Audio signal encoding method, audio signal decoding method, encoding device, decoding device, audio signal processing system, audio signal encoding programme and audio signal decoding programme

FIELD: information technology.

SUBSTANCE: when a frame immediately preceding a target encoding frame to be encoded by a first encoding unit operating according to a linear predictive coding scheme is encoded by a second encoding unit operating according to a coding scheme different from the linear predictive coding scheme, the target encoding frame can be encoded according to the linear predictive coding scheme by initialising the internal state of the first encoding unit. Consequently, encoding processing performed according to a plurality of coding schemes including the linear predictive coding scheme and a coding scheme different from the linear predictive coding scheme can be realised.

EFFECT: improved speech quality.

7 cl, 5 dwg

The technical field to which the invention relates.

The present invention relates to a method of encoding an audio signal, the method of decoding an audio signal, the encoding device, decoding device, the processing system audio program audio encoding and decoding program audio.

The level of technology

Encoding method for compressing speech/music signals (audio signals) with a low bit rate is important to reduce the costs incurred if bilateral transmission, broadband transmission and preservation of speech and music signals. In order to effectively encode both speech and music signals, an effective coding scheme of a hybrid type, in which the selectively use a coding scheme suitable for speech signals, and encoding scheme suitable for musical signals. The encoding scheme of a hybrid type effectively performs encoding by switching the encoding schemes in the encoding process of autopaletteselect, even when the characteristics of the input signals change in time.

The encoding scheme of a hybrid type usually includes as a component diagram of the CELP coding (CELP: encoding with linear prediction and code excitation)that is appropriate to the financing of speech signals. Usually in order to encode the residual signal obtained by applying the inverse filter with linear prediction to the input signal, the encoder performs CELP scheme, retains information about the past of the residual signals in the adaptive code book. Because of the adaptive code book is used for coding, achieves high coding efficiency.

Method for coding speech signals and music signals described in patent literature 1. In patent literature 1 describes an algorithm for encoding as speech signals and music signals, etc. the Method described in patent literature 1 uses a filter synthesis with linear prediction (LP), which operates in the usual way to encode speech signals and music signals. Filter synthesis with LP switch between the generator excitation speech and generator excitation conversion in accordance with the coded whether the speech signal or a music signal, respectively. For coding of speech signals using traditional CELP method, and coding of music signals applied a new asymmetric method of overlapping summation. When performing a conventional filter synthesis with LP interpolation coefficients LP is relative to the signal in the areas of operation overlap summation.</>

When switching from the coding scheme different from the CELP encoding scheme, encoding scheme, carrying out the CELP scheme, the encoding process of autopaletteselect, information about residual signal corresponding to the speech coming before the switch is not stored in the adaptive code book in the encoder. As a result, the coding efficiency deteriorates when encoding frame, coming directly after switching the coding scheme that leads to the problem of reducing the quality of the reproduced speech. The solution in the prior art it is known as adaptive multirate wideband plus (AMR-WB+, non-patent literature 1), which is a coding scheme, a standardized partnership Project 3rd generation (3GPP), in which the internal state of the encoder performs CELP scheme, initialize using the encoded result obtained according to the encoding scheme that is different from the CELP scheme. Encoder AMR-WB+ receives the residual signal by inverse filtering with linear prediction with respect to the input signal and then selectively encodes the residual signal using two coding schemes, i.e. schemes CELP scheme coded excitation conversion (TLC). When switching from the scheme on TLC scheme CELP coder AMR-WB+ updates the adaptive code is the first book in the CELP scheme using the excitation signal in the scheme of TLC.

List of links

Patent literature

Patent literature 1: Japanese laid patent application No. 2003-44097.

Non-patent literature

3GPP TS 26.290 “Audio codec processing functions; Extended Adaptive Multi-Rate-Wideband (AMR-WB+) codec; Transcoding functions”. [online].[retrieved 5 March 2009]. Found on the Internet:<URL:http://www.3gpp.org/ftp/Specs/html-nfo/26290.htm>

Disclosure of inventions

Technical problem

However, according to the encoding scheme of a hybrid type, in which selectively using a coding scheme based on the CELP scheme, and a coding scheme that does not use coding with linear prediction, it is difficult to obtain the excitation signal of the encoding process performed according to the coding scheme that does not use coding with linear prediction. Consequently, when switching from the coding scheme that does not use coding with linear prediction, coding scheme, based on the CELP scheme, it is difficult to initialize the adaptive code book in the CELP scheme with the excitation signal corresponding to the speech that comes to switching. The present invention is to initialize to an appropriate value of the internal state means the encoding or the decoding means performing encoding using the encoding with linear prediction in order, therefore, to improve the speech quality, reproducibility is th frame, coming directly after switching, when switching from the coding scheme that does not use linear prediction, coding scheme using linear prediction.

