By correcting the quantized global gain so that the product of n (m) and s (n) is added to the quantized global gain g ⁇ , the quantized global gain g ⁇ and the adjacent quantized gain Correction depending on the distance to the global gain candidate, that is, the quantization step width can be performed on the quantized global gain.
  • combination part 160 is recorded on a recording medium, and the said information read from the said recording medium Is also allowed to be input to the decoding device 2.
  • the decoding device 2 (see FIG. 13) of the first embodiment includes a normalized signal decoding unit 107, a global gain decoding unit 106, a gain correction amount decoding unit 230, a decoded signal sequence generation unit 250, and a sorting unit 260.
  • the decoding device 2 may include a separation unit 210 and a time domain conversion unit 270 as necessary.
  • the bit stream transmitted from the encoding device 1 is input to the decoding device 2.
  • the separation unit 210 extracts a normalized signal code, a global gain code, and a gain correction amount code from the bit stream.
  • ⁇ Normalized signal decoding unit 107 A normalized signal code is input to the normalized signal decoding unit 107.
  • the normalized signal decoding unit 107 applies a decoding method corresponding to the encoding method performed by the normalized signal encoding unit 120 of the encoding device 1 to decode the normalized signal code and decode the normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] is obtained (step S1d).
  • ⁇ represents an index of discrete frequency
  • the normalized signal decoding unit 107 performs the same operation as the normalized signal decoding unit 107 of FIG. 1 described in the “Background Art” column.
  • the sorting unit 260 sorts the decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] so that the number of significant samples included in each range is as equal as possible.
  • N is a predetermined integer greater than or equal to 2
  • the set of discrete frequency indexes of the decoded normalized signal sequence X ⁇ Q ( ⁇ ) is ⁇ L min , ..., L max ⁇
  • the decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min, ...
  • the number of partitions N is set in advance in, for example, the partition unit 150 of the encoding device 1 and the partition unit 260 of the decoding device 2 so as to have a value common to the number of partitions N in the partition unit 150 of the encoding device 1. .
  • the partitioning process performed by the segmenting unit 260 on the decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] is performed by the segmenting unit 150 of the encoding device 1 by quantization normalization.
  • the segmentation unit 150 and the decoding device 2 of the encoding device 1 are performed so that the same processing as the segmentation processing performed on the completed signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] is performed. Is set in advance with the sorting unit 260.
  • ⁇ Details of Sorting Process Performed by Sorting Unit 260 The partitioning process performed by the segmenting unit 260 on the decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] is performed by the segmenting unit 150 of the encoding device 1 by quantization normalization. This is the same as the segmentation process performed on the completed signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ]. That is, the sorting process based on the “criteria for sorting so that the number of significant samples included in each range is as equal as possible” is, for example, the n-th range (n is 1 to N ⁇ 1) of the decoded normalized signal sequence.
  • the number of samples whose absolute value of samples is greater than or equal to a predetermined value among all samples included in the first range to the nth range of the decoded normalized signal sequence is decoded
  • the absolute value of the sample among all the samples included in the normalized signal sequence is the minimum number of samples that is n or more of N / N of the number of samples that is greater than or equal to the predetermined value.
  • the number of samples whose sample energy is greater than or equal to a predetermined value among all samples included in the first range to the n-th range of the decoded normalized signal sequence is the decoding normal
  • the maximum number of samples is less than n / N of the number of samples whose sample energy is greater than or equal to or greater than a predetermined value among all the samples included in the digitized signal sequence.
  • the number of samples whose absolute value of samples is greater than or equal to a predetermined value among all samples included in the first range to the nth range of the decoded normalized signal sequence is decoded
  • the absolute value of the sample among all the samples included in the normalized signal sequence is the maximum number of samples that is n or less than N / N of the number of samples that is greater than or equal to the predetermined value. Seeking By setting a range other than the first range to the (N-1) th range in the decoded normalized signal sequence as the Nth range of the decoded normalized signal sequence, N decoded normalized signal sequences are obtained. This is done by dividing the range.
  • the classification process exemplified above realizes the classification based on the “criteria for classifying so that the number of significant samples included in each range is as equal as possible” by a method of sequentially determining each range. . According to the classification process exemplified above, it is possible to realize classification according to “a criterion for classifying so that the number of significant samples included in each range is as equal as possible” with a small amount of calculation processing.
