MXPA98010783A

MXPA98010783A - Audio signal encoder, audio signal decoder, and method for encoding and decoding audio signal

Info

Publication number: MXPA98010783A
Application number: MXPA/A/1998/010783A
Authority: MX
Inventors: Norimatsu Takeshi; Tsushima Mineo; Ishikawa Tomokazu
Original assignee: Matsushita Electric Ind Co Ltd
Priority date: 1997-05-15
Filing date: 1998-12-15
Publication date: 1999-10-14

Abstract

An audio signal encoder and an audio signal decoder which can reproduce audio signals even when the decoder does not make full use of information from the encoder. A quantizing section constituting the encoder (1) is constituted to have a first quantizing subsection (501) composed of low-, intermediate-, and high-band quantizing subsections, a second quantizing subsection (502) which further quantizes the quantizing error of the first quantizing subsection (501), and a third quantizing subsection (503) which furthermore quantizes the quantizing errors after the processing in the first and the second quantizing subsections (501, 502).

Description

CODING DEVICE, AODIUM SIGNAL DECODER DEVICE AND AUDIO SIGNAL CODING AND DECODING METHOD Technical Field The present invention relates to an audio signal coding apparatus, for encoding an audio signal, such as a voice signal or a music signal and more particularly, to an apparatus for transforming a time domain audio signal. , frequency domain, using a method such as orthogonal transformation and efficiently encoding the signal transformed in such a way, that it is expressed with fewer code streams, compared to the original audio signal, and to a decoder apparatus that has a structure capable of encoding a high-quality and wideband audio signal, using all or only a portion of the code streams, which are encoded signals. Moreover, the invention relates to a method for encoding and decoding an audio signal.

REF .: 28977 Background in Art Many methods have been proposed to efficiently encode and decode audio signals. Especially for an audio signal that has a frequency band that exceeds 20 kHz., Such as a music signal, an MPEG audio method has been proposed in recent years. In the coding method represented by the MPEG method, a digital audio signal on the time axis is transformed to data on the frequency axis, using an orthogonal transformation, such as a cosine transformation and the data that is in the Frequency axis are encoded in an aurally important way, by using the characteristic of auditory sensibility of the human being, where the data that are not aurally important or are redundant data, are not encoded. In order to express an audio signal, with a quantity of data considerably less than the amount of data of the original audio signal, there is a coding method that uses a vector quantization method, such as the TC-VQ method. The audio methods MPEG and TC-WVQ, are described in the "I? -11172-3 standard of ISO / EIC" and in "T. Moriya, H. Suga, An 8 Kbits transform coder for noisy channels, Proc. ICASSP 89, pages 196-195", respectively. Hereinafter, the structure of a conventional audio coding apparatus will be explained, using Fig. 24. In Fig. 24, reference numeral 1601 denotes an FFT unit, which transforms an input signal into frequency; numeral 1602 denotes an adaptive bit allocation calculation unit, which performs an adaptive calculation of the bit allocation, by calculating an audible minimum limit and a masking characteristic, such that the specified band of the signal of input transformed into its frequency, be coded; numeral 1603 denotes a sub-band division unit, which divides the input signal, into multiple bands; numeral 1604 denotes a normalization unit of the scale factor, which normalizes each component of the band divided into multiple bands, using a scale factor; and numeral 1605 denotes a scalar quantization unit, which performs a scalar quantification of the normalized output from the normalization unit of the 1604 scale factor, according to the allocation of the bit, from the adaptive attribution calculation unit of 1602 bits. A description of the operation is given. An audio signal is input to the FFT unit 1601 and the division unit in sub-bands 1603. In the FFT unit 1601, the audio signal is subjected to a frequency conversion and the output is input to the adaptive allocation unit. of bits 1602. In the adaptive allocation unit of bits 1602, the amount of data that will be provided to a specific band component is calculated, based on a minimum limit audible, which is defined according to the auditory characteristic of human beings and the characteristic of masking, and to the attribution of quantity of data of each band, is codified in the form of an Index. On the other hand, in the division unit in sub-bands 1603, the input signal is divided, for example, into 32 bands, to be discharged. In the normalization unit of the scale factor 1604, a normalization of each of the components obtained from the division unit in sub-bands 1603 is carried out, with a representative value- The normalized value is quantified in the form of an index . In the scalar quantization unit 1605, based on a bit allocation calculated by the adaptive bit allocation unit 1602, the output from the normalization unit of the scaling factor 1604 is scaled quantized and the quantized value is encoded in the form of an IND2 index. In the conventional audio signal coding apparatus, constructed as described above, the MPEG method is generally used in such a way that the coding is effected with a data amount of 64000 bits / sec. for each channel. With a smaller amount of data than this, the reproducible frequency bandwidth and the subjective quality of the decoded audio signal are sometimes considerably degraded. The reason is the following. As in the example shown in Fig. 37, the coded data is divided into three main parts, i.e., the bit allocation, obtained by the adaptive allocation unit of bits 1602, the representative band value, obtained by the unit standardization of the 1604 scale factor and the quantized value, obtained by the scalar quantization unit 1605. Thus, when the compression ratio is high, not enough data is attributed to the quantized value. Furthermore, in the conventional audio signal coding apparatus, it is normal for an encoder and a decoder to be constructed with the amount of data to be encoded and the amount of data to be decoded, equal to one another. For example, in a method where a data amount of 128000 bits / sec is encoded. , a data amount of 128000 bits is decoded in the decoder. However, in conventional audio signal encoding and decoding devices, encoding and decoding must be carried out with a fixed amount of data, in order to obtain a satisfactory sound quality and consequently, it is impossible to obtain a high sound quality. with a high compression ratio.

The present invention is designed to solve the aforementioned problems and aims to provide coding and decoding apparatus for audio signals and a coding and decoding method. of an audio signal, with which a high quality and a broad reproduction frequency can be obtained, even when coding and decoding are carried out with a small amount of data and also, the amount of data in the sodification and decoding can be be variable and not fixed. Moreover, the object of the present invention is to provide audio signal encoding and decoding devices and an audio signal encoding and decoding method, which can significantly improve the efficiency of quantification.

Discovery of the Invention In order to solve the aforementioned problems, according to the invention of Claim 1, an audio signal encoding apparatus for encoding an audio signal is provided, by means of performing a vector quantization to a signal sequence characteristic of an audio signal. frequency, which is obtained by subjecting an input audio signal to a frequency transformation, and the apparatus comprises: a multi-stage quantization means, comprising at least one first-stage vector quantifier, to quantify vectorially the signal sequence characteristic of frequency or a portion of it, and a Vector quantizer of second stage to quantify vectorly a quantization error component, generated by the first stage vector quantifier; and each one of the quantification means of each stage, of the means of quantification in multiple stages, have at least one divided vectorial quantifier, which quantifies vectorially to a train of coefficients in any of the multiple frequency bands, obtained by means of dividing the frequency characteristic signal sequence, into at least two frequency bands that may have an overlapped portion between the multiple stages, using a division form for each stage. According to the invention of Claim 2, the audio signal encoding apparatus defined in Claim 1 further comprises standardizing means for normalizing the frequency characteristic signal sequence and adding its output to the quantization means. in multiple stages. According to the invention of Claim 3, in the audio signal Sodifying Apparatus, defined in Claims 1 and 2, the quantization means of each stage appropriately selects, as a frequency band divided from the signal sequence. frequency characteristic to be quantified, a band having a large sum of energy addition of the quantization error, and thus quantifies the selected band.

According to the invention of Claim 4, in the audio signal coding apparatus defined in Claims 1 and 2, the quantification means of each stage, selects a frequency band, in the manner of a frequency divided from the Signal sequencing characteristic of frequency that will be quantified, based on a facet of auditory sensibility that shows the auditory nature of human beings, and quantifies the selected band, whose frequency band selected has a large sum of power admission of a suantifisation error, measured by giving a large value to a band that has a high importance within the characteristic of auditory sensitivity. According to the invention of Claim 5, the audio signal coding apparatus, defined in any of claims 1 to 4, further comprises: a first quantization band selection unit, between the first stage vector quantizer and the vector quantifier of second stage, which are constituents of the means of quantification in multiple stages; and a second quantization band selection unit, between the second stage vector quantizer and the terrestrial stage vector quantizer; wherein the first quantization band selection unit receives, as an input, the output of the quantization error from the vector quantifier of first stage and select a band to be quantified, by the second stage vector quantifier; and the second quantization band selection means receive, as an input, the output of the quantifisation error from the second stage vector quantizer and select a quantized eer band, by the third stage vector quantifier. According to the invention of Claim 6, in the audio signal encoding apparatus defined in Claim 5 the multistage quantization means comprises: a plurality of divided vector quantifiers of stage i, which independently quantify the trains of respective coefficients in the respective frequency bands, in which the sequential frequency signal signaling is divided; and a vestigial quantifier of stage j, the sual serves as a quantifier of the entire band, to quantify, once at least, all the respective frequency bands of the input signal to be quantized. According to the invention of Claim 7, in the audio signal encoding apparatus defined in Claim 1, in the multistage quantization means, the vector quantizer of the previous stage, a quantifisation error in the vector sutifisation , using a vector quantification method with a codebook and the vector quantifier of the stage subsesuente, performs a vectorial quantification in addition to the calculated quantification error. According to the invention of Claim 8, in the audio signal encoding apparatus defined in Claim 2, in the multistage quantization means, the vector quantizer of the previous stage calculates a quantifisation error in the vector quantization , using a vestifial quantifisation method are a book of codes and the vector quantifier of the subsequent stage, performs an additional vector quantifisation to the calculated quantification error. According to the invention of Claim 9, in the audio signal encoding apparatus defined in Claim 8, when calculating distances between codes, used to remove an optimal code in the codebook during a vector quantization, the vector quantifier of the means of quantification in multiple stages, calculates the distances used, in the form of weight, by the normalized components of the output of the input signal from the normalization means, and extracts a code having the minimum distance. According to the invention of Claim 10, in the audio signal coding apparatus defined in Claim 9, the vestorial suantifier in the multistage quantization means calculates the distances using, in weight form, both normalized components of the output of the signal sequence characteristic of frequency, coming from the normalization means and a value based on the auditory characteristic that shows the auditory nature of the human being and extracts a code that has a minimum distance. According to the invention of Claim 11, in the audio signal encoding apparatus defined in any of Claims 2 and 8 to 10, the normalization means is provided with a frequency scheme normalization unit, which normalizes to broad strokes the scheme of the signal sequence frequency characteristic. According to the invention of claim 12, in the audio signal encoding apparatus defined in any of Claims 2 and 8 to 10, the normalization means is provided with a bandwidth normalization unit, which divides the the signal sequence characteristic of frequency, in a plurality of components of continuous unit frequency bands and normalizes the train of coefficients in each unit band, by means of dividing them with a single value. According to the invention of Claim 13, in the audio signal coding apparatus defined in Claim 1, the multi-step quantization means comprise: a vector sub-quantifier, which quantifies the respective coefficient trains of the resonant frequency bands in which the frequency characteristic signal sequence is divided, independently by the divided vector quantifiers; and a vector quantizer, which serves as a quantifier of the entire band to quantify, once at least, all the respective frequency bands of the input signal to be quantized. In accordance with the invention of Claim 14, in the audio signal coding apparatus defined in Claim 13, the means for multi-stage sutification comprise: a first vector quantifier, comprising a vector quantizer divided by a low band, a quantizer divided vector of intermediate band and a vector quantifier divided by high band; a second vector quantifier connected next to the first vector quantizer; and a third vectorial quantifier conestado next of the second vectorial quantifisador; wherein an input of the frequency characteristic signal sequence directed to the multistage quantization means is divided into three bands, and the frequency characteristic signal sequence of the low band component that is between the three bands is quantified by the looped vestifial quantifier of the low band, the signal frequency characteristic of the intermediate band component that is between the three bands, is quantified by the quantifier divided vector of intermediate band, and the frequency characteristic signal sequencing of the high band component that is between the three bands, is quantized by the high band divided vector quantizer, independently; each of the divided vector quantifiers constituting the first vector quantizer calculates a quantization error, with respect to the frequency characteristic signal sequence, and sends this error to the second subsequent vector quantizer; the second vector quantifier performs a quantification, so that a band amplitude is quantized by the second vector quantifier, calculates a quantization error, with respect to the input of the second vector quantifier and exits this error towards the third vestifial quantifier; and the third vectorial quantifier effects a quantification, so that a band amplitude is quantified by the third vectorial quantifier. According to the invention of Claim 15, the audio signal encoding apparatus defined in Claim 14 further comprises a first band quantization selection unit disposed between the first vector quantizer and the second vector quantizer, which are constituents of the quantification means in multiple stages, and a second band quantization selection unit, disposed between the second vector quantifier and the third vector quantifier; wherein the first band-suffixing selection unit receives, in the form of an input, the output from the first vector quantizer and selects a band to be quantized by the second vector quantizer; the second vestorial quantifier effects a suantifisation of a bandwidth, to be quantified by the second vector quantifier, with respect to the quantization errors from the first vector quantifier, which comprise the three vector quantifiers, in the band selected by the first band quantization selection unit and so, calculates an error of suantification with respect to the input of the second vector suantifisator and sends this error to the second unit of selection of band quantization; the second band quantization selection unit receives, in the input form, the quantization error from the second vector quantizer and selects a band, to be quantified by the third-vector quantizer; and the third vector quantifier performs a quantization of a bandwidth, to be quantified by the third vector quantifier, with respect to the quantization error from the second vector quantizer, in the band selected by the second band quantization selection unit.