Solution

A method of coding an audio signal of the present invention encodes the audio signal includes multiple frames, using the first encoding means working according to the encoding scheme with linear prediction, and second encoding means working according to the encoding scheme that is different from the encoding scheme with linear prediction. A method of coding an audio signal of the present invention includes a step of switching to switch encoding means for encoding the second frame immediately following the first frame from the second encoding means on the first means of coding, after the first frame of the audio signal has been encoded using a second encoding means, the initialization phase to initialize the internal state of the first encoding means, in accordance with a predefined method, after the step switch.

In accordance with the encoding method of the audio signal of the present invention, even when the second frame should be encoded according to the encoding scheme with linear predicted the eat, while the first frame has been encoded using an encoding scheme that is different from the encoding scheme with linear prediction, the second frame may be encoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first encoding means working according to the encoding scheme with linear prediction. This can be realized by the processing of the encoding performed with many encoding schemes, including coding scheme with linear prediction and the encoding scheme that is different from the encoding scheme with linear prediction.

In the present invention, the internal state of the first encoding means preferably contains the content of the adaptive codebook or the value stored by means of the delay elements of the filter synthesis with linear prediction to determine the response to zero input. During initialization of the internal state of the first encoding means preferably initialize using the first frame. In the initialization phase, the first encoding means preferably initialize using the residual signal obtained by applying the inverse filter with linear prediction or to the first frame, still subject to encoding by the second encoding means or to the first frame, decoded back, after encoding by the second encoding means. In the initialization phase inverse filter with linear prediction is preferably applied either to the first frame, still subject to encoding by the second encoding means, or to the first frame, the decoded back, after encoding by the second encoding means, using the linear prediction coefficients used by the first encoding means to encode the third frame before the first frame. Alternatively, in the initialization phase, when the linear prediction coefficients of the first frame is included in the codes of the second frame, the inverse filter with linear prediction is preferably applied either to the first frame, still subject to encoding by the second encoding means, or to the first frame, the decoded back, after encoding by the second encoding means, using the linear prediction coefficients included in the codes of the second frame. In the present invention, during initialization of the internal state of the first encoding means can be initialized using the internal state, which had first encoding means, when the first encoding means encodes the frame, predshestvuyuschih the first frame. As for the coefficients of the linear prediction filter synthesis with linear prediction to determine the response to a zero input, it is desirable to use the linear prediction coefficients used when the first encoding means encodes the third frame before the first frame, or when the linear prediction coefficients of the first frame is included in the codes of the second frame, the linear prediction coefficients of the first frame is calculated, when the encoded second frame, or the linear prediction coefficients obtained by applying the perceptual weighting filter to the computed coefficients of linear prediction.

The method of decoding an audio signal of the present invention decodes the audio signal includes multiple frames, using the first decoding means operating according to the encoding scheme with linear prediction, and the second decoding means operating according to the encoding scheme that is different from the encoding scheme with linear prediction. The method of decoding an audio signal includes a step switch for switching the decoding means for decoding the second frame immediately following the first frame, second means for decrypting the first decoding means, after the first CA is p of the encoded audio signal has been decoded by the second decoding means, and phase initialization to initialize the internal state of the first decoding means in accordance with a predefined method, after the step switch.

In accordance with the method of decoding an audio signal of the present invention, even when the second frame should be decoded using the coding scheme with linear prediction, while the first frame was decoded using the encoding scheme that is different from the encoding scheme with linear prediction, the second frame can be decoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first decoding means operating according to the encoding scheme with linear prediction. Therefore can be implemented in the decoding processing performed according to the number of encoding schemes, including coding scheme with linear prediction and the encoding scheme that is different from the encoding scheme with linear prediction.