  • Specific examples of the sorting process performed by the sorting unit 260 are specific examples of “first example of sorting process” to “sixth example of sorting process” that are specific examples of the sorting process performed by the sorting unit 150 of the encoding device 1.
  • the quantized normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] is decoded and decoded normalized signal series X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min , ..., L max ⁇ ].
  • ⁇ Gain correction amount decoding unit 230 A gain correction amount code is input to the gain correction amount decoding unit 230.
  • the gain correction amount decoding unit 230 decodes the gain correction amount code to obtain a gain correction amount corresponding to each divided range (step S3d).
  • gain correction amount candidates ⁇ 1 (m),..., ⁇ N (m) in the first to Nth ranges and codes idx () for specifying these gain correction amount candidates M sets (m) are stored (M is a predetermined integer of 2 or more).
  • M is a predetermined integer of 2 or more.
  • a set of ⁇ 1 (1), ..., ⁇ N (1) and idx (1), a set of ⁇ 1 (2), ..., ⁇ N (2) and idx (2), ..., ⁇ 1 (M),..., ⁇ N (M) and idx (M) are stored in the storage unit as a gain correction amount code book.
  • the gain correction amount code book stored in the storage unit is the same as the gain correction amount code book stored in the storage unit of the encoding device 1.
  • the gain correction amount decoding unit 230 refers to the gain correction amount code book, and from the first range associated with idx (I) that is the same code as the gain correction amount code idx in the gain correction amount code book. Gain correction amounts ⁇ 1 (I),..., ⁇ N (I) corresponding to the respective Nth ranges are obtained.
  • the decoding process performed by the gain correction amount decoding unit 230 is basically a process corresponding to the encoding process performed by the gain correction amount encoding unit 140. Further, it may be a vector quantization decoding process or a scalar quantization decoding process.
  • a global gain code is input to the global gain decoding unit 106.
  • the global gain decoding unit 160 decodes the global gain code and outputs a decoded global gain g ⁇ (step S4d).
  • the decoding process performed by the global gain decoding unit 106 is a process corresponding to the encoding process performed by the global gain encoding unit 105, and the global gain decoding unit in the [Background Technology] column. This is a well-known technique as described in 106.
  • ⁇ Decoded signal sequence generation unit 250> The decoded signal sequence generation unit 250, a decoding normalized signal sequence X ⁇ Q ( ⁇ ), and the gain correction amount delta n (I), a decoding global gain g ⁇ , classification information is input.
  • the decoded signal sequence generation unit 250 corrects the decoding gain obtained by correcting the decoding global gain g ⁇ by the gain correction amount ⁇ n (I) and decoding for each range obtained by the processing of ⁇ Step S2d> in the sorting unit 260.
  • a signal sequence obtained by multiplying the value of each sample of the normalized signal sequence X ⁇ Q ( ⁇ ) is output as an output signal sequence X ⁇ ( ⁇ ) (step S5d). Since this output signal sequence X ⁇ ( ⁇ ) corresponds to the input signal sequence X ( ⁇ ) of the encoding device 1, it can also be said to be a decoded signal sequence.
  • the first example of the decoded signal sequence generation processing is an example in which a correction gain is obtained by adding the decoded global gain g ⁇ and the gain correction amount.
  • the decoded signal sequence generation unit 250 includes a decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] and a gain correction amount ⁇ n (I) [n ⁇ ⁇ 1, .., N ⁇ ], the decoding global gain g ⁇ , and the classification information.
  • the sample number on the lowest side of the nth range specified by the classification information is L (n-1)
  • L min is L (0)
  • L max is L (N) -1.
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (n ⁇ 1) , .., L (n) -1 ⁇ ] represents each sample of the correction gain obtained by correcting the decoded global gain g ⁇ with the gain correction amount ⁇ n (I) and the decoded normalized signal sequence X ⁇ Q ( ⁇ ). It is obtained by multiplying by the value of. That is, each sample value of the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ] in the n-th range is obtained by Expression (28).
  • X ⁇ ( ⁇ ) (g ⁇ + ⁇ n (I)) X ⁇ Q ( ⁇ ) (28)
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (1) ⁇ 1 ⁇ ], X ⁇ ( ⁇ ) [ ⁇ ⁇ L (1) , ..., L (2) -1 ⁇ ], ..., X ⁇ ( ⁇ ) [ ⁇ ⁇ L (N-1) , ..., L (N) -1 ⁇ ],
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (N) ⁇ 1 ⁇ ] that is, X ⁇ ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ].
  • the decoded global gain g ⁇ and the sum of squares of the values of all the samples in the frame of the decoded normalized signal sequence are included in the divided range of the decoded normalized signal sequence.
  • a value obtained by multiplying a value obtained by dividing the value of all the samples by the sum of squares by a gain correction amount is added as a correction gain.
  • the decoded signal sequence generation unit 250 includes a decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] and a gain correction amount ⁇ n (I) [n ⁇ ⁇ 1, .., N ⁇ ], the decoding global gain g ⁇ , and the classification information.