According to the invention of Claim 16, in the audio signal encoding apparatus defined in Claim 14, instead of the first vector quantizer, the second vector quantizer or the third vestorial quantizer are constructed, using the vector split-band quantizer. low, to the divided vector quantifier of intermediate band, or to the vector quantifier divided of high band. According to the invention of Claim 17, there is provided an audio signal decoding device that receives, as an input, the output of codes from the audio signal encoding apparatus defined in Claim 1 and deodifying these codes to exit a signal corresponding to the original input audio signal and the decoding apparatus comprises: an inverse quantization unit, which performs an inverse quantization, using at least a portion of the code output from the quantifisation means of the apparatus audio signal encoder; and a frequency inverse transform unit, which transforms the output of a fresuensia character signal sequence from the inverse quantization unit, to a signal corresponding to the original audio input signal. According to the invention of Claim 18, an audio signal decoder apparatus is provided which receives, as an input, the output of codes from the audio signal encoding apparatus defined in Claim 2 and decoding these codes to output a signal corresponding to the original input audio signal and the decoding apparatus comprises: a unit of Inverse quantization, which reproduces a frequency characteristic signal sequence; an inverse normalization unit, which reproduces the normalized components, based on the code output from the audio signal sodifisher, using the frequency signal output characteristic of the inverse sutifisation unit, and multiplises the signal frequency and the normalized components; and a reverse frequency transformation unit for receiving the output from the inverse normalization unit and transforming the frequency signal signal sequencing into a signal corresponding to the original audio input signal. In accordance with the invention of Claim 19, there is provided an audio signal decoding device which, as an input, outputs the codes from the audio signal encoding apparatus defined in Claim 13, and decodes these codes. to output a signal corresponding to the original audio input signal, and the decoding apparatus comprises: an inverse quantization unit, which perform a reverse quantization using the output codes, suando-- all the codes are graduated from all the vector quantifiers that constitute the quantification means of the audio signal encoding device, or some of these. According to the invention of Claim 20, in the audio signal decoding apparatus defined in Claim 19, subsequent to the inverse quantifisation of the quantized codes in a preset band at a specified stage, the reverse quantization unit executes, alternately , an inverse quantifisation of the quantized codes in the preset band in a next step, and the inverse quantization of the quantized codes in a band different to the preset band in the specified stage; when quantized codes no longer exist in the pre-stable band in the next step, the inverse quantization unit continuously executes an inverse quantization of the quantized codes in a different band; and when quantized codes no longer exist in the band different from the pre-stable band, the inverse-quantization unit continuously executes an inverse quantifisation of the quantized codes in a next stage. According to the invention of Claim 21, an audio signal decoder apparatus is provided which receives, as an input, the code output from the audio signal encoding apparatus defined in Claim 14 and decodes these codes to output a signal corresponding to the original audio input signal, and the decoding apparatus comprises: a unit of inverse quantization, which performs an inverse quantization, using only the code output from the low-band divided vector quantifier, as a constituent of the first vector quantizer, even though all or some of the three divided vector quantifiers constitute the first vestifial quantifier in the output codes of the audio signal encoding device. According to the invention of Claim 22, in the audio signal decoding apparatus defined in Claim 21, the inverse quantization unit performs an inverse quantization, using the code output from the second vector quantifier, in addition to the output of codes from the low-band divided vector quantizer, as a constituent of the first vector quantizer. According to the invention of Claim 23, in the audio signal decoding apparatus defined in Claim 22, the inverse quantization unit performs an inverse quantization, using the code output from the divided vector quantizer of intermediate band, as a constituent of the first vector quantifier, in addition to the code output from the low band divided vestifial quantifier, as a sonic of the first vector quantifier and the code output from the second vector quantifier. According to the invention of Claim 24, in the audio signal decoder apparatus defined in Claim 23, the inverse quantization unit performs an inverse quantization, using the code output from the third vector quantifier, in addition to the output of codes from the low band divided vector quantizer, as a constituent of the first vector quantizer, the code output from the second vector quantizer and the code output from the intermediate band divided vector quantizer, as a constituent of the first vector quantizer. According to the invention of Claim 25, in the audio signal decoder apparatus defined in Claim 24, the inverse quantization unit performs an inverse quantization, using the code output from the high band divided vector quantizer, as a constituent. of the first vector quantifier, in addition to the code output from the low-band divided vector quantizer, as a constituent of the first vector quantifier, the code output from the second vector quantifier, the code output from the intermediate band divided quantifier, as a constituent of the first vestigial quantifier and the code output from the third vector quantifier. According to the invention of Claim 26, there is provided an audio signal encoding and decoding method, which receives a signal sequence frequency characteristic, obtained by means of frequency transformation of an input audio signal, coding and outputting this signal and decoding the encoded output signal, to reproduce a signal corresponding to the original audio input signal, wherein: the frequency characteristic signal sequence is divided into coefficient trains, corresponding to at least two bands of fresuensia, and these coefficient trains are independently quantified and graduated; and from the quantized received signal, the data of an arbitrary band corresponding to the divided band, are inversely quantized, consequently to reproduce a signal corresponding to the original audio input signal. According to the invention of Claim 27, in the audio signal coding and decoding method defined in Claim 26, the subletting is carried out in stages, in such a way that an error of quantified salculation is quantified additionally; and the inverse quantification is carried out by means of repeating, alternately, the quantification aimed at expanding the band and the quantification aimed at deepening the quantification stages, in the quantification. According to the invention of Claim 28, in the audio signal encoding and decoding method defined in Claim 27, the inverse quantization, directed to expand the band is carried out, by expanding the band to the sense with respect, to the auditory physiological characteristics of human beings. According to the invention of Claim 29, in the audio signal coding and decoding method defined in any of Claims 26 to 28, at the termination of the coding, after the frequency characteristic signal sequence is normalized , the signal sequence characteristic of frequency is divided into trains of coefficients, corresponding to at least two frequency bands, and the respective coefficient trains are independently quantized and graduated; and at the termination of the decoding, using the codes in relation to the normalization, coming from the termination of the coding, the codes coming from the termination of the encoding is subjected to an inverse normalization and thus, the inversely normalized codes are subjected to an inverse quantization, in such a way that the data that are in an arbitrary band corresponding to the divided band are inversely quantified and with this, reproduce a signal corresponding to the original audio input signal. According to the invention of claim 30, an audio signal sounding apparatus is provided to encode an audio signal, by performing a normalization and a vestorial quantization to a frequency-phasic signal sequence, the signal being obtained at subjecting an input audio signal to a freshness transformation, and the apparatus assumes: a multistage quantification means comprising, at least, a unit of vector quantization and first stage normalization, which normalizes and quantifies vectorially to the signal sequence characteristic of frequency 6 to a portion of this, and a second unit of vector quantization and of second stage normalization, which normalizes and quantifies vestorially to a somponent of error of suantification, generated by the first unit of vector quantification and first stage normalization; and means for quantifying each stage of the quantification means in multiple stages, having at least one unit of vector quantization and split normalization, which normalizes and quantifies vectorially a train of coefficients, of any of the multiple frequency bands, obtained by dividing the signal sequence characteristic of frequency in, at least, two frequency bands, which may have an overlapped portion between the multiple stages, using a type of division for each stage. According to the invention of Claim 31, the audio signal encoding apparatus defined in Claim 30 further comprises: a first band quantization selection unit, between the vector quantization unit and the first stage normalization unit; unit of vector quantification and second stage normalization, which are constituents of the means of quantification in multiple stages; and a second band quantization selection unit between the vector quantifisation unit and the second stage normalization unit and the vestifial quantification and standardization unit of the third stage; wherein the first band quantifisation selection unit receives, as an input, the output of the quantifisation error proceeding from the vector quantization unit and the first stage normalization, selects a band to be quantized by the vector quantization unit and second stage normalization, and exit a quantization error in the selected band, towards the unit of quantifisasidn vectorial and of normalization of stage second; and the second band-sucking selection means receive, as input, the quantization error output from the vector quantization unit and the second-stage normalization unit, select a band to be quantized by the vector quantization unit and standardization of third stage, and exit a quantification error in the selected band, towards the vector quantifisation unit and third stage normalization. According to the invention of Claim 32, in the audio signal coding apparatus defined in Claim 31, one of the vector quantization and normalization units, within and after the second stage of the quantifisation means in stages multiples, appropriately seles a frequency band having a large sum of energy addition of the output of the quantization error, coming from the unit of vector quantization and normalization of the previous stage, from the frequency bands divided from the sequence of signal frequency characteristic that has to be normalized and quantified, and then the unit normalizes and quantifies the selected band. According to the invention of Claim 33, in the audio signal encoding apparatus defined in Claim 31, each of the quantization units vector and normalization, within and after the second stage of the means of quantification in multiple stages, appropriately selessiona frequency band, from the frequency bands divided from the signal sequence characteristic of frequency to be normalized and quantified, based on the characteristic of auditory sensitivity, which shows the auditory nature of human beings, whose frequency band selected, has a large sum of energy addition of the quantization error output, coming from the vector quantification unit and of normalization of previous stage and measurement, when giving a large value to a band that has a high importance of sensitivity of auditory sensibility and thus, the unit normalizes and quantifies the selected band. According to the invention of Claim 34, there is provided an audio signal decoding apparatus that receives, as an input, the output of codes from audio signal coding devices, as defined in Claims 30. to 33, and decoding the codes to output a signal corresponding to the original audio input signal, and the decoding apparatus comprises: units of inverse quantization, to receive signals from their respecitive quantifiers, in the sutifisation unit of the encoder apparatus. audio signal, and playing the signals corresponding to the coefficient trains of the respective frequency bands, in which the sequencing of signal frequency characteristic is divided; a plurality of inverse normalization units provided correspondingly to the respective inverse quantization units, for multiplying the coefficient trains of the frequency-phasing-frequency signal sequence outputs, from the respec- tive inverse quantization units, by means of reproduced standardized somatics at the base of the codes related to the normalization and to the output from the audio signal coding apparatus, and outputting the signals corresponding to the respective coefficient trains, of the signal sequence characteristic of frequency, before being encoded; and a reverse frequency transformation unit, for receiving the outputs from the multiple inverse normalization units, and transforming these, in signals corresponding to the original audio input signal. According to the invention of Claim 35, there is provided a method of encoding and de-broadcasting audio signal to receive a signal sequence of freshness, obtained by means of a frequency transformation of an input audio signal, coding and outputting this signal. , and decoding the encoded output signal, to reproduce a signal corresponding to the original audio input signal, where: the characteristic signal sequence of frequency is divided into trains of soefficients, corresponding to at least two frequency bands, and these trains of soefisients are normalized, quantized and graduated independently; and from the received quantized signal, using the codes related to the normalization and the output of the finished coding, the data of an arbitrary band corresponding to the divided band, are normalized and inversely quantized, in such a way that a corresponding signal is reproduced to the original audio input signal. According to the invention of Claim 36, in the audio signal encoding and decoding method defined in Claim 35, normalization and quantization are carried out in stages, such that the salivated quantifisation errors are standardized and quantified in an ad hoc manner; and the normalization and the inverse quantification, are carried out by means of repeating, alternately, the normalization and the quantifisation directed to expand the band and the inverse normalization and quantification directed to deepen the quantification stages, in the quantification. According to the invention of Claim 37, in the method of sodification and signal decoding of The audio defined in Claim 36, the inverse normalization and quantification, aimed at expanding the band is carried out, by expanding the band to the sense with respect, to the auditory physiological characteristics of human beings. In accordance with the invention of Claim 38, there is provided an audio signal encoding apparatus for encoding an audio signal, by performing a vector quantization to a signal sequence characteristic of frequency, which is obtained by submitting a signal. audio input, to a frequency transformation, and the apparatus comprises: frequency transformation, and the apparatus comprises: a multi-stage quantization means, comprising at least one first-stage vector quantizer, to quantitatively quantify the sequence of frequency characteristic signal or a portion thereof, and a second stage vector quantifier to vestifially quantify a quantization error component, generated by the first stage vector quantifier; and each one of the means of quantification of each stage, of the means of quantification in multiple stage, comprises a vestigial quantifier of a full band, to quantify vectorially, all the sequencies of signal sarasteristicae of frequency of all the components of quantifisation error, which come out of the quantification means of the previous stage. As described above, according to the coding apparatus and the audio signal decoding device, or the audio signal deodification and deodification method of the present invention, a structure capable of quantifying is provided for quantification. , even at high data compression ratios, by using, for example, a vector quantization method and used to attribute a quantity of data during a quantification, to attribute, alternately, a quantity of data that contribute to the expansion of data. a reproduced band and a quantity of data that contribute to the improvement of the quality. First, in the encoding apparatus, and as a first stage, an input audio signal is transformed to a signal in the frequency domain and a portion of the frequency signal is encoded; in the second stage, a portion of the uncoded frequency signal and a quantization error signal are coded and added to the codes obtained in the first stage; in the third stage, the other portion of the frequency signal without coding, and the quantization error signals of the first and second stages, are coded and added to the codes obtained in the first and second stages; followed by a similar coding in the most advanced stages. On the other hand, in the decoding apparatus, any of the following decoding forms is possible: decoding using only the codes decoded in the first stage, decoding using the codes deodified in the first and second stages, and decoding using the decoded codes of the first stage to the third or more stage. The order of decoding, ee to decode, alternatively, the codes that contribute to the expansion and the codes that contribute to the improvement of quality. Because of this, a satisfactory sound quality is obtained, even when the coding and decoding are carried out without a fixed amount of data. In addition, a high quality of eonide is obtained at a high pressure rejection. Moreover, according to the coding apparatus and the audio signal decoding device, or the audio signal coding and decoding method of the present invention, normalization means are provided, rather than quantization means, and the normalization of an audio signal input, is carried out before the quantization. Accordingly, the normalization means and the quantization means carry out a codification, while exhibiting all of their capacities, so that a highly efficient quantization is carried out, without the loss of the amount of data, that the audio signal possessed. original and consequently with fewer errors of quantification and this effect being dramatically elevated, according to the type of audio signal. Moreover, as described above, the attribution of the quantity of data during quantifisation is carried out in such a way that an amount of data is attributed, alternately, that contributes to the expansion of a reproduced band and an amount of data that contributes to the improvement of quality. When a quantity of data is limited in a receiving terminal, an inverse quantization is performed only in a narrow band and in a surface region. However, by expanding the inverse quantization alternately in one direction, to expand the band and in a direction to increase the depth of the inverse quantization to increase the amount of data in a receiving terminal, a desired amount of data may be deodified. the encoded audio signal, despite the amount of data transmitted from the coding device. As a result, by varying the sanctity of data to be decoded, according to the communication environment or circumstance similar, in the receiving terminal, stable high definition sound quality can be obtained, even when a telephone network is used. ordinary public Moreover, in accordance with the encoding apparatus and the audio signal decoding device, or the audio signal coding and deodification method of the present invention, standardization means are provided. before each of the means of quantification that effect a quantification of multiple stages, and the normalization is carried out in each of the divided frequencies and for each stage of the quantification, followed by the quantification. Consequently, the normalization of each frequency domain allows an appropriate sodification according to the amount of data that the audio signal possessed in each frequency domain, that is, the means of quantization and the normalization means carry out a coding, while they exhibit their total capacities, therefore, a quantification is made with a high efficiency, without the loss of the amount of data that the original audio signal had and thus, with fewer errors of quantifisation, with the result of obtaining a quality of stable high definition sound, the effect being dramatically high, according to the type of audio signal. Additionally, at the end of the coding, when the inverse normalization and the inverse quantifission are carried out in both the direction of increasing the band, and in the direction of increasing the depth of the quantization, alternatively, an amount can be decoded. of the desired data of the encoded audio signal, despite the amount of data that is transmitted from the encoding apparatus. Accordingly, by varying the amount of data that will be decoded, according to the environment of somnolence or similar circumstances, in the terminal of Reception, you can obtain a stable high definition sound output, even when an ordinary public telephone network is used.