In the present invention, the internal state of the first decoding means preferably contains the content of the adaptive codebook or the value stored by means of the delay elements of the filter synthesis with linear prediction to determine the response to zero input. In the initialization phase within Annee state of the first decoding means preferably initialize using the first frame. In the initialization phase, the first decoding means preferably initialize using the residual signal obtained by applying the inverse filter with linear prediction to the first frame decoded by the second decoding means. In the initialization phase inverse filter with linear prediction is preferably applied to the first frame decoded by the second decoding means, using the linear prediction coefficients used by the first decoding means to decode the third frame before the first frame. Alternatively, in the initialization phase, when the linear prediction coefficients of the first frame is included in the codes of the second frame, the inverse filter with linear prediction is preferably applied to the first frame decoded using the second encoding means, using the linear prediction coefficients included in the codes of the second frame. In the present invention, during initialization of the internal state of the first decoding means can be initialized using the internal state, which had first decoding means when the first decoding means has decoded frame preceding the first frame.

Device to the financing of the present invention includes first encoding means, working according to the encoding scheme with linear prediction, and second encoding means working according to the encoding scheme that is different from the encoding scheme with linear prediction, and encodes the audio signal using the first encoding means and the second encoding means. The encoding device includes first means for determining the encoding, which determines whether the first or the second encoding means to encode the target frame coding, which is included in the audio signal. The encoding device of the present invention further comprises second means for determining the encoding, which determines if the first means for determining the encoding determined that the target frame encoding must be encoded using the first encoding means, whether encoded frame immediately preceding the target frame coding using the first encoding means or the second encoding means, and a means of calculating the internal state of the encoding, which decodes, if the second means for determining the encoding has determined that the immediately preceding frame has been encoded using a second encoding means, the encoded result of the immediately preceding frame, and calculates the internal state of the first media is TBA encoding using the decoded result. The encoding device of the present invention further comprises means for initializing coding, which initializes the internal state of the first encoding means using the internal state calculated by means of calculating the internal state of the encoding. First encoding means encodes the target frame after encoding means initialization encoding.

In accordance with the encoding device of the present invention, even when the target frame encoding must be encoded using the first encoding means working according to the encoding scheme with linear prediction, while the frame immediately preceding the target frame encoding, encoded using the second encoding means working according to the encoding scheme that is different from the encoding scheme with linear prediction, the target frame encoding may be encoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first encoding means. Therefore, it may be implemented processing of the encoding performed according to the number of encoding schemes, including coding scheme with linear prediction and the encoding scheme that is different from the encoding scheme with linear prediction.

The device on the encoding of the present invention includes first decoding means, working according to the encoding scheme with linear prediction, and the second decoding means operating according to the encoding scheme that is different from the encoding scheme with linear prediction, and decodes the encoded audio signal using the first decoding means and the second decoding means. The decoding device includes first means for determining the decoding, which determines whether the first decoding means or the second means of encoding to decode the target frame decoding, which is included in the encoded audio signal. The decoding device also includes a second means for determining the decoding, which determines if the first means for determining the decoding has determined that the target frame decoding must be decoded by the first decoding means, whether the decoded frame immediately preceding the target frame decoding, using the first decoding means or the second decoding means. The decoding device further comprises means of calculating the internal state of the decoder, which computes, if the second means for determining the decoding has determined that the immediately preceding frame decoded using the second means, decode the simulation, the internal state of the first decoding means using the decoded result of the immediately preceding frame, and means for initialization of the decoding, which initializes the internal state of the first decoding means using the internal state calculated by means of calculating the internal state of the decoder. First decoding means decodes the target frame decoding, after its internal state is initialized using the initialization decoding.

In accordance with the decoding device of the present invention, even when the target frame decoding must be decoded by the first decoding means operating according to the encoding scheme with linear prediction, while the immediately preceding frame has been decoded by the second decoding means operating according to a decoding scheme that is different from the encoding scheme with linear prediction, the target frame decoding can be decoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first decoding means. This can be implemented in the decoding processing performed according to the set of CX is m coding including coding scheme with linear prediction and the encoding scheme that is different from the encoding scheme with linear prediction.

System for processing an audio signal of the present invention includes an encoding device and the decoding device. The decoder decodes the encoded audio signal encoded by the encoder.