  • the sample number on the lowest side of the nth range specified by the classification information is L (n-1)
  • L min is L (0)
  • L max is L (N) -1
  • the classification part Each of the first range to the Nth range obtained in 260 will be described as the nth range.
  • the decoding global gain g ⁇ and the multiplication value of the nth range gain correction amount ⁇ n (I) and s (n) are added.
  • the signal sequence obtained in this way is the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ] in the n-th range .
  • each sample value of the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ] in the n-th range is obtained by Expression (30).
  • X ⁇ ( ⁇ ) (g ⁇ + s (n) ⁇ n (I)) X ⁇ Q ( ⁇ ) (30)
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (1) -1 ⁇ ], X ⁇ ( ⁇ ) [ ⁇ ⁇ L (1) , ..., L (2) -1 ⁇ ], ..., X ⁇ ( ⁇ ) [ ⁇ ⁇ L (N-1) , ..., L (N) -1 ⁇ ],
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (N) ⁇ 1 ⁇ ] that is, X ⁇ ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ].
  • the decoded global gain g ⁇ and the number of samples in which the energy of the samples in the frame of the decoded normalized signal sequence is larger than a predetermined value are classified into the decoded normalized signal sequence.
  • the gain obtained by adding the value obtained by multiplying the gain correction amount by the value obtained by dividing the energy of the sample within the range by the number of samples larger than the predetermined value is used as the correction gain.
  • the decoded signal sequence generation unit 250 includes a decoded normalized signal sequence X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] and a gain correction amount ⁇ n (I) [n ⁇ ⁇ 1, .., N ⁇ ], the decoding global gain g ⁇ , and the classification information.
  • the sample number on the lowest side of the nth range specified by the classification information is L (n-1)
  • L min is L (0)
  • L max is L (N) -1
  • the classification part Each of the first range to the Nth range obtained in 260 will be described as the nth range.
  • the number of samples c (n) whose energy is greater than a predetermined value is obtained. Further, the sum of c (1) to c (N) is obtained. This sum is a sample of all samples X ⁇ Q ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (N) -1 ⁇ ] of the decoded normalized signal sequence whose energy is greater than a predetermined value.
  • the number of The predetermined value may be 0 or a value greater than or equal to 0, or may be a value obtained by multiplying the decoded global gain g ⁇ by a predetermined value ⁇ .
  • the decoding global gain g ⁇ and the multiplication value of the nth range gain correction amount ⁇ n (I) and s (n) are added.
  • the signal sequence obtained in this way is the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ] in the n-th range .
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (1) ⁇ 1 ⁇ ], X ⁇ ( ⁇ ) [ ⁇ ⁇ L (1) , ..., L (2) -1 ⁇ ], ..., X ⁇ ( ⁇ ) [ ⁇ ⁇ L (N-1) , ..., L (N) -1 ⁇ ],
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (N) ⁇ 1 ⁇ ] that is, X ⁇ ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ].
  • the sample number on the lowest side of the nth range specified by the classification information is L (n-1)
  • L min is L (0)
  • L max is L (N) -1.
  • the pre-quantization width multiplication gain candidates ⁇ 1 (m),..., ⁇ N (m) of the first range to the Nth range and their pre-quantization width multiplication gains are stored.
  • M sets (M is a predetermined integer equal to or greater than 2) are stored as M sets of codes idx (m) that specify correction amount candidates.
  • a set of ⁇ 1 (1), ..., ⁇ N (1) and idx (1), a set of ⁇ 1 (2), ..., ⁇ N (2) and idx (2), ..., ⁇ 1 (M),..., ⁇ N (M) and idx (M) are stored in the storage unit as a gain correction amount code book.
  • the decoded signal sequence generation unit 250 When the decoded signal sequence generation unit 250 performs the decoded signal sequence generation processing of the fourth example, for example, candidates for gain correction amount before quantization width multiplication included in a gain correction amount codebook stored in a storage unit (not shown) The absolute value of all values of ⁇ n (m) is set to be less than 1.
  • the gain correction amount code book stored in the storage unit is the same as the gain correction amount code book stored in the storage unit of the encoding device 1.
  • the gain correction amount decoding unit 230 refers to the gain correction amount code book and performs gain correction in the gain correction amount code book.
  • the decoded signal sequence generation unit 250 uses the quantization step width step of the decoded global gain g ⁇ in the global gain decoding unit 106 to determine the decoded global gain g ⁇ and the first number for each of the first range to the Nth range.
  • the signal sequence obtained by multiplying the value of each sample of Q ( ⁇ ) [ ⁇ ⁇ L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ]] is the output signal sequence X ⁇ in the nth range.
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (1) ⁇ 1 ⁇ ], X ⁇ ( ⁇ ) [ ⁇ ⁇ L (1) , ..., L (2) -1 ⁇ ], ..., X ⁇ ( ⁇ ) [ ⁇ ⁇ L (N-1) , ..., L (N) -1 ⁇ ],
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (N) ⁇ 1 ⁇ ] that is, X ⁇ ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ].
  • the quantization step width step corresponding to the decoded global gain g ⁇ is a difference value between the decoded global gain and the adjacent candidates in the global gain decoding unit 106.