Brief Description of the Drawings Fig. 1 is a diagram illustrating the complete structure of an audio signal encoding and decoding apparatus, according to a first embodiment of the present invention. Fig. 2 is a block diagram illustrating an example of a normalization unit, as a constituent of the audio signal encoding apparatus described above. Fig. 3 is a block diagram illustrating an example of a frequency scheme normalization unit, as a constituent of the audio signal encoding apparatus described above. FIG. 4 is a block diagram illustrating a quantization unit, as a constituent of the above-described audio signal coding apparatus, according to a first embodiment of the present invention. FIG. 5 is a block diagram illustrating a quantization unit, as a constituent of the audio signal encoding apparatus described above, according to a second embodiment of the present invention.

FIG. 6 is a block diagram illustrating a quantization unit, as a constituent of the audio signal coding apparatus described above, according to a third embodiment of the present invention. Fig. 7 is a block diagram illustrating an inverse quantization unit, as a constituent of the audio signal decoding apparatus described above, according to a broad embodiment of the present invention. FIG. 8 is a block diagram illustrating an example of an inverse quantization unit, as a constituent of the audio signal decoding apparatus described above, according to a fourth embodiment of the present invention. FIG. 9 is a block diagram illustrating another example of a reverse quantization unit, as a constituent of the audio signal decoding apparatus described above, according to a fourth embodiment of the present invention. Fig. 10 is a block diagram illustrating yet another example of an inverse quantization unit, as a constituent of the audio signal decoding apparatus described above, according to a fourth embodiment of the present invention. FIG. 11 is a block diagram illustrating yet another example, of a reverse quantization unit, as a constituent of the signal decoding apparatus of FIG. audio described above, according to a fourth embodiment of the present invention. Fig. 12 is a block diagram illustrating one embodiment of a reverse normalization unit. Fig. 13 is a diagram for explaining the detailed operation of a quantization unit within the coding apparatus. FIG. 14 is a diagram for explaining the detailed operation of an inverse quantization unit within the decoding apparatus. FIG. 15 is a diagram for explaining the performance of its inverse quantization, performed by the inverse quantization unit, according to FIG. fourth mode. Fig. 16 is a block diagram illustrating one embodiment of a frequency scheme inverse-quantization unit. Fig. 17 is a diagram for explaining a quantization method employed in the audio signal encoding apparatus, according to the prior art and the present invention. Fig. 18 is a block diagram illustrating a quantization unit within an audio signal encoding apparatus, according to a fifth embodiment of the present invention.

Fig. 19 is a waveform diagram for explaining the complete operation of the audio signal coding apparatus, according to the fifth embodiment of the present invention. Fig. 20 is a diagram illustrating the schematic structures of the audio signal encoding apparatus, according to the first to the third modalities (20 (a)) and the fifth modality (20 (b) and (c) ) of the present invention.

Fig. 21 is a waveform diagram for explaining the complete operation of the audio signal coding apparatus, according to the fifth embodiment of the present invention. Fig. 22 is a diagram for explaining an example of the normalization and quantification method, performed by the normalization and quantization means from first stage to third stage, in the audio signal coding apparatus, according to the fifth embodiment of the invention. present invention.

Fig. 23 is a diagram for explaining the structure of an audio signal decoding apparatus, corresponding to the audio signal encoding apparatus, according to the fifth embodiment of the present invention. Fig. 24 is a diagram illustrating the acoustics of the conventional audio signal encoding apparatus.

Lesser Ways to Execute the Invention Hereinafter, the embodiments of the present invention will be described using the drawings.

Modality 1 Fig. 1 is a diagram illustrating the complete estrustura of a coding and deodifying devices of audio data, according to a first embodiment of the invention. In Fig. 1, the reference numeral denotes an encoding apparatus and 2 denotes a decoding apparatus. In the coding apparatus 1, the reference numeral 101 of note to a frame division unit, which divides an input signal into a preset number of frames; 102 denotes a window multiplication unit that multiplies the input signal in a window function on the time axis; 103 denotes an MDCT unit that performs a modified discrete cosine transformation; 104 denotes a normalization unit that receives both the output signal from the time axis, and the output of the MDCT coefficients from the MDCT unit 103 and normalizes the MDCT coefficients; and 105 denotes a quantification unit, which receives the normalized MDCT coefficients and quantifies them. In spite of the fact that the MDCT transformation is effected in a transformation of In this modality, a Fourier Transformation (DFT) can also be used. In the deodifying apparatus 2, the reference numeral 106 denotes a reverse quantization unit, which receives an output (index I D2) of signal from the encoding apparatus 1 and inveighly quantifies this signal; 107 denotes a reverse normalization unit, which inversely normalizes the output from the inverse quantization unit 106, using an INDI Index, from the normalization unit 104 of the sodifier apparatus 1; 108 denotes an inverse MDCT unit, which performs a modified discrete cosine transformation at the output of the reverse normalization unit 107; 109 denotes a window multiplication unit, which performs a window multiplication to the proceeding output of the reverse MDCT unit 108; and 110 denotes an overlapping unit of supers 110, which effects an overlap of supers at the output of window multiplication unit 109. A description is given of the operation of the audio signal encoding and decoding apparatus, constructed as It has been described above. It is assumed that the input signal to the coding apparatus 1 is a digital signal sesuensia that is temporarily sontinuous. For example, it is a digital signal obtained by means of quantizing a 16-bit audio signal at a sampling frequency of 45 kHz. This signal of The input is assumed in a division unit of suds 101, until a preset sampling number is alsanse and is withdrawn, when the accumulated sampling number reaches a defined frame length. Here, the frame length of the frame dividing unit 101 is, for example, any of the 108, 256, 512, 1024, 2048 and 4096 samples. In the frame division unit 101, it is also possible to graduate the signal, they are a variable frame length, according to the characteristics of the input signal. In addition, the partition division unit 101 is constructed to estimate an output for the specified length change. For example, in the case where the frame length is 4096 samples, when half of a change length as long as the frame length is determined, the frame division unit 101 leaves the last 4096 samples, each time The frame length reached 2048 samples. Of course, even when the frame length of the milling frequency varies, it is possible to have a structure in which the change length is determined at half the frame length. The output from the frame division unit 101 is output to the window multiplication unit 102 and the normalization unit 104. In the window multiplication unit 102, the output signal from the frame division unit 101 is outputted from the frame division unit 101. , is multiplied by a window function on the time axis and the result is graduated from the window multiplication unit 102. This way is shown, for example, in the formula (1). hx? = h? xi -1 = 1,2 --V (1) where xl is the output from the frame division unit 101, hi is the window function and -hxi is the output from the window multiplication unit 102. In addition, i is the time suffix. The window function hi shown in formula (1) is an example, and the window function is not restricted to that shown in formula (1). The selection of a window function depends on the characteristic of the input signal to the window multiplication unit 102, the frame length of the frame division unit 101, and the shape of the window functions in the frames that they are located temporarily before and after the painting that is being processed. For example, assuming that the frame length of the frame division unit 101 is N, as the characteristic of the input signal to the window multiplication unit 102, the average energy of the input signals in each JV is calculated. / 4 and when the average energy varies significantly, the calculation shown in formula (1) is axesuited is a length of sub-frame smaller than N. In addition, it is desirable to properly select the window function, according to the form of the window function of the previous frame and with the form of the window function of the subsequent frame , in such a way that the shape of the window function of the current frame is not distorted. Then, the output from the window multiplication unit 102 is input to the MDCT unit 103, where the modified discrete cosine transformation is executed and the MDCT coefficients are released. A general formula for the transformation of modified dietary cosine is represented by formula (2). Í2p (k + l / 2) (n ^ yk =? Hxn "o) • CH N ~ (2): -V / 4 + l / 2 (A = 0.1 ..., - V / 2- l) Assuming that the output of the DCT coefficient proceeding from the MDCT unit 103 is expressed by yk in the formula (2), the output from the MDCT unit 103 shows the frequency and eetae characteristics linearly correspond to the lowest frequency component, in so that the variable k of yk, tends to 0, while this corresponds to a higher frequency component, while the variable k, tends to N / 2-1 from 0. The unit Normalization 104 resides, both at the output xi of the time axis signal from the frame division unit 101, and at the output yk of the MDCT coefficients from the MDCT unit 103, and normalizes the MDCT coefficients using several parameters. To normalize the MDCT coefficients, variations in the values of the MDCT coefficient are suppressed, whose values are considerably different between the low band component and the high band component. For example, when the low band component is considerably larger than the high band component, a parameter having a larger value in the low band component and a small value in the high band component, and the MDCT coefficients are selected. are divided by this parameter, to suppress the variations of the coefficient MDCT. In the normalization unit 104, the INDI Indices that express the parameters used for normalization are encoded. The quantization unit 105 receives the MDCT coefficients normalized in the normalization unit 104 and quantifies the MDCT so-cients. At this time, the quantization unit 105 outputs a code index having a minimum difference, between the differences of the quantized values and the respective quantized outputs, corresponding to the multiple code indices, in a codebook. In this case, a difference between the quantified value, made by the unit of quantification 105 and the value corresponding to the output of the Code Index, from the quantification unit 105, it is a quantification error. On the other hand, in the decoding apparatus 2, the decoding is carried out using the INDI indices, coming from the normalization unit 104 of the sodifisator 1, and indices IND2, proceeding from the suitability unit 105- In the reverse quantifisation 106, the normalized MDCT coefficients are reproduced, using indices IND2 from the quantization unit 105. In the inverse quantization unit 106, the reproduction of the MDCT soeffects can be carried out using all or some of the indices. Of course, the output from the normalization unit 104 and the output from the reverse quantization unit 106 are not always identical to those before the quantization, since the quantization made by the quantization unit 105 is assisted with quantification errors. In the reverse normalization unit 107, the parameters used for normalization in the coding apparatus 1 are restored from the output of the INDI indices, from the normalization unit 104 of the coding apparatus 1, and the output of the unit of inverse quantization 106, is multiplied by these parameters, to restore to the MDCT coefficients. In the MDCT reverse unit 108, the output of the MDCT coefficients from the reverse normalization unit 107, are subjected to a reverse MDCT transformation, whereby the signal of the frequency domain is restored in time domain signal. The inverse MDCT transformation calculation is represented, for example, by means of formula (3). -, 2 ----- t, 1 (2p (k -r \ l 2) (n + n¡>) xx (n) =? 'CH Ñ (3) - >. "= -V / 4 + 1/2 where y? k are the coefficient coefficients MDCT restored in the reverse normalization unit 107 and xx (n) are the MDCT coefficients that are graduated from the reverse MDCT unit 108. The window multiplication unit 109 performs a window multiplication using the output xx (n) from the reverse MDCT unit 108. The window multiplication is carried out using the same window that is used for the window multiplexing unit 102 of the coding apparatus 1, and a process is performed which is shown, for example, in formula (4). -. (.) = xx (?) • h¡ (4) wherein z (i) is the output from the window multiplication unit 109. The frame overlap unit 110 reproduces the audio signal, using the output from the window multiplication unit 109. Since the output from of the window multiplication unit 109 is a temporarily overlapping signal, the frame overlap unit 110 provides an output signal from the decoding apparatus 2 using, for example, the formula (5). sahda (?) = - ", (/) + zm ¡(¡+ CHANGE) (5) where zm (l) is the output signal z (i) from the window multiplication unit 109, in the time frame m, zm-¡(í) is the output signal from the unit window multiplication 109, in the time frame (ml), CHANGE is the sample number corresponding to the change length of the coding apparatus, and output (i) is the output signal from the decoding apparatus 2, in the time m, of the frame overlap unit 110. An example of the normalization unit 104 will be described in detail, using Fig. 2. In Fig. 2, the reference numeral 201 denotes a scheme normalization unit. of frequency, which receives the outputs coming of the frame division unit 101 and the MDCT unit 103; and 202 denotes a bandwidth normalization unit, which resides the output from the frequency scheme normalization unit 201 and executes a normalization with reference to a band table 203. A given description of the operation. The frequency scheme normalization unit 201 calculates a frequency scheme, i.e., a gross form of the frequency, using the data of the time axis output, from the frame division unit 101 and divides the output of the MDCT coefficients, from the MDCT unit 103, by this. The parameters used to express the frequency scheme are encoded as INDI indices. The bandwidth normalization unit 202 receives the output signal from the frequency scheme normalization unit 201 and performs a normalization for all bandwidth that is shown in the band table 203. For example, assuming that the output of the MDCT soeficientes from the frequency scheme normalization unit 201, are det (i) (1 = 0"2047) and that the band table 203 is as shown in Table 1, an average value of the amplitude in each band, using formula (6), for example.