In accordance with the system processing of the audio signal of the present invention, even when the target frame encoding must be encoded using the first encoding means working according to the encoding scheme with linear prediction, while the immediately preceding frame has been encoded using a second encoding means working according to the encoding scheme that is different from the encoding scheme with linear prediction, the target frame encoding may be encoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first encoding means. Even when the target frame decoding must be decoded by the first decoding means operating according to the encoding scheme with linear prediction, while the immediately preceding frame has been decoded by the second decoding means operating in accordance with the scheme is tiravanija, other than coding scheme with linear prediction, the target frame decoding can be decoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first decoding means. This can be implemented processing of encoding and decoding processing performed according to the number of encoding schemes, including coding scheme with linear prediction and the other encoding scheme that is different from the encoding scheme with linear prediction.

Program audio encoding of the present invention for encoding an audio signal using a first encoding means working according to the encoding scheme with linear prediction, and second encoding means working according to the encoding scheme that is different from the encoding scheme with linear prediction, makes a computer device to function as the first means for determining the coding to determine whether the first encoding means or the second encoding means to encode the target frame coding, which is included in the audio signal; second means for determining the encoding to determine if the first means for determining the encoding determined that the target frame encoding should be the coding for the van with by using the first encoding means, was there a coded frame immediately preceding the target frame coding using the first encoding means or the second encoding means; a means of calculating the internal state encoding for decoding, if the second means for determining the encoding has determined that the immediately preceding frame has been encoded using a second encoding means, and encoding of the immediately preceding frame and calculating the internal state of the first encoding means using the decoded result; means initialization coding to initialize the internal state of the first encoding means using the internal state calculated by means of calculating the internal state of the encoding; and encoding means for coding the target frame encoding using the first means of coding, once initialized its internal state.

In accordance with the program audio encoding of the present invention, even when the target frame encoding must be encoded using the first encoding means working according to the encoding scheme with linear prediction, while the immediately preceding frame has been encoded using a second tool Cody is Finance, working according to the encoding scheme that is different from the encoding scheme with linear prediction, the target frame encoding may be encoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first encoding means. Therefore, it may be implemented processing of the encoding performed according to the number of encoding schemes, including coding scheme with linear prediction and the encoding scheme that is different from the encoding scheme with linear prediction.

The program is decoding the audio signal of the present invention for decoding the encoded audio signal using a first decoding means operating according to the encoding scheme with linear prediction, and second decoding means operating according to the encoding scheme that is different from the encoding scheme with linear prediction, makes a computer device to function as a first means for determining the decoding to determine whether the first decoding means or the second decoding means to decode the target frame decoding, which is included in the encoded audio signal; second means for determining the decoding to determine if the first means for determining the decoding has determined that the spruce frame decoding must be decoded with by using the first decoding means, whether the decoded frame immediately preceding the target frame decoding, using the first decoding means or the second decoding means; a means of calculating the internal state of the decoder to calculate, if the second means for determining the decoding determines that the immediately preceding frame has been decoded by the second decoding means, the internal state of the first decoding means using the decoded result of the immediately preceding frame; and means for initializing the decoding to initialize the internal state of the first decoding means using the internal state calculated by means of calculating the internal state of the decoder; and decoding means for decoding target frame to be decoded by the first decoding means, after you have initialized its internal state.

In accordance with the program for decoding the audio signal of the present invention, even when the target frame decoding must be decoded by the first decoding means operating according to the encoding scheme with linear prediction, while the immediately preceding frame has been decoded using the second CPE is of the decoding, working according to the encoding scheme that is different from the encoding scheme with linear prediction, the target frame decoding can be decoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first decoding means. Therefore can be implemented in the decoding processing performed according to the number of encoding schemes, including coding scheme with linear prediction and the encoding scheme that is different from the encoding scheme with linear prediction.

Advantages results inventions

In accordance with the present invention, when switching from an encoding scheme that does not use linear prediction, coding scheme, using coding with linear prediction, the internal state of the means of coding or the decoding means performing encoding using the encoding with linear prediction can be initialized with appropriate values, and can be improved speech quality, reproducible from shot coming directly after switching.

Brief description of drawings

Figure 1 - diagram depicting the configuration of the encoding device and the decoding device in accordance with the embodiment;

2 is a diagram depicting the configuration of the CA the device coding in accordance with the embodiment;

figure 3 - block diagram of the sequence of steps to describe the operation of the encoder, in accordance with the embodiment;

4 is a diagram depicting the configuration of a decoding device in accordance with the embodiment; and

5 is a block diagram of a sequence of steps to describe the operation of the decoding device, in accordance with the embodiment.