  • the decoding global gain is corrected by correcting the decoding global gain so that the product of this quantization step width step and ⁇ n (I) whose absolute value is less than 1 is added to the decoding global gain g ⁇ .
  • the gain obtained by correcting g ⁇ with the gain correction amount can be between the decoded global gain g ⁇ and a candidate for the decoded global gain adjacent thereto.
  • the pre-quantization width multiplication gain correction amount candidate ⁇ n (m) stored in a storage unit may be generated by learning. In this case, but it may also include those that are not less than 1 candidate before quantization width multiplied gain correction amount ⁇ n (m).
  • the candidate quantization step width step and the quantization width multiplied before gain correction amount delta n By correcting the decoding global gain so that the product of m) is added to the decoding global gain g ⁇ , the distance between the decoding global gain g ⁇ and the decoding global gain candidate adjacent thereto, that is, the quantum It is possible to perform correction depending on the conversion step width on the decoded global gain.
  • the modified example of the fourth example obtains a gain correction amount depending on the quantization step width corresponding to the decoded global gain g ⁇ , and calculates the decoded global gain g ⁇ and all the samples in the frame of the decoded normalized signal sequence. Whether the value obtained by dividing the sum of squares of the values by the sum of squares of the values of all the samples within the divided range of the decoded normalized signal sequence and the value obtained by multiplying the gain correction amount is used as the correction gain.
  • the number of samples whose energy of samples in a frame of the decoded normalized signal sequence is larger than a predetermined value is the number of samples whose energy of samples in the divided range of the decoded normalized signal sequence is larger than a predetermined value.
  • a value obtained by multiplying the value obtained by dividing the gain correction amount by the gain correction amount is used as the correction gain.
  • the pre-quantization width multiplication gain candidates ⁇ 1 (m),..., ⁇ N (m) of the first to Nth ranges and their pre-quantization width gains are stored.
  • M sets (M is a predetermined integer equal to or greater than 2) are stored as M sets of codes idx (m) that specify correction amount candidates.
  • a set of ⁇ 1 (1), ..., ⁇ N (1) and idx (1), a set of ⁇ 1 (2), ..., ⁇ N (2) and idx (2), ..., ⁇ 1 (M),..., ⁇ N (M) and idx (M) are stored in the storage unit as a gain correction amount code book.
  • the absolute values of all the values of the gain correction amount candidates ⁇ n (m) before quantization width multiplication included in the gain correction amount codebook stored in a storage unit are set to 1 / N Keep it below.
  • the gain correction amount code book stored in the storage unit is the same as the gain correction amount code book stored in the storage unit of the encoding device 1.
  • the gain correction amount decoding unit 230 refers to the gain correction amount code book and includes the gain correction amount code book.
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ in the nth range is obtained by the expression (34) using s (n) of the second example or the third example.
  • L (n ⁇ 1) ,..., L (n) ⁇ 1 ⁇ ] are obtained.
  • X ⁇ ( ⁇ ) (g ⁇ + step s (n) ⁇ n (I)) X ⁇ Q ( ⁇ ) (34)
  • the output signal sequence X ⁇ ( ⁇ ) [ ⁇ ⁇ L (0) ,..., L (N) ⁇ 1 ⁇ ] that is, X ⁇ ( ⁇ ) [ ⁇ ⁇ L min ,..., L max ⁇ ] is obtained.
  • the average value of s (n) in the second example or the third example is N. Therefore, for example, the product of the quantization step width step, the candidate ⁇ n (m) of the gain correction amount before the quantization width whose absolute value is less than 1 / N, and s (n) whose average value is N is decoded.
  • the gain obtained by correcting the decoding global gain g ⁇ with the gain correction amount is the decoding global gain g ⁇ and the decoding adjacent to the decoding global gain g ⁇ . Can be between global gain candidates.
  • the pre-quantization width multiplication gain candidate ⁇ n (m) included in the gain correction amount codebook stored in a storage unit may be generated by learning.
  • the gain correction amount candidate before quantization width multiplication ⁇ n (m) is not less than 1 / N.
  • the quantization step width step and the gain correction amount candidate ⁇ before quantization width multiplication ⁇ By correcting the decoding global gain so that the product of n (m) and s (n) is added to the decoding global gain g ⁇ , the decoding global gain g ⁇ and a decoding global gain candidate adjacent thereto are corrected. It is possible to perform correction on the decoded global gain depending on the distance between the decoded global gain and the quantization step width.
  • the output signal sequence X ⁇ ( ⁇ ) is input to the time domain conversion unit 270 provided as necessary.
  • the time domain transform unit 270 applies a frequency-time transform to the output signal sequence X ⁇ ( ⁇ ) and outputs a time domain signal sequence z F (t) in units of frames.
  • the frequency-time conversion method is an inverse conversion corresponding to the time-frequency conversion method used in the frequency domain conversion unit 101.
  • the frequency-time conversion method here is IMDCT (Inverse Modified Discrete Cosine Transform) or IDCT (Inverse Discrete Cosine Transform).