(Table 1) s? anJ =? dct (?) F bjbajo < < bjalto (6) promj = bjalto - bjbajo -r 1, where bjbajo and bjalto are the lowest band index i and the highest band index i, respectively, in which det (i) in band j is shown to belong to the band table 203. In addition, p is norm in the calculation of distance, which is desirable to be 2, and propij is the average of the amplitude in each band number j. The bandwidth normalization unit 202 quantifies the prom-, to obtain the gpropij and normalizes it, using for example, the formula (7). n_dct (?) = det (/) / qpromJ bjbajo < t < billet (7) To quantify the proj, a scalar quantization can be used d can be carried out a vector quantification, using the codebook. The bandwidth normalization unit 202 encodes the INDI indices of the parameters used, to express the qpromj-. Although the normalization unit 104 of the coding apparatus 1 is constructed using both the frequency scheme normalization unit 201, and the band amplitude normalization unit 202, as shown in FIG. shown in Fig. 2, this can be constructed, so that it uses any of the units, frequency scheme normalization unit 201 or bandwidth normalization unit 202. In addition, when there is no significant variation between the component of low band and the high band somponente of the output of the MDCT soeficientes, from the MDCT unit 103, the output from the MDCT unit 103 can be entered directly into the quantization unit 105, without the need to use the units 201 and 202. The freshness scheme normalization unit 201 shown in FIG. Fig. 2, will be described in detail, using Fig. 3. In Fig. 3, the reference numeral 301 denotes a unit of linear forecasting analysis, which resides the output from the frame division unit 101.; 302 denotes a schema quantization unit, which receives the output from the linear predictive analysis unit 301; and 303 denotes to an envelope normalization unit, the signal resides the forward output of the MDCT unit 103. A description is given of the operation of the frequency scheme normalization unit 201. The linear predictive analysis unit 301 receives the audio signal on the time axis, from the frame division unit 101 and performs a linear predictive coding. The linear predictive coefficients (coefficients LPC) can be obtained, as a result of the linear predictive coding, generally by calcu- lating an autocorrelation function of a signal multiplied in a window, such as a H mming window and solving a normal equation or something similar. When the linear predistive soephisients are stored, they are converted to linear spectral torque coefficients (LSP coefficients) or similar and are quantized in the quantization unit of scheme 302. As a quantification method, vector quantization or scalar quantization can be used. Then, the frequency transfer characteristic expressed by the quantized parameters, by means of the scheme quantizing unit 302, is calculated in the envelope characteristic normalization unit 303 and the output of the MDCT coefficients from the MDCT unit 103, is divided by the characteristic that will be normalized. To be more specific, adding the linear predictive soefficients equivalent to the parameters quantified by the scheme quantification unit 302, are qlpc (i), the frequency transfer characteristic measured by the enveloping normalization unit 303, is expressed for example , by the formula (8). qlpc (i) O = i = ORDER OR ORDER + \ = ¡< n (8) e-.v (;) = l Iffif) where ORDER is desired to be 10 ~ 40 and fft () means transformation -Fourier of high speed. Using the calculated transference characteristic of e-nv (i), the envelope normalization unit 303 performs a normalization using, for example, formula (9) as follows. fcd1 (?) = mcdtii) I env (?) (9) wherein mcdt (1) is the output signal from the MDCT unit 103 and fcdt (i) is the normalized output signal from the envelopefacesatural normalization unit 303. Next, the quantization unit 105 of the encoding apparatus 1, will be described in detail, using Fig. 4. In Fig. 4, the reference numeral 401 denotes a first subquantification unit, 402 denotes a second subquantification unit, which receives the output from the first subquantification unit , and 403 denotes a terrestrial sub-unit of evaluation, which receives the output from the second subquantification unit 402.

Then, the operation of the quantisation unit 105 will be described. An input signal to the first subquantification unit 401 is the output from the normalization unit 104 of the coding apparatus, ie the standardized MDCT soepents. However, in the stricture which does not have a normalization unit 104, it will be the output from the MDCT unit 103. In the first subquantification unit 401, the input MDCT coefficients are subjected to a vector quantization or a scalar quantifisation, and the indices that express the parameters used for the quantification are coded. Next, the quantization errors with respect to the input MDCT coefficients, are calcined due to the quantization, and these are output to the second sub-quantization unit 402. In the first subsurface detection unit 401, all the MDCT coefficients can be quantized or also only a portion of these can be quantified. Of course, if only a portion of these is quantized, the quantization errors of the bands that are not quantized by the first subquantifisation unit 401 will become MDCT input coefficients of the non-quantized bands. Next, the second subquantifisation unit 402 receives the quantization errors of the MDCT coefficients, obtained in the first subquantification unit 401, and quantifies them. In this suantification, dome in the first subquantification unit 401, vector quantization or scalar quantization can be used. The second subquantification unit 402 encodes the parameters used for quantification, in the form of indexes. Next, it calculates the quantification errors, due to the quantification, and outputs them to the third subquantification unit 403. This third subquantification unit 403, is identical in its estrustura, to the second sub-quantification unit. The numbers of the MDCT coefficients, that is, the bandwidths, which are to be quantified by the first subquantification unit 401, the second subquantifisation unit 402 and the terrestrial sub-quantization unit 403, are necessarily equal to each other , and the bands that are going to be quantified, are not necessarily the same. Considering the auditory characteristic of human beings, it is desired that both the second subquantification unit 402, and the third subquantification unit 403, be determined in such a way as to quantify the band of the MDCT coefficients, which shows the low frequency component. As described above, in the audio signal encoding apparatus, according to the first embodiment, when the quantization is carried out, the quantization unit is provided in stages, that is, multi-stage quantization means are constituted, and the bandwidth which will be quantified by means of the unit of quantification, varies between the adjacent stages and thus, the coefficients of an arbitrary band that is between the input MDCT coefficients, for example, the coefficients corresponding to the component of low frequency, which is auditorily -important for human beings, they are quantitatively preponderantly- Therefore, even when an audio signal is encoded at a low bitrate, that is, a high compression ratio, it is possible to perform a high definition audio reproduction, in a reseption terminal.

Mode 2 Next, an audio signal coding apparatus according to a second embodiment of the invention will be described using Fig. 5. In this second embodiment, only the structure of the quantization unit 105 of the encoding apparatus 1 is different from that of the first embodiment and therefore, only the structure of the quantization unit will be described hereinafter- In Fig. 5, the reference numeral 501 denotes a first subquantifisation unit, 502 denotes a second unit of subsuantifisation, and 503 denotes a third subquantification unit. This fifth modality is different in its structure to that of the first modality, in that the first 501 quantization unit, divides the input MDCT coefficients, into three bands (high band, intermediate band and low band) and quantifies the respective bands independently, and the quantization sections for the respective bands, which constitutes the first unit of subquantification 501, corresponding to the so-called "divided vector quantifiers". Generally, when the quantification is carried out using a vector quantization method, a vector is constituted by means of extracting certain elements from the input MDCT coefficients, whereby the vectorial quantification is effected. In the first sub-quantization unit 501, according to this seventh embodiment, when a vector is constituted by means of extracting certain elements of the input MDCT coefficients, the quantification of the low band is made using only the elements of the low band, the quantifisation of the intermediate band is carried out using only the elements of the intermediate band and the quantification of the high band is carried out using only the elements of the high band, whereby the respective bands are subjected to a vectorial quantization. However, in this second embodiment, a method for dividing the band to be quantized into three bands, ie, low band, intermediate band and high band, is described by way of example, the number of divided bands can be different to three. Next, with respect to the second sub-subsurface unit 502 and the third sub-quantization unit 503, as well as the first sub-quantization unit 501, the band to be quantized can be divided into several bands. As described above, in agreement they are the second embodiment of the invention, in the means of quantification in multiple stages, initially in the first stage, the MDCT input coefficients are divided into three bands and independently quantified, in such a way that the Once the important auditory band has been tested, it can be carried out in the first-order quantification. In addition, in the subsequent quantization units 502 and 503, the MDCT coefficients of the auditory-important band are additionally quantified in stages, with which the quantization error is further reduced and a higher-definition audio reproduction is performed in the reception terminal.

Mode 3 An audio signal encoding apparatus according to a third embodiment of the present invention will be described using Fig. 6. In this third embodiment, since only the structure of the quantization unit 105 of the encoder 1, is different from that of the first mode and therefore, only the structure of the quantization unit will be described hereinafter. In Fig. 6, the reference numeral 601 denotes a first subquantification unit, 602 denotes a first quantization band selection unit, 603 denotes a second subquantifisation unit, 604 denotes a second selection unit of band of quantification, 605 denotes a third subquantifisation unit. This third embodiment is different from the structure of the second embodiment, in that they are added to the first selection unit of the quantization band 602 and to the second selection unit of the quantization band 604. From here on, the operation will be described. The first quantization band unit 602 outputs a band, of which MDCT coefficients will be quantized by the first quantization band selection unit 602, using the quantization error output (-fdcterr (i)), of the first subquantification unit 601. For example, j that maximizes esu-m (j) given in the formula (10) is calculated and a band is quantified that fluctuates from j * SHIFTING TO J * SHIFTING + BANDWIDTH . esum (j) =?, ßdt "(t) 2 (10) where DISPLACEMENT is the constant and ANCHOBANDA is the total sample corresponding to the bandwidth to be quantified by the second subquantification unit 603. The first band sorting unit 602 encodes, for example, the j that gives the maximum value in the formula (10), in the form of index IND2. The second sub-quantization unit 603 receives the index IND2 and quantizes the selected band by the first quantization band selection unit 602. The second quantization band selection unit 604 is implemented by the same structure as the first selection unit of quantization band 602, except that its input is the quantization error output, from the second subquantification unit 603 and the band selected by the second quantisation band selection unit 604, is the input that goes to the third subquantification unit 605. Although in the first quantization band selection unit 602 and in the second quantization band selection unit 604, a band that will be quantized by the next quantization unit is selected, using the formula (10), a band that has to be quantified can be calculated, using a value obtained by multiplying a value used for normalization made by the normalization unit 104 by the value by virtue of the faces of the auditory sensitivity of the human race, relative to the fresuensias, as shown in the formula (11). esum (j) ---- «- • (- •)} ( eleven ) wherein env (i) is obtained by dividing the output from the MDCT unit 103 between the output from the normalization unit 104, and zxc (i) is table by virtue of the auditory characteristic of human beings relative to lae frequensiae , and an example of this is shown in (Figure 2). In the formula (11), zxc (i) can always be 1, in such a way that it is not considered- (Graph 2) 500 1000 1500 2000 2500 - > frequency Also, it is not necessary to provide multiple quantization band selection unit steps, that is, only the first quantization band selection unit 602 or the second quantization band selection unit 604 can be used. As described above, according to the third embodiment, when the quantization is performed in multiple stages, a unit of quantization band selection is placed between the adjacent stages of the quantifisation units, so that the band is a quantized variable and is appropriately selectable. In consecuense, the band to be quantified can be varied, according to the input signal and the degree of freedom in the quantification is increased, with which the efficiency of the quantification is significantly improved, by predominantly quantifying a portion that needs to be quantified. Hereinafter, a description of the detailed operation of the quantization method of the quantization unit 106 included in the coding apparatus 1 is given, according to any of the first to third embodiments, using Fig. 1 and Fig. 13. From the entry of the MDCT coefficients 1401 to each unit of subquantification, some of these are extracted to constitute the sub-vectoree of source of aeonide 1403. In the same way, assuming that the coefficient trains, the suals are obtained by dividing the MDCT coefficients input to the normalization unit 104, with the output of the MDCT coefficients 1401 from the normalization unit 104, are the normalized components 1402, some of these components are extracted according to the same rule for extracting the sound source sub-vectors from the MDCT coefficients 1401, whereby they constitute the weight sub-vectors 1404. The rule for extracting sound ub- Emission source vectors 1403 and weight sub-vectors 1404 from the MDCT coefficients 1401 and the normalized components 1402, respectively, are shown, for example, in the formula (14). (VTOTAL vecto? - ¡+ j CR ITOTAL subvecto? (j) -? + j < TOTAL (14) CR Í? OTAL + j = TOTAL CR where the j element of the sound source sub-vector i is the subvector (J), the MDCT soephisients are the vector (), the total number of elements of the MDCT coefficients 1401 is TOTAL and the number of elements of the sound source sub-vectors 1403 is CR, and VTOTAL is determined at a value equal to or greater than TOTAL and VTOTAL / CR, and an integral number must be given. For example, when TOTAL is 2048, CR = 19 and TOTAL = 2052, or CK = 23 and VTOTA - = 2070, or CR = 21 and VTOTAi = 2079. The weight sub-vectors 1404 can be extracted by means of the process of the formula (14). The vector quantifier 1405 selects, from the sddigo gateway of the codebook 1409, a code vector having a minimum distance between it and the sound source sub-vector 1403, after having been heavy by the weight sub-vectors 1404. Then, the vector quantizer 1405 outputs an index IND2 of the code vector having a minimum distance and a residual sub-vector 1404, which corresponds to the quantization error between the code vector have a minimum distance and the input of the sound source sub-vector 1403. An example of the effective calculation method will be described on the premise that the vector quantizer 1405 is composed of three constituents: distance calculation means 1406, a code decision means 1407 and a residue generating means 1408. The distance calculation means 1406 calculates the distance between the sound source sub-vector 1403 and the code vector k of the codebook 1409, using for example, the formula (15). dik =? «(Subvector, (j) - Ck (j)) s (15) where wj is the element of the sub-vectors of weight, ck (j) is the element of the -fc vestor of sddigo, R and S are norms for the calculation of distances and it is desired that the values of -R and S are 1, 1.5, 2. These R and S standards for distance calculation can have different values. Also, dik is the distance from the k code vector to the i sound source sub-vector. The means of Code decision 1407 selects a code vector having a minimum distance, from the distances calculated by the formula (15) d a similar formula and codifies the Indise of this (IND2). For example, when day is the minimum value, the index that must be encoded by the i sub-vector ee u. The waste generating means 1408 generates the residual sub-vectors 1410, using the code vectors selected by the code decision means 1407, according to the formula (16). ras, (y) = S "* i'í? cto7; (y) - < -?" (/) (16) wherein the element of the residual residual sub-vector 1410 is) and the element of the u code vector, selected by the decision means of scrip 1407, is cu (j '). The residual sub-vestores 1410 are retained in the form of MDCT coefficients to be quantified by the subsequent quantization units, by executing the inverse process of the formula (14) or a similar one. However, when a band is being quantified, it has no influence on the subsequent subquantifisation units, that is, when the subsequent subquantification units are not required to quantify, the waste generating means 1408, the sub-vectors residuals 1410 and generation of MDCT transformation 1411 are not necessary. Even though you number the code vectors, which had the sddigos book 1409 is not specified, when the memory capacity calculating in time and the related thing is considered, it is desired that the number be approximately 64. As another modality of the vector quantifier 1405, the following structure is available. That is, the distance calculation means 1406 calculate the distance, using formula (17). wherein K is the total number of code vectors used to remove codes from the codebook 1409. The code-rejection means 1407 selects k, which gives a minimum value of the distance dik, punctuated in formula (17), u encodes the index of this (IND2). Here, k is a value that is in the range of 0 to 2J ----- 1. The residue generating means 1408 generates residual sub-vectors 1410 using formula (18). resif) =. { subvector, (j) - Cu (j) O = k < K (18) res? (J) =. { subvector, (j) - C "(j) K = k < 2K Although the number of code vectors that codebook 1409 has is not restricted, when the memory capacity that calculates in time and what is related is considered, it is desired that the number be approximately 64. Also, although the weight sub-vectors 1404 are generated from the standard components 1402, it is possible to generate the weight eub-vectors, by multiplying the weight sub-vespers 1404 by a weight under, of the auditory characteristic of human beings.