The implementation of the invention

Preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. In the description of the drawings the same elements are marked with the same reference numbers, if possible, and their description is not repeated. System for processing an audio signal, in accordance with the embodiment includes a device 10 coding, which encodes the input audio signal and the device 20 decoding which decodes the encoded audio signal encoded using the device 10 coding. Figure 1 and figure 2 is a diagram depicting the configuration of the device 10 coding, in accordance with the embodiment. The device 10 coding encodes the input speech/audio signal (audio signal), and outputs the encoded signal. Speech/music first audio signal is divided into frames is within a finite duration, and after that enter in the device 10 coding. The device 10 coding performs encoding by using the first coding scheme, when the speech/music signal is a speech signal, and performs encoding using the second coding scheme, when the speech/music signal is a musical signal. The first coding scheme may be a CELP scheme, such as ACELP-based encoding with linear prediction with adaptive code book. The second coding scheme is a coding scheme different from the first coding scheme and does not use linear prediction. For example, the second coding scheme may be a scheme of encoding conversion, such as AAC.

The device 10 coding physically includes a computer device that includes a CPU 10a, a ROM 10b, RAM 10c, the storage device 10d, the device 10E connection and the like. The CPU 10a, a ROM 10b, RAM 10c, the storage device 10d and the device 10E connection connected to the bus 10f. The CPU 10A performs Central management device 10 coding by performing a pre-installed computer programs (for example, audio encoding, designed to perform the process depicted in the block diagram of the sequence of steps of the method figure 3), which is stored in the external the second memory, such as a ROM 10b, and loads it into the RAM 10C. Storage device 10d is memory available for recording and for reading and stores a set of computer programs, many of the data required to execute a computer program (for example, the adaptive code book and the linear prediction coefficients used for encoding according to the first encoding scheme, and, in addition, various parameters required for encoding according to the first encoding scheme and the second coding scheme, and a predetermined number of pre-coded and coded frames). Storage device 10d stores at least a frame of speech/music signal, coded most recently (the last coded frame).

The device 10 encoding functionally includes a module 12 switch coding scheme (first means for determining the coding, second means for determining the encoding), the first encoding module 13 (first encoding means), the second encoding module 14 (the second encoding means), the module 15 of the code multiplexing module 16 calculation of the internal state (the means of calculating the internal state of the encoding) and module 17 specify how the initialization of the internal state (the tool initialization code). The module 12 is pereklucheniya coding scheme, the first encoding module 13, a second encoding module 14 module 15 of the code multiplexing module 16 calculation of the internal state and the module 17 specify how the initialization of the internal state are functions implemented using the CPU 10A executing the computer program stored in the external memory device 10 coding, such as a ROM 10b to control each component of the device 10 coding is depicted in figure 1. The CPU 10A executes the process shown in the flowchart of the sequence of steps of the method of figure 3, with the execution of the program audio encoding (using module 12 switch coding scheme, the first encoding module 13, the second encoding module 14 module 15 of the code multiplexing module 16 calculation of the internal state and module 17 specify how the initialization of the internal state).

Next, referring to figure 3 describes the operation of the device 10 coding. Speech/music signal is first divided into blocks having a finite duration, and then injected into the device 10E communication device 10 coding. When the speech/music signal is injected through the device 10E communication module 12 switch coding scheme determines, based on a target frame coding (frame, which is the purpose of coding) speech/music is a high signal, are you using the first coding scheme or a second coding scheme to encode the target frame coding on the basis of this definition, sends the target frame encoding either the first encoding module 13, which performs the first encoding scheme to encode speech/music signal or the second encoding module 14, which performs the second encoding scheme (step S11; the first phase switching). At step S11 module 12 switch coding scheme determines that the encoding should be performed using the first coding scheme, if the target frame coding is a speech signal, and that the encoding should be performed using the second coding scheme, if the target frame coding is a musical signal. Then, after the first stage switch, perform first initialization phase (steps S12 through S18) to initialize the internal state of the first encoding module 13 (hereinafter in this application referred to as includes the contents of the adaptive code book or the value stored by means of the delay elements of the filter synthesis with linear prediction, which computes the response to a zero input, and so on).

If the module 12 switch coding scheme determines in step S11 that the target frame coding is a musical signal is al, and the target frame encoding must be encoded using the second coding scheme (step S11: the second encryption module), module 12 switch coding scheme sends the target frame encoding in the second encoding module 14, and the second module 14 coding encodes the target frame coding, sent