  • Second Embodiment the surplus bits of the normalized signal code are used as the gain correction amount code.
  • the number of consumed bits may be smaller than the specified number of bits. is there.
  • the normalized signal encoding unit 120 outputs the surplus bit number U obtained by subtracting the number of consumed bits from the specified number of bits to the gain correction amount encoding unit 140.
  • the gain correction amount encoding unit 140 outputs a U-bit gain correction amount code based on the input surplus bit number U.
  • the gain correction amount candidate number M used in the gain correction amount encoding unit 140 may be 2 U
  • the code idx (m) for specifying each gain correction amount candidate may be U bits.
  • the normalized signal decoding unit 107 uses the consumption bits that are the number of bits of the actual normalized signal code from the specified number of bits defined as the maximum value of the number of bits of the normalized signal code.
  • the surplus bit number U obtained by subtracting the number is output to the gain correction amount decoding unit 230.
  • the gain correction amount decoding unit 230 can decode the input U-bit gain correction amount code. Specifically, the gain correction amount candidate number M included in the gain correction amount codebook used in the gain correction amount decoding unit 230 is set to 2 U, and a code idx (m) for specifying each gain correction amount candidate is set to U bits. In other words, idx (I) having the same code as the U-bit gain correction amount code idx may be obtained.
  • bits that are prepared for the normalized signal code but are not actually used for the normalized signal code are used for the gain correction amount code.
  • the third embodiment is an example in which information corresponding to the number N of divided ranges is transmitted from the encoding device 1 to the decoding device 2.
  • the sorting unit 150 of the encoding device 1 determines the number N of ranges after sorting based on some standard or information transmitted from outside the sorting unit 150, and performs sorting processing so that the number of ranges after sorting becomes N. .
  • the division unit 150 of the encoding device 1 also outputs an auxiliary code that can specify N, which is the number of ranges after the division.
  • the sorting unit 260 of the decoding device 2 receives the auxiliary code, and performs the sorting process so that the number of ranges after the division becomes the number N specified by the auxiliary code.
  • the encoding device, the encoding method, the decoding device, and the decoding method according to the present invention are not limited to the above-described embodiments, and can be appropriately changed without departing from the spirit of the present invention. Is possible.
  • the processing described in the above embodiment may be executed not only in time series according to the order of description but also in parallel or individually as required by the processing capability of the apparatus that executes the processing. .
  • the processing functions in the encoding device / decoding device are realized by a computer
  • the processing contents of the functions that the encoding device / decoding device should have are described by a program.
  • the processing functions of the encoding device / decoding device are realized on the computer.
  • the program describing the processing contents can be recorded on a computer-readable recording medium.
  • a computer-readable recording medium for example, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used.
  • a magnetic recording device a hard disk device, a flexible disk, a magnetic tape or the like, and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only) Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical recording medium, MO (Magneto-Optical disc), etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory), etc. Can be used.
  • this program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Furthermore, the program may be distributed by storing the program in a storage device of the server computer and transferring the program from the server computer to another computer via a network.
  • a computer that executes such a program first stores a program recorded on a portable recording medium or a program transferred from a server computer in its own storage device.
  • the computer reads a program stored in its own recording medium and executes a process according to the read program.
  • the computer may directly read the program from a portable recording medium and execute processing according to the program, and the program is transferred from the server computer to the computer.
  • the processing according to the received program may be executed sequentially.
  • the program is not transferred from the server computer to the computer, and the above-described processing is executed by a so-called ASP (Application Service Provider) type service that realizes a processing function only by an execution instruction and result acquisition. It is good.
  • the program in this embodiment includes information that is used for processing by an electronic computer and that conforms to the program (data that is not a direct command to the computer but has a property that defines the processing of the computer).
  • the encoding device and the decoding device are configured by executing a predetermined program on the computer.
  • at least a part of the processing contents may be realized by hardware. Good.