According to the third embodiment of the invention described above, the first quantization band selection unit is arranged between the first subquantification unit and the second subquantifisation unit, and the second quantization band selection unit is disposed between the second subquantification unit and the third subquantification unit, thereby causing the bands to be quantized by the second subquantification unit and the third variable subquantification unit and appropriately selecsionable. Therefore, the band to be quantified can be varied, according to the input signal and thus, the degree of freedom is improved in the suantification, whereby the efficiency of the quantification is significantly improved, by predominantly quantifying a portion that needs to be quantified-Mae still, according to the first to third modalities described above, in the audio signal coding apparatus for coding An audio signal, when performing a vector quantization to a sequencing of the frequency characteristic signal, obtained by means of the frequency transformation of an input audio signal, is provided with a multistage quantization means having at least one , a first-stage vector quantifier for vectorially quantizing the signal sequence of the frequency faces or a portion thereof, and a second-stage vector quantizer for vectorially quantizing a quantization error component, derived from the first-stage vector quantifier; and each stage of the means of quantification in multiple stages, is provided with at least one divided vector quantifier, which quantifiates vectorially the trains of coefficients of any of the multiple frequency bands, which are obtained by dividing the Sequential sequence of frequency faces, in at least two frequency bands, which may have an overlapped portion between the multiple stages, according to a form of division for each stage. In this structure, among the coefficients Input MDCT, the coefficients of an arbitrary band, for example, coefficients that correspond to a low-frequency band component, which is aurally important to humans, are quantified predominantly at a desired depth. On the other hand, at the end of the coding, the coding can be carried out using the codes decoded in the multiple stages, and the decoding order can be decoding alternately the codes that contribute to the expansion of the band and the codes that contribute to the improvement of quality. Therefore, even when the audio signal is encoded at a low bitrate, that is, at a high compression ratio, or even when coding and decoding are carried out, without a fixed amount of data, it is possible to perform a high-definition audio reproduction in the reception terminal.

Modality 4 Next, a conventional audio decoder apparatus, with a fourth embodiment of the present invention, will be described using Fig. 1 and Figs. 7 ~ 11. The output of the indices from the coding apparatus 1 is divided widely into the output of the INDI indices from the normalization unit 104 and the output of the IND2 indices, from the quantization unit. 105. The output of the INDI indices from the normalization unit 104 is decoded by the reverse normalization unit 107, and the output of the IND2 indices from the quantization unit 105 is decoded by the inverse quantization unit 106. The The inverse quantization unit 106 can perform a decoding, using only a portion of the output of the indices IND2, from the quantization unit 105. Thus, assuming that the quantization unit 105 of the coding apparatus 1 has the structure shown in Fig. 5, is a description of the case where the inverse quantization is carried out, using the inverse quantisation unit having the stricture of Fig. 7, in the apparatus decoder 2. In Fig. 7, the reference numeral 701 designates a first inverse quantifisation unit of low band somponent. The inverse-sub-quantization unit of low-band component 701 performs a decoding, using only the indices IND21 of the low-band component of the first sub-quantization unit 501. By means of the structure described above, that is, by performing the decoding using only the Indices of the low band component from the first subquantification unit 501, in the reverse quantization unit of low band somponent 701, can be decoding a desired amount of data of the encoded audio signal, despite the amount of data that is transmitted from the coding apparatus 1. That is, even when the amount of data of the data that is vain is decoded at the receiving location , be restricted, a quantity of data can be decoded by making the amount of data to be encoded different from the amount of data to be decoded. Accordingly, the amount of data to be decoded, may vary according to the communication environment or similar circumstances, in the receiving terminal, a stable high definition sound quality can be obtained, even when using a public telephone network ordinary Fig. 8 is a diagram showing the structure of the inverse quantization unit, included in the audio signal decoding apparatus, which is employed when the inverse quantization is carried out in two stages. In Fig. 8, the reference numeral 704 denotes a second inverse quantization unit. This second inverse quantization unit 704 performs a decoding using the indices IND3, from the second sub-quantization unit 502. As a result, the output IND21 'from the inverse quantization unit of low-band component 701 and the output IND3' from the second unit of inverse quantification 704, are added and its summation is graduated from the unit of quantification Inverse 106. This addition is made in the same band as the band that was quantified for each subquantification unit, in the quantification. As discussed above, the IND21 indices from the first subquantification unit (low band) are decoded by the inverse quantization unit of low band somponent 701 and, when the IND3 indices, coming from the second unit of subquantification are inversely quantized, the output IND21 'from the inverse quantization unit of low band component 701, added to the object of the inverse quantization, whereby the inverse quantization is carried out in two stages. Therefore, audio signal quantified in multiple stages can be exactly deodified, resulting in a higher sound quality. Next, FIG. 9 is a diagram illustrating the structure of the inverse quantization unit included in the audio signal decoding apparatus, in which the band object of the process is extended when the inverse quantization is carried out in two stages. In Fig. 9, reference numeral 702 denotes a first inverse quantization unit of intermediate band component. This first inverse quantization unit of intermediate band component 702 performs a decoding, using indices IND22 of the intermediate band somponent, from the first subquantization unit 501.

Accordingly, the output IND21 'from the inverse quantization unit of low-band component 701, the output IND3' from the second inverse quantization unit 704, and the output IND22 'from the first inverse quantization unit of component of intermediate band 702 are added and their sum is output from the inverse quantifisation unit 106. This addition is made in the same band as the band that was quantified by each subquantification unit, in the quantification. Therefore, the reproduced sound band is extended and an audio signal of a higher quality is reproduced. Next, Fig. 10 is a diagram showing the structure of the inverse quantization unit, included in the audio signal decoding apparatus, in which the inverse quantization is carried out in three steps, by means of quantification. having the structure of Fig. 9. In Fig. 10, the reference numeral 705 denotes a third unit of quantification inverea 705. The third unit of inverse quantization 705 performs a decoding, using the indices from the third unit sub-quantization 503. Accordingly, the output I-FD21 'from the inverse quantization unit of low-band component 701, the output IND3' from the second inverse quantization unit 704, the output IND22 'from the first unit of inverse quantization of intermediate band component 702, and output IND4 'from the third unit of quantification inverea 705, are added and their sum is output from the inverse quantization unit 106. This addition is made to the same band, which the band that was quantified by each subquantification unit, in the quantification. Next, Fig. 11 is a diagram illustrating the structure of the inverse quantization unit included in the audio signal decoding apparatus, in which the band object of the process is extended when the quantization in three stages is carried out , in the inverse quantization unit having the structure of Fig. 10. In Fig. 11, the reference numeral 703 denotes a first inverse quantization unit of high band component. This first inverse quantization unit of high band component 703 performs a decoding, using the indices of the high band component, from the first subquantification unit 501. Accordingly, the output IND21 'from the quantization unit inverea low band 701, the output IND3 'from the second inverse quantization unit 704, the output IND22' from the first inverse quantization unit of intermediate band component 702, the output IND4 'from the third inverse quantization unit 705 , and the output IND23 'coming from the first inverse quantization unit of high band component 703, are added and its sum is output from the inverse quantization unit 106. This addition is made to the same band, as the band that was quantized by each subquantification unit, in the quantifisation.

While this fourth embodiment is described for the case where the inverse quantization unit 106 inversely decodes the data quantized by the quantization unit 105, having the structure of Fig. 5, a similar inverse quantization can be performed, yet when the quantization unit 105 has the structure shown in Fig. 4 or 6. Moreover, when coding is carried out using quantization unit having the structure shown in Fig. 5, and the decoding is carried out using the inverse quantization unit having the structure shown in Fig. 11, as shown in Fig. 15, after the low band indices from the first subquantifisation unit are inversely quantized, the indices from the second sub-quantization unit 502 in the next step, are inversely quantized, and after the intermediate band indices from the first unit of subquantifisation are inversely quantified. In this way, the inverse quantification to extend the band or the inverse quantization to reduce the errors of quantification, are repeated alternately. However, when a signal encoded by the quantization unit having the structure shown in Fig. 4 is decoded using the inverse quantization unit having the structure shown in Fig. 11, since there are no bands divided in the structure of Fig. 4, the suantifixed coefficients are successively deodified by the inverse quantization unit in the next step. A detailed description of the operation of the inverse quantisation unit 106 is given, as is a constituent of the decoding apparatus 2, using Fig. 1 and Fig. 14. For example, the inverse quantization unit 106 is composed of the unit of Inverse quantization of low band component 701, when this has the inverse quantization unit shown in Fig. 7, and is composed of two inverse quantization units, i.e., the inverse quantization unit of low band component 701 and the second inverse quantization unit 704, when this has the inverse quantization unit shown in Fig. 8. The inverse vector quantizer 1501 reproduces the MDCT coefficients, using the IND2 indices from the vector quantification unit 105 of the apparatus encoder 1. When the subquantification unit has the structure shown in Fig. 7, the inverse quantization is carried out as follows, an index number is decoded from indexes I-TD21 and a code vector having the number is selected from codebook 1502. It is assumed that the content of codebook 1502, is identical to the codebook of the coding apparatus 1. From the selected code vector, a reproduced sub-vector 1503 is obtained and this is converted into an MDCT coefficient inversely quantized (i, j) 1504, in the reverse process of the formula (14). When the subsurfacing unit has the structure shown in Fig. 8, inverse quantization is carried out as follows. An index number -fc is deodi fi ed from indices IND21 and IND3, and a code dress having a number u, populated in the formula (19), is selected from the code book 1502. a reproduced sub-vector is generated using the formula (20).

C, (J) u = k resitj) - (20) - QO) u? k where the j element of the sub-vector reproduced is resi (j). Next, a description of the detailed structure of the inverse normalization unit 107 is given, as a constituent of the decoding apparatus 2, using Fig. 1 and Fig. 12. In Fig. 12, the reference numeral 1201 denotes to a unit of inverse quantization of frequency scheme, 1202 denotes a unit of inverse normalization of bandwidth, 1203 denotes a table of bands. The inverse quantization unit of frequency scheme 1201 receives IND11 indices from the inverse quantization unit of frequency scheme 1201, reproduces the frequency scheme and multiplies the forward output of the inverse quantization unit 106, by the frequency scheme . The band-width inverse normalization unit 1202 receives the indices IND12 from the band-amplitude normalization unit 202, and restores the amplitude of each band shown in the band table 1203, by means of multiplication. Assuming that the value of each restored band, using the IND12 Indices from the bandwidth normalization unit 202, is qpromj, the operation of the band-width inverse normalization unit 1202 is given by the formula (12) . dct (i) = n _dct (i) qprom¡ bjbaja = i = bjalta (12) wherein the output from the frequency scaling reverse quantization unit 1201 is ¡-d-ct (i) and the output from the inverse band normalization unit 1202 is dct (i). In addition, the band table 1203 and the band table 203 are identical. Then, a description of the detailed structure of the invequantization unit of frequency scheme 1201 is given as a constituent of the decoding apparatus 2, using Fig. 16. In Fig. 16, reference numeral 1301 designates a unit of invequantization of scheme, and 1302 denotes a unit of invequantization of envelope caracterletisa. The scheme invesharpening unit 1301 restores the parameters that show the frequency scheme, for example, linear prediction coefficients, using the index IND13 proceeding from the linear predictive analysis unit 301 of the coding apparatus 1. When the restored coefficients are coefficients For linear prediction, the quantum envelope faces E13 are re-corrected when calculating in formula (8), similarly. When the restored coefficients are not linear prediction coefficients, for example, when they are LSP coefficients, the envelope sarasteristisas E13 are restored, by transforming them into sarasterlsticas of frequency. The inverse quantization unit of envelope characteristic 1302 multiplies the envelope features E13 by output IND16, from the inverse quantization unit 106, as shown in formula (13), and outputs the result to the inverse normalization unit. of bandwidth 1201. mcdt (i) = fcdt (i) - env (i) (13) according to the fourth embodiment described above, standardization means are provided, before the quantization means, and the normalization of an input audio signal is carried out before surtification. Therefore, the normalization means and the means of quantification, perform a coding, at the same time to exhibit all of their layers, resulting in a high efficiency quantification, with fewer errors of quantification, without the loss of the amount of data who owned the original audio eefial. Likewise, when the amount of data in the reception terminal is limited, the inverse quantization is carried out only in a narrow band and in a surface region. However, by extending the inverse quantification alternately in the direction of expanding the band and in the direction of increasing the depth of internal quantification, to increase the amount of data in the receiving terminal, a desired amount of data of the encoded audio signal may be encoded, despite the amount of data that is transmitted from the encoding apparatus 1 - In agreement, by varying the amount of data to be encoded , in accordance with the communication environment of similar cirsunstansias, in the reseption terminal, a stable high definition sound output can be obtained, even if an ordinary public telephone network is used.

Modality 5 Next, an audio signal encoding apparatus according to a fifth embodiment of the present invention will be described using Fig. 18. In this fifth embodiment, since only the structure of the quantifisation unit 105 of the sodifisator 1, it is different from the modalities mentioned above, only the structure of the unit of quantification will be described and the descriptions of other constituents will be omitted. In Fig. 18, reference numeral 1801 denotes a first normalization unit, 1802 denotes a first subquantification unit, 1803 denotes a first band sorting unit, 1804 denotes a second normalization unit, 1805 denotes a second subquantification unit, 1806 denotes a second quantization band selection unit, 1807 denotes a third normalization unit, 1808 denotes a third subquantification unit, and 1809 denotes a third quantization band selection unit. This fifth modality is different, in structure, to the third modality, in which the second and third standardization units 1804 and 1807 are added. Next, the constituents of the fifth mode will be described. The first, second and third normalization units 1801, 1804 and 1807 are implemented by the same structure as the normalization unit 104, of asorde with the first mode. The first, second and third subquantification units 1802, 1805 and 1808 are implemented by the same structure as the first sub-quantization unit 601, according to the third embodiment. The first, second and third quantization band selection units 1803, 1806 and 1809, are implemented by the same structure as the first quantization band selection unit 602, according to the third embodiment. While in this fifth embodiment, the encoding apparatus has three sets of combinations, each comprising a normalization unit, a sub-quantization unit and a quantization band selection unit, the number of games is not restricted to three, it can be four or more, or it can be two .. in addition, the unit of selection of Band quantification of the game of the last stage, is not always necessary, that is, it can be dispensed with.

A description of the operation of the audio signal encoding apparatus is given in accordance with the fifth embodiment shown in FIG. 18. In FIG. 18, the MDCT coefficients of an audio signal input, in FIG. In the fifth mode, they are subjected to a normalization in the first normalization unit 1801 t and the normalized MDCT coefficients are graduated. The first sub-quantization unit 1802 quantifies the normalized MDCT coefficients, corresponding to the output signal from the first normalization unit 1801. The first subquantization unit 1802 outputs, in the form of indices, the parameters used for the quantization and later , this exits a quantization error generated in the quantification, to the first quantifisation band selection unit 1803, in the next stage. The first quantization band selection unit 1803 calculates a band, from which the MDCT coefficients will be quantized by the second subquantization unit 1805, using the output from the first subquantization unit 1802. The second standardization unit 1804 normalizes to the output of the MDCT coefficients from the first sub-quantization unit 1802, based on the result selected by the first band selection unit of 1803 quantification, with a review of the selected band. The second sub-quantization unit 1805 suffixes the output from the second normalization unit 1804 and outputs, in the form of indices, the parameters used for the quantization, and a quantization error is generated, in the quantization. The second quantization band selection unit 1806 calculates a band, from which the MDCT coefficients will be quantized by the third subquantization unit 1808, using the output from the second subquantifisation unit 1805. The third standardization unit 1807 normalizes to the output of the MDCT coefficients from the second subquantifisation unit 1805, based on the result selected by the second quantization band selection unit 1806, with respect to the selected band. The third subquantifisation unit 1808 quantifies the output from the third standardization unit 1807, and outputs, in the form of indices, the parameters used for the quantization and a quantization error generated in the quantifisation. The third quantization band selection unit 1809 shown in the figure is required only when there is a fourth subquantification unit (not shown) in the subsequent stage. If the fourth subquantifisation unit is present, the third unit of quantization band 1809 calculates a band, from which the MDCT coefficients will be quantified by the fourth subquantification unit, using the output from the third subquantization unit 1808. The first, second and third standardization units 1801, 1804 and 1807 are output, in the form of indices, the parameters used for normalization, as in the quantization unit 105 according to the first modality. Hereinafter, the operation and the functional characteristic of the coding apparatus according to the fifth embodiment, will be described in comparison, with the coding apparatuses according to the first to the third modalities. In the structure having normalization means in an audio signal encoding apparatus, according to any one of the first to third embodiments, an audio signal waveform is shown in FIG. 19 (a). the time axis is transformed to a waveform in the axis of freshness, by means of a MDCT transformation, FFT and the normalization means carry out the normalization A, that is to say the extraction and deviation of the scheme of the scheme extracted by an amplitude value , for the full frequency domain of the waveform that is on the frequency axis, as shown in Fig. 20 (a). Subsequently, the normalized output is subjected to a quantization x, y, z for one of the three domains of frequency, it is desir, low band, intermediate band and high band, which are obtained by dividing the dominance of somnsense frequency, respectively, they are what is obtained a suantifidated output, A (x + y + z). On the other hand, in this fifth embodiment, as shown in Fig. 20 (b), the normalization means a, ß, and? they are placed before the divided quantization means, respectively, and initially, as shown in Fig. 19 (d), a waveform is divided into the frequency axis, in a plurality of frequency bands. Next, normalization and quantifisation are carried out for each of the fresuensia bands, and as a result, a suantified output, ax + ßy +? Z, is obtained. The entire process is shown in Fig. 21. Generally, when there is a considerable polarization of the frequency characteristics of an audio signal, for example, when it is a signal in which the frequency is shown in the low band in the voice information, if this signal is coarse and completely normalized, a portion of the low-band characteristic, can not be normalized or quantified preponderantly. That is, coarse and complete normalization does not take an envelope of a portion of the signal where the signal change is small, so that the data of the portion where the signal change is small, are lost. By Subsequently, when the quantization is performed after normalization, even when the quantifier exhibits its total capacity, it merely effects a quantification of the signal, which does not have the data in the portion, where the signal change is small, that is, it makes a useless quantification. In other words, the quantifier performs a quantification in which, the effect made by the normalization and the quantifisasión sombinadas, is difísilmente obtained. As soon as the means of normalization are provided, it is necessary that both the normalization means, as well as the means of their stabilization, exhibit the totality of dreams. In sontraste are this, suando is entered a very coarse signal, even when all this signal is grossly normalized, the result will not differ much. Fig. 20 (a) shows the relationship that exists between the normalization means and each of the quantization means, in the audio signal encoding apparatus having the normalization unit 104 shown in Fig. 1, of agreement with any of the first to third modalities. In said structure in which the normalization means A, normalizes the entire sarasteristic frequency signal sequence of an input audio signal, when the input audio signal is a signal having a polarization with respect to the Often, for example, a signal whose frequency is concentrated in the low band will not be able to display all of its capacidade nor the means of standardization, as well as the means of quantification. In contrast to this, in the estrustura of I agree with the fifth modality, where the quantifism means x, y, yz, are provided with normalization means a, ß, and Y, since the normalization is carried out for each of the signals object of the quantification, made by each of the means of quantification, each of the normalization means carries out the appropriate normalization with respect to the load of each one of the means of quantifisation that effect the quantification, is say, the normalization in which the level of signal that has to be quantified, is brought to the level at which, each of the means of quantification can exhibit its total capacity, thereby providing the maximum effect produced by the means of quantification and standardization means. To be more specific, the structures of the normalization unit and the quantization unit of the coding apparatus, according to the fifth embodiment, are as shown in Fig. 20 (b). Initially, in a train of coefficients in a frequency band A, which is obtained by dividing an expensive frequency signal sequence, obtained by a frequency transformation of an input audio signal, or by dividing the frequency band of the characteristic signal sequence Frequency, normalization and quantification are carried out by means of the normalization means to the first stage and by means of quantification x. Then, by means of the second stage normalization means ß and by the quantization means and, a train of coefficients of a frequency band B, adjacent to the train of coefficients of the first stage frequency band A, is subjected to standardization and quantification. Also, through the means of standardization? of the third stage and by means of quantifisation z, a train of coefficients of a frequency band C, adjacent to the train of coefficients of the frequency band B, is subjected to normalization and to quantification. Alternatively, as shown in FIG. 20 (c), the second stage normalization means ß and the quantification means and, perform a normalization and quantization for the frequency band B, which is adjacent to, and partially overlapped, with the frequency band A of the first stage, in such a way that the output of the quantization error, coming from the first stage, is normalized and quantified, in the overlapped portion, while the train of coefficients of the frequency band B of the signal sequence characteristic of frequency, is normalized and quantified in the other portion; and the means of standardization? of the third stage and the means of quantization z, effect a normalization and a quantification for the band of fresuensia C, which is adjacent to, and partially overlapped, with the frequency band B of the second stage, in such a way that the output of the quantization error, of the second stage, it is normalized and quantified, in the overlapped portion, while the train of soefficients of the frequency band C of the signal sequence characteristic of frequency, is normalized and suffused in the other porsión. In the structure shown in Fig. 20 (b) or 20 (c), since the normalization is carried out before each quantification, made by each of the means of quantification, each of the normalization means effects a normalization with respect to the load of one of the means of quantification, whereby each of the means of normalization and each of the means of quantification, can perform a quantification, at the same time of exhibiting the totality of dream layers, resulting in a significantly improved quantification efisiensia. The freshness bands processed by the normalization and suffusion means in the resis- tant stages and the depths of their anti-profiling are not restricted to those shown in Figs. 20 (b) and 20 (c). These bands and depths can be adjusted arbitrarily.

Next, a description is given of an example of a normalization and quantification method, made by the normalization and quantization means of first to third stages, using Figs. 18 and 22. In this fifth embodiment, although the first, second and third standardization units 1801, 1804 and 1807 are implemented with the same structure of the normalization unit 104 according to the first embodiment, these units can be implemented, using a parameter normalization calculation method, different from the one mentioned above. For example, the normalization units can be constructed using standardization parameters, LPC coefficients or LSP coefficients, directly calculated from the input of the MDCT coefficients, to the respective normalization units. In Fig. 22, TI, T2 and T3 are normalization tables, used by the normalization means a, ß and? respective, and these tables are obtained as follows. In the MDCT coefficients of several types of sound source signals, which can be entered as an input audio signal, an LPC (Linear Predistive Coding) analysis is performed, to obtain LSP sobjects (Line Spectral Pair). This operation is repeated for each sound source and the LSP coefficients are obtained for each frame. So, all the LSP coefficients are resoled and subjected to a group analysis, to obtain a plurality of representative envelope patterns and envelope patterns, are used as in the normalization table T2, for the second stage ß normalization means. Next, the normalization table T3, for the normalization means? of the third stage, it is obtained in the same way as described above. In this way, the normalization tables TI, T2 and T3 are obtained for the normalization means, which allow the respective quantification means, from the first to the third stages, to produce an optimal quantization, in which the normalization and the quantification exhibits the totality of its capacities, with respect to the load of each one of the means of suantificación. In the audio signal encoding apparatus constructed in such a way, when an audio signal, the sual considered by these various types of sound source signals, is input to this apparatus, the input audio signal is transformed from data in the time axis, to data in the frequency axis, by means of a MDCT transformation, FFT, as shown in Figs. 19 (a) through 19 (b). The data transformed to data on the frequency axis is subjected to a scheme extraction, as shown in Figs. 19 (b) to 19 (c). This extraction of The scheme is effected using the normalization table TI of the normalization means a of the first stage, and the scheme obtained as a result made by approximately, for example, 20 pieces of LSP polynomials. Then, the signal transformed to data on the frequency axis is divided by means of the obtained scheme, whereby the normalization a is carried out. After normalization, the normalized signal is quantified by means of the quantization means x, which completes the normalization and quantification of the first stage. As a result, efficient quantization is effected with respect to the characteristics of various types of sound source signals, described above. Then, through the second stage quantization and normalization means, the second stage quantization and normalization is performed on a train of coefficients of a frequency band B, different to that of the coefficient trains of a band of fresuensia A, the signal is divided from a frequency signal characteristic and is subjected to a first stage normalization and quantization, or to a quantization error output, as a result of the normalization and quantification of the first stage, using the second stage normalization table T2. Therefore, an effi- cient quantifisation is made, with respect to the characteristics of several types of signals from sources of sound, giving priority to the desired frequency band, or to a depth of quantization of a desired portion in a frequency band. Through the means of quantifisation and third stage normalization, the quantization and normalization of the third stage is carried out on a train of coefficients of a frequency band, different from that of the trains of coefficients of the frequency band. for the signal sequence frequency characteristic and subjected to a first and second stage normalization and quantization, or to an error output of quantifisasidn, as a result of the normalization and second stage quantization, using the third stage normalization table T3. Accordingly, an efficient quantization is performed, with respect to the faces of several types of signals of sound sources, giving priority to the desired frequency band, or to a depth of quantization of a desired portion in a frequency band. As described above, since normalization a and quantifisation x, normalization ß and quantification y, and normalization? and quantization z in their respective stages, are carried out using standard tables TI, T2 and T3, which are created by the method described above, it is possible that they perform a normalization that reduces the excessive load of Normalization, according to the characterization of the signal is the object of the quantification, whereby the efficiency of the quantification is significantly improved and the quality of reproduction is also, signifisantely improved. In continuation, the crack of a deodifying apparatus, corresponding to the coding apparatus according to the fifth embodiment, is shown in Figs. 13 (b) and 23 (c). Corresponding to the stricture of the sodifisator apparatus shown in Figs. 20 (b) and 20 (s), is provided with the inverse quantization units x ', y', and z ', which receive the signal output proceeding from the quantifiers of the quantifier unit of the encoder, and reproduces the signals corresponding to the trains of coefficients of the respective frequency bands, in which the sequencing of the frequency signal is divided; a plurality of inverse normalization units', ß 'and?', which are provided for the reverse scaling units, multiply the coefficient trains of the output of the frequency characteristic signal sequence from the quantization units Inverse, by the normalized components, which are reproduced based on the codes that relate to the output of the normalization coming from the audio signal encoding device, and to the respective signals corresponding to the trains of respective coefficients, of the frequency characteristic signal sequence, before being coded; and an inverse freshness transformation unit (not shown), which receives the outputs from the respective inverse normalization units, and outputs a signal corresponding to the original audio input signal. Subsequently, when inverse normalization and inverse quantization are performed in the decoding apparatus, alternately directed to, extending the band to be quantized and increasing the depth of the quantization, as described in the fourth embodiment, a desired amount of data of the encoded audio signal, in spite of the amount of data that is transmitted from the encoding device. That is, the amount of data that is going to be decoded, can be variable, in accordance with the environment of comunisasidn or similar circumstances, in the receiving terminal, a stable high definition sound quality can be obtained, even when it is used an ordinary public telephone network. In this fifth embodiment, each of the first, second and third quantization band selection units 1803, 1806 and 1809 can be constructed in such a way that they leave a frequency band to be quantified, which is determined in advance. In this case, each of the first, second and third quantization band selection units 1803, 1806 and 1809, is not If there is no need to obtain a band to be sutured and a band of freshness to be quantified, previously established, they are what the structure is simplified. More than one, each of the first, second and third units of selection of band of their antisense 1803, 1806 and 1809, can be constructed in such a way that it exploits a band separation, using two sarasterlsticas, that is, the minimum audible characteristic in silence and the masking characteristic, which makes sounds of frequency components, in the vicinity of a non-audible input frequency component, in order to obtain, as an output, a band of frequency to be quantified, in the basis of the auditory faces of human beings. According to the audio signal coding apparatus of the fifth embodiment, standardization means are provided before each quantifisation means that carry out quantifisation in multiple stages, and normalization is carried out for each of the divided frequencies and for each stage of quantification, followed by quantification. Accordingly, the normalization for each frequency domain allows an appropriate sodification according to the amount of data that the original audio signal possessed in each frequency domain, that is, the normalization means and the quantifisation means effect an encoding , at the same time to exhibit the whole dream capacities, with which a quantification of high efficiency is carried out, without losing the amount of data that the original signal had and in coneesuencia, with minor quantification errors, with the result of obtaining stable high-definition sound quality. In the first to fifth modalities, the multi-stage quantification means has, at least, a first-stage vector quantifier, to vestifially quantify the characteristic frequency signal sequence or a portion of it, and a second-stage vector suantifier. to quantitatively quantify a quantization error component, generated by the first stage vector quantifier; and each of the means of quantification of multiple stages, is provided with at least one divided vector quantifier, which quantifies vectorially a train of coefficients of any of the multiple bands of fresuensia, the suals are obtained by dividing the sequence In the case of frequency frequencies in the two frequency bands, the suals may have an overlapped portion between the multiple stages, according to a form of division for each stage. However, each one of the means gives quantification of each stage, of the means of quantification in multiple stages, it can be a vector suantifier of complete band, to quantify vectorially, all the sequence of eeñal characteristics of fresuensia of all the quantification error components, which come out of the means of suantifisation of the previous stage. Even when multistage quantization is carried out by the full band vestifial quantifier, to encode an audio signal at a low bitrate, that is, at a high compression ratio, as described above, or even when Coding and deodification are carried out without a fixed amount of data, high definition sound can be played at the receiving terminal. As described above, according to the coding apparatus and the audio signal decoding apparatus, or the audio signal coding and decoding method of the present invention, a structure capable of screening the quantification is provided for quantification. , even at high data compression constraints, by using, for example, a method of vectorial quantification and used to attribute a quantity of data during a quantification, to attribute, alternatively, a sanctity of data that are distributed to the expansion of data. a reproduced band and a quantity of data that contribute to the improvement of the quality. First, in the encoding apparatus, and as a first step, an input audio signal is transformed to a signal in the frequency domain and a portion of the frequency signal is coded; in the second stage, a portion of the frequency signal without coding and a quantization error signal, are encoded and added to the codes obtained in the first stage; in the third stage, the other portion of the unencrypted frequency signal, and the quantization error signals of the first and second stages, are coded and added to the codes obtained in the first and second stages; followed by a similar coding in the most advanced stages. On the other hand, in the decoding apparatus, any of the decoding forms following is possible: decoding using only the codes decoded in the first stage, deodification using the codes decoded in the first and second stages, and deodification using the decoded codes of the decoder. the first stage to the third or more stages. The order of decoding is to deodify, alternately, the codes that contribute to the expansion and the codes that contribute to the improvement of quality. Because of this, a satisfactory sound quality is obtained, even when the coding and decoding are carried out without a fixed amount of data. In addition, a high sound output at a high compression ratio is obtained. Moreover, according to the coding apparatus and the audio signal decoding device, or the method of audio signal decoding and decoding of the present invention, normalization means are provided, rather than quantization means, and the Normalization of an input audio signal is carried out before quantization. Therefore, the normalization means and the means of quantification carry out a sodifisation, while they exhibit all their capacities, so that a highly efficient quantization is carried out, without the loss of the amount of data, that the audio signal possessed. original and consequently are less errors of suantifisasión and this efesto being dramatically elevated, according to the type of audio signal. Moreover, as described above, the attribution of the quantity of data during the quantification is carried out in such a way that an amount of data is attributed, alternately, that contributes to the expansion of a reproduced band and an amount of data that contributes to the improvement of quality. When a quantity of data is limited in a receiving terminal, an inverse quantization is performed only in a narrow band and in a surface region. However, by expanding the inverse quantization alternately in one direction, to expand the band and in a direction to increase the depth of the inverse quantization to increase the amount of data in a reseption terminal, a desired amount of data may be desodified. the audio signal encoded, despite the amount of data transmitted from the encoding device. Accordingly, by varying the amount of data that will be decoded, according to the communication environment or similar circumstances, in the reception terminal, a stable high definition sound quality can be obtained, even when an ordinary public telephone network is used- Moreover, in accordance with the coding apparatus and the decoder apparatus of audio signal, the audio signal coding and deodification method of the present invention, standardization means are provided before each of the quantization means that effect a multistage quantization, and normalization is performed on each one of the divided frequencies and for the stage of the quantification, followed by the quantification. Consequently, the normalization of each frequency domain allows an appropriate coding according to the amount of data that the audio signal had in each frequency domain, that is, the quantization means and the normalization means efestuen an encoding, while they exhibit their total capacities, which makes a quantification with a high efficiency, without the loss of the amount of data that the original audio signal possessed and consequently, with less quantification errors, with the result of obtaining a quality of stable high-definition sound, and the effect is dramatically high, according to the type of audio signal-Additionally, at the completion of the coding, when the normalization invends and the inverse quantization are in the direction of increasing the band, as well as in the direction of increasing the depth of the quantifisation, alternatively, a desired amount of data can be decoded from the encoded audio signal, despite the amount of data that is transmitted from the coding apparatus. Accordingly, by varying the amount of data to be decoded, in accordance with the communication environment or similar circumstances, in the reseption terminal, stable high-definition sound quality can be obtained, even when a telephone network is used. ordinary public Applicability in the Industry As described above, according to the coding apparatus and the audio signal deodifying device, or the sodifisation and decoding method of the present invention, even if the encoding and decoding are carried out without a fixed amount of data, a satisfactory sound quality is obtained, in addition to obtaining a high sound quality at a high compression ratio. Moreover, the normalization of each frequency domain allows an appropriate coding according to the amount of data that the audio signal had in each frequency domain, with which the means of normalization means perform an encoding, while displaying all of their capabilities, which is done quantification with high efficiency, without losing the amount of data that had the original audio signal and in sonsesuencia, with fewer errors of quantifisation. This effect is dramatically increased according to the type of audio signal. Accordingly, by varying the amount of data to be decoded, according to the communication environment or similar circumstances, stable high-definition sound quality can be obtained at the receiving terminal, even when using a public telephone network ordinary It is noted that, as they relate to this date, the best method conosido by the requested, to carry out the present invention, is that which is clear from the present, discovering the invention. Having described the invention as above, the content of the following is claimed as property.

Claims

CLAIMS An audio signal coding device for sounding an audio signal, by means of performing a vector quantization to a frequency characteristic signal sequence, which is obtained by subjecting an input audio signal to a frequency transformation, Considered because it suffers: a multi-stage suantifisation means, including at least one first-stage vector suantifisator, to quantitatively quantify the frequency signal sequence or a portion of it, and a second-stage vector quantifier to quantify vectorially a quantification error component, generated by the first stage vector quantifier; and each of the means of quantifisation of each stage, of the quantifiable means in multiple stages, have at least one vector quantifier divided, the vector quantifies vectorially to a train of coefficients in any of the multiple frequency bands, obtained by means of dividing the signal sequence characteristic of frequency, in at least two frequency bands that may have an overlapped portion between the multiple stages, using a division form for each stage. The audio signal coding apparatus according to claim 1, characterized in that it further comprises a normalization means for normalizing the characteristic frequency signal sequence and adding its output to the multistage suantiflowing means. The audio signal encoding apparatus, of soundness, are Claims 1 and 2, characterized in that the means of quantization of each stage appropriately select, as a frequency band divided from the signal sequence frequency characteristic which is to be quantified. , a band having a large sum of energy addition of the quantization error, and thus quantifies the selected band. The conformance audio signal encoding apparatus are Claims 1 and 2, characterized in that the means of quantification of stage, selessionan a band of fresuensia, in the manner of a fresuensia divided from the sequential signal of characteristic fresuensia that will be quantified, based on a characteristic of auditory sensitivity that shows the auditory nature of human beings, and quantifies the selected band, whose selected frequency band has a large sum of energy addition of a quantization error, measured by giving a large value to a band that has a high importance within the auditory sensitivity facet. The audio signal encoding apparatus according to any of Claims 1 to 4, characterized in that it further comprises: a first quantization band selection unit, between the first stage vector quantizer and the second stage vector quantizer, the which are constituent of the means of quantification in multiple stages; and a second quantization band selection unit, between the second stage vector quantizer and the terrestrial stage vestifial quantifier; wherein the first quantization band selection unit receives, as an input, the quantization error output from the first stage vestifial quantifier and selects a band to be quantized, by the second stage vector quantizer; and the second quantization band selection means receive, as an input, the quantization error output from the second stage vector quantizer and select a band to be quantized, by the third stage vector quantizer. The audio signal coding device of claim 5, characterized in that the multi-stage quantization means comprise: a plurality of divided vector quantifiers of stage i, which quantify independently, the respective coefficient trains in the frequency bands respective, in which the frequency characteristic signal sequence is divided; Y a vector quantizer of stage j, which serves as a quantifier of the entire band, to quantify, once at least, all the respective frequency bands of the input signal to be quantized. The audio signal coding apparatus according to claim 1, characterized in that in the multistage quantization means, the vectorial quantifier of the previous stage calculates a quantifisation error in vestigial quantifisation, using a vestorial quantification method is a The codebook and the vector quantifier of the subsequent stage perform an additional vector quantization to the calculated quantization error. The audio signal coding apparatus according to Claim 2, characterized in that in the multistage quantization means, the vectorial quantifier of the previous stage calculates a quantification error in vestigial quantification, using a vector quantification method with A codebook and the vectorial quantifier of the subsequent stage perform a vectorial quantification additional to the calculated quantification error. The audio signal coding apparatus according to Claim 8, characterized in that when calculating distances between codes used to remove an optimal code in the codebook during a vector quantization, the vector quantizer of the multistage quantization means, calculates the distances used, in the form of peo, by the normalized components of the output of the input signal from the normalization means, and extracts a code that has the minimum diet. The audio signal coding apparatus according to claim 9, characterized in that the vector quantizer of the multistage quantization means calculates the distances that use, in the form of weight, both normalized components of the output of the signal sequence. Characteristic of frequency, coming from the means of normalization and a value based on the auditory facetry that shows the auditory nature of the human being and extracts a code that has a minimum distance. The audio signal encoding apparatus according to any one of the Claims 2 and 8 to 10, characterized in that the normalization means are provided with a frequency scheme normalization unit, which grossly normalizes the scheme of frequency signal samplerization. The audio signal encoding apparatus according to any of Claims 2 and 8 to 10, characterized in that the normalizing means is provided with a band-width normalization unit, which divides the signal sequence, the frequency characteristic, in a plurality of components of continuous unit frequency bands and normalizes the train of coefficients in each unit band, by dividing them with a single value. The sonicity audio signal coding apparatus is Claim 1, characterized in that the means of quantifisation in multiple stages include: a vector suantifier, which quantifies the respective coefficient trains of the respective frequency bands, in which the sequence of Frequency facet signal is divided, independently by divided vector quantifiers; Y a vector quantizer, which serves as a full-band suantifisator to quantify, once at least, all the respective frequency bands of the input signal to be quantized. The audio signal coding apparatus according to claim 13, characterized in that the quantification means in multiple stages comprise: a first vector quantizer, comprising a vector splitter of low band, a vector quantizer divided by intermediate band and a quantifier vector divided from high band; a second vectorial quantifier conestado next of the first vectorial suantificador; and a vesturial suantifiser terser connected immediately after the second vectorial quantifier; where an entry of the sequencing of the signaling function of the fresuensia directed to the means of quantification in multiple stages, is divided into three bands, and the signal sequence characteristic of frequency of the low band component which is between the three bands, is quantified by the low-band divided vector quantifier, the signal sequence frequency characteristic of the intermediate band component that is between the three bands, is quantized by the intermediate band divided vector quantizer, and the frequency characteristic signal sequence of the band component. high band that is between the three bands, is quantified by the vector quantifier divided by high band, independently; each of the divided vector quantifiers that make up the first vector quantifier calculates a quantization error, with respect to the signal sequence characteristic of frequency, and sends this error to the second subsequent vestorial quantifier; the second vector quantifier performs a quantification, so that a bandwidth is quantified by the second vestorial quantifier, a quantification error results, with respect to the input of the second vector quantifier and this error exits towards the third vestifial quantifier; and the vector quantizer terser performs a quantization, so that a bandwidth is quantified by the third vector quantifier. The audio signal encoding apparatus according to Claim 14, characterized in that it further comprises; a first band quantization selection unit, disposed between the first vector quantifier and the second vector quantizer, which are constituents of the quantification means in multiple stages, and a second unit of selection of band quantization, arranged between the second vector quantifier and the third vector quantifier; wherein the first band quantifisation selection unit resides, in the form of an input, the prosecutorial output of the first vector suantifier and selects a band to be quantified by the second vector suantifier; the second vector quantifier effects a quantification of an amplitude of band, to be quantified by the second vector quantifier, with respect to the suantification errors from the first vector quantifier, which comprise the three vector suantifisers, in the band selected by the first band-sorting unit and thus , it outputs a quantization error with respect to the input of the second vector quantizer and sends this error to the second band quantization detection unit; the second band quantization selection unit receives, in the input form, the quantization error from the second vector quantizer and selects a band, to be quantified by the vector quantizer terser; and the third vector quantifier performs a quantifism of a bandwidth, to be quantified by the third vector quantifier, with respect to the quantization error from the second vector quantifier, in the band selected by the second band quantization selection unit. The audio signal coding apparatus, according to claim 14, characterized in that instead of the first vector quantizer, the vector quantizer or the third vector quantizer are constructed, using the low-band divided vector quantizer, to the vector divided quantizer of intermediate band, or the high band divided vector quantifier. An audio signal decoding apparatus that receives, as an input, the output of codes from the audio signal encoding apparatus, in accordance with Claim 1 and decoding these codes to output a signal corresponding to the input audio signal. original, characterized in that it comprises: an inverse quantization unit, which performs an inverse quantization, using at least a portion of the code output from the quantization means of the audio signal encoding apparatus; and a reverse frequency transformation unit, which transforms the output of a Signal frequency signaling from the inverse quantization unit, to a signal corresponding to the original audio input signal-18. An audio signal decoding device which, as an input, resides the output of the following codes of the audio signal sodipisers, according to Claim 2 and decoding these codes to output a signal corresponding to the input audio signal. original, characterized in that it comprises: a unit of inverse quantization, which reproduces a signal sequence characteristic of frequency; an inverse normalization unit, which reproduces the normalized components, based on the code output from the audio signal encoding apparatus, using the output of the frequency characteristic signal sequence, from the inverse quantization unit, and multiplies the frequency characteristic signal sequence and the normalized components; and a reverse transformation frequency unit, to support the output from the reverse normalization unit and tracing the the signal sequence characteristic of frequency, in a signal corresponding to the original audio input signal. - An audio signal decoding apparatus that receives, as an input, the output of the codes coming from the audio signal encoding apparatus according to Claim 13, and decoding these codes to output an eefial corresponding to the input signal of the original audio, and the decoding apparatus comprises: a reverse quantization unit, which performs an inverse quantization using the output codes, when all the codes are graduated from all the vector quantifiers that constitute the quantification means of the coding apparatus of audio signal, or some of this. . The audio signal decoder apparatus according to Claim 19, characterized in that: subsequent to the inverse quantization of the quantized codes in a preset band in a specified stage, the inverse quantization unit executes, alternately, a inverse quantification of the quantized codes in the preset band in a next stage, and the inverse quantifisation of the quantized codes in a band different to the preset band in the specified stage; When there are no longer any subletized codes in the preset band in the next step, the inverse quantization unit continuously executes an inverse quantifisation of the suantified codes in a different band; and when there are no longer suantified codes in the band other than the preset band, the inverse suturing unit continuously executes an inverse quantization of the quantized codes in a next stage. An audio signal deodifying device which, as an input, resides the code output from the audio signal encoding device according to Claim 14 and decodes these codes to output a signal corresponding to the original audio input signal. , characterized in that it comprises: an inverse quantization unit, which performs an inverse quantization, with only the code output from the low band divided vector quantizer, as a constituent of the first vector quantizer, even though all or some of the three divided vector quantifiers constitute the first vector quantizer in the output codes of the audio signal encoding apparatus. The audio signal decoder apparatus according to Claim 21, characterized in that the inverse quantifisation unit effects an inverse superslift, by outputting codes from the second vector quantifier, in addition to the code output from the divided vector quantizer of low band, as a constituent of the first vector quantifier. The audio signal decoding apparatus according to Claim 22, characterized in that the inverse quantization unit performs an inverse quantification, using the code output from the intermediate band divided vector quantifier, as a sonic of the first vector quantifier, in addition to the output of codes from the low-band divided vector quantizer, as a constituent of the first vector quantifier and to the code output from the second vector quantifier. The audio signal decoding apparatus according to Claim 23, characterized in that the inverse quantization unit performs an inverse quantization, using the code output from the third vectorial quantifier, in addition to the code output from the divided vector quantizer of low band, as a constituent of the first vestigial quantifier, the output of codes from the second vestifial quantifier and the code output from the intermediate band divided vector quantizer, as a constituent of the first vector quantifier. The audio signal decoding apparatus according to Claim 24, characterized in that the inverse quantization unit performs an inverse quantization, using the code output from the high-band divided vector suantifisator, as a constituent of the first vector quantifier, in addition to the output of the prosedente codes of the low-band divided vector quantizer, as a constituent of the first vector quantifier, the code output from the second vector quantizer, the code output from the intermediate band divided vector quantizer, as a constituent of the first vector quantizer and the code output from the third vector quantifier-A signal coding and decoding method audio, which receives a signal sequence characteristic of frequency, obtained by means of frequency transformation of an input audio signal, coding and outputting this signal and decoding the output encoded signal, to reproduce a signal corresponding to the input signal of original audio, characterized in that: the frequency characteristic sequence is divided into coefficient trains, corresponding to at least two frequency bands, and these soefficient streams are independently quantized and graduated; and from the quantized signal received, the data of an arbitrary band corresponding to the divided band are inversely quantized, preferably to reproduce a signal corresponding to the original audio input signal. The audio signal coding and decoding method according to claim 26, characterized in that: the quantization is carried out in stages, in such a way that a calculated quantization error is further quantified; and the inverse quantification is carried out by means of repeating, alternately, the sutifisation directed to expand the band and the quantification directed to deepen the quantification stages, in the quantification. The encoding and decoding method of audio signal according to Claim 27, characterized in that the inverse quantifisation, aimed at expanding the band is carried out via the expansion of the band to the sense in respect to the auditory physiological characteristics of the band. The seree humanoe. The audio signal encoding and decoding method in accordance with any of the Claims 26 to 28, characterized in that: at the termination of the coding, after the signal sequence characteristic of frequency is normalized, the signal sequence characteristic frequency is divided into trains of soefficients, corresponding to at least two frequency bands, and the trains of resilient coefficients are independently sutured and graduated; and in the termination of the deodifisation, using the codes in relation to the normalization, coming from the termination of the codification, the codes coming from the termination of the codification are subjected to an inverse normalization and thus, the inversely normalized codes are subject to a quantifisasión inverea, in such a way that the data that are in an arbitrary band corresponding to the divided band are inversely quantized and with this, they reproduce a signal corresponding to the original audio input signal. An audio signal coding apparatus for encoding an audio signal, by performing a normalization and vector scaling to a frequency characteristic signal sequence, which is obtained by subjecting an input audio signal to a signal transformation. frequency, characterized in that it comprises: multi-stage quantification means comprising, at least, one unit of vector suantifisasión and first stage normalization, sual normalizes and vestorially quantifies the signal frequency characteristic signal or a portion of it, and a second unit of vector quantization and second stage normalization, sual normalizes and suantifisa vectorially to a somantifier error of suantifisation, generated by the first unit of vector quantification and of first stage normalization; and means for quantifying each step of the means for quantification in multiple stages, having at least one unit of vector quantization and of divided normalization, which normalizes and quantifies vectorially a train of coefficients, of any of the multiple bands of frequency, obtained by dividing the characteristic frequency signal sequence into, at least, two frequency bands, which may have an overlapped portion between the multiple stages, using a division type for each stage. The audio signal coding apparatus according to Claim 30, characterized in that it further comprises: a first unit of selection of band suantifisation, between the unit of vector quantization and of first stage normalization, and the unit of vector quantification and of second stage normalization, which are constituents of the means of quantification in multiple stages; and a second band quantization selection unit between the vector quantization unit and the second stage normalization unit and the vestigial quantifism unit and the third stage normalization unit; wherein the first band quantization selection unit receives, as an input, the output of the quantization error proceeding from the vector quantization unit and the first stage normalization, selects a band to be quantified by the vector quantization unit and second stage normalization, and a quantization error in the selected band, towards the unit of vector quantization and normalization of the second stage; Y the second band quantization selection means receive, as input, the output of the quantization error proceeding from the vestorial quantization unit and second stage normalization, selects a band to be quantized by the vector quantization unit and of third stage normalization, and a quantization error in the selected band, towards the unit of vector quantization and standardization of third stage. The audio signal encoding apparatus according to Claim 31, characterized in that each of the vector quantization and normalization units, within and after the second stage of the multistage quantifisation means, appropriately selects a frequency band. having a large sum of energy addition of the output of the quantization error, proceeding from the unit of vectorial suantifisation and of normalization of the previous stage, from the frequency bands -parted from the signal sequence characteristic of frequency that has to be standardized and suantifidated, and then the unit normalizes and suantifies the selected band. The audio signal coding apparatus according to Claim 31, characterized in that it outputs one of the vector standardization and normalization units, within and after the second stage of the multistage quantization means, appropriately selects a frequency band. , from the frequency bands divided from the signal sequence characteristic of frequency that has to be normalized and suanified, based on the characteristic of auditory sensitivity, which shows the auditory nature of human beings, whose frequency band is seized, it has a large sum of energy input from the output of the error of its quantification, proceeding from the unit of vestigial quantification and of normalization of previous and measured stage, when giving a large value to a band that has a high importance of characteristic of auditory sensibility and thus, the unit normalizes and quantifies the selected band. An audio signal decoder that receives, as an input, the output of codes from the audio signal encoding apparatuses, according to claims 30 to 33, and decoding the codes to output a signal corresponding to the original audio input signal, costing because somprende: units of inverse quantization, to receive signals of its respective quantifiers, in the quantization unit of the audio signal coding apparatus, and reproducing the signals corresponding to the coefficient trains of the respective frequency bands, in which the frequency characteristic signal sequence is divided; a plurality of inverse normalization units correspondingly provided to the respective inverse quantization units to multiply the coefficient trains of the frequency characteristic signal sequence outputs from the respective inverse quantization units by means of reproduced standardized components at the base of the codes related to the normalization and to the output from the audio signal encoding device, and outputting the signals corresponding to the coefficient trains respective, of the frequency signaling frequency sarasteristy, before being codified; and a fresh inverse transform unit, to receive the outputs from the multiple inverse normalization units, and transforming them, into signals corresponding to the original audio input signal. An audio signal coding and decoding method for receiving a signal sequence characteristic of frequency, obtained by frequency transformation of an audio signal of input, Sodifying and outputting this signal, and deodifying the output encoded signal, to reproduce a signal. signal corresponding to the original audio input signal, characterized in that: the signal sequence characteristic of frequency is divided into trains of coefficients, corresponding to at least, two frequency bands, and this trene of coefficients are normalized, quantified and graduated independently; and from the quantifiedae document received, using the codes related to the normalization and the output of the finished coding, the data of an arbitrary band corresponding to the divided band, are normalized and inversely quantized, in such a way that a signal corresponding to the original audio input signal is reproduced. The audio signal coding and decoding method according to claim 35, characterized in that: the normalization and the quantization are carried out in stages, in such a way that the calculated quantization errors are normalized and quantified further; and normalization and inverse quantifisation, are carried out by means of repeating, alternately, the normalization and the suantifisation directed to expand the band and the normalization and the inverse surtization directed to deepen the quantification stages, in the quantification. The audio signal encoding and decoding method according to Claim 36, characterized in that the inverse normalization and quantization, aimed at expanding the band are carried out, by expanding the band to the sense are respected, at auditory physiological characteristics of humans. An audio signal coding device for sounding an audio signal, by performing vestorial quantization at a signal frequency characteristic signal, which is obtained by submitting an input audio signal, to a frequency transformation, and the apparatus appears: freshness transformation, characterized in that it comprises: quantification means in multiple stages, comprising at least one first-stage vector quantifier, to vestorially quantify the frequency characteristic signal sequence or a portion of it, and a Vector quantifier of the second stage to quantify a quantity quantifier error quantifier, generated by the first stage vector quantifier; and each one of the quantification means of each stage, of the means of quantification in the multiple stage, comprises a vector quantifier of a complete band, to quantify vectorially, all the frequency signal sequences of all frequencies. Quantification error components, which come out of the quantification means of the previous stage.