Landscapes

  • Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Computational Linguistics (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Compression, Expansion, Code Conversion, And Decoders (AREA)

Abstract

L'invention concerne une méthode d'encodage composée des éléments suivants : un processus d'encodage de signal de normalisation permettant d'obtenir un code de signal de normalisation obtenu en encodant une séquence grâce à un signal dans lequel chaque échantillon de signal d'entrée compris dans la séquence de signal d'entrée a été normalisé, et d'obtenir une séquence de signal normalisé de quantification correspondant au code de signal de normalisation ; un processus d'encodage de gain global permettant d'obtenir un gain global de quantification correspondant à la séquence de signal d'entrée, et permettant d'obtenir un code de gain global correspondant au gain global de quantification ; un processus de division permettant de diviser la séquence de signal normalisé de quantification en une pluralité de plages ; et un processus d'encodage de quantité de correction de gain permettant d'obtenir un code de correction de gain qui sert à spécifier une quantité de correction de gain pour chacune des plages divisées et pour lequel la corrélation entre la séquence de signal et la séquence de signal d'entrée est la plus élevée, ou l'erreur est la plus faible, ladite corrélation étant obtenue en multipliant le gain obtenu en corrigeant le gain global de quantification pour chacune des plages divisées par une quantité de correction de gain, avec les valeurs pour chaque échantillon de la séquence de signal normalisé de quantification.
PCT/JP2013/052914 2012-02-07 2013-02-07 Méthode d'encodage, dispositif d'encodage, méthode de décodage, dispositif de décodage, programme et support d'enregistrement WO2013118835A1 (fr)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2013557574A JP5786044B2 (ja) 2012-02-07 2013-02-07 符号化方法、符号化装置、復号方法、復号装置、プログラム及び記録媒体

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2012024176 2012-02-07
JP2012-024176 2012-02-07

Publications (1)

Publication Number Publication Date
WO2013118835A1 true WO2013118835A1 (fr) 2013-08-15

Family

ID=48947590

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2013/052914 WO2013118835A1 (fr) 2012-02-07 2013-02-07 Méthode d'encodage, dispositif d'encodage, méthode de décodage, dispositif de décodage, programme et support d'enregistrement

Country Status (2)

Country Link
JP (1) JP5786044B2 (fr)
WO (1) WO2013118835A1 (fr)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005004113A1 (fr) * 2003-06-30 2005-01-13 Fujitsu Limited Dispositif de codage audio
JP2006010817A (ja) * 2004-06-23 2006-01-12 Victor Co Of Japan Ltd 音響信号符号化装置

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005004113A1 (fr) * 2003-06-30 2005-01-13 Fujitsu Limited Dispositif de codage audio
JP2006010817A (ja) * 2004-06-23 2006-01-12 Victor Co Of Japan Ltd 音響信号符号化装置

Also Published As

Publication number Publication date
JP5786044B2 (ja) 2015-09-30
JPWO2013118835A1 (ja) 2015-05-11

Similar Documents

Publication Publication Date Title
US20070106502A1 (en) Adaptive time/frequency-based audio encoding and decoding apparatuses and methods
US8665945B2 (en) Encoding method, decoding method, encoding device, decoding device, program, and recording medium
WO2012102149A1 (fr) Procédé d'encodage, dispositif d'encodage, procédé de détermination de quantité de caractéristique périodique, dispositif de détermination de quantité de caractéristique périodique, programme et support d'enregistrement
CN110444215B (zh) 编码装置、编码方法、记录介质
JP2011522291A (ja) オーバーラッピング変換の２つのブロック変換への因数分解
JP6457552B2 (ja) 符号化装置、復号装置、これらの方法及びプログラム
WO2011045926A1 (fr) Dispositif de codage, dispositif de décodage, et procédés correspondants
JP5663461B2 (ja) 符号化方法、符号化装置、プログラム、記録媒体
JP5786044B2 (ja) 符号化方法、符号化装置、復号方法、復号装置、プログラム及び記録媒体
JP5734519B2 (ja) 符号化方法、符号化装置、復号方法、復号装置、プログラム及び記録媒体
JP6224827B2 (ja) 分配量子化及び符号化を使用した累積和表現のモデル化によるオーディオ信号包絡符号化、処理及び復号化の装置と方法
JP5800920B2 (ja) 符号化方法、符号化装置、復号方法、復号装置、プログラム及び記録媒体
JP6224233B2 (ja) 分配量子化及び符号化を使用したオーディオ信号包絡の分割によるオーディオ信号包絡符号化、処理及び復号化の装置と方法
CN110291583B (zh) 用于音频编解码器中的长期预测的系统和方法
JP5320508B2 (ja) 符号化装置、復号装置、これらの方法、プログラム及び記録媒体
JP5840101B2 (ja) 符号化方法、符号化装置、復号方法、復号装置、プログラム及び記録媒体
JP2011009860A (ja) 符号化方法、復号方法、符号化器、復号器、プログラム
JP5738480B2 (ja) 符号化方法、符号化装置、復号方法、復号装置及びプログラム
JP5635213B2 (ja) 符号化方法、符号化装置、復号方法、復号装置、プログラム及び記録媒体
WO2013002238A1 (fr) Procédé d'encodage, dispositif, programme et support d'enregistrement
JP5714172B2 (ja) 符号化装置、この方法、プログラムおよび記録媒体
JPWO2018225412A1 (ja) 符号化装置、復号装置、平滑化装置、逆平滑化装置、それらの方法、およびプログラム
WO2013129439A1 (fr) Dispositif de codage, procédé de codage, programme et support d'enregistrement

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 13746300

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2013557574

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 13746300

Country of ref document: EP

Kind code of ref document: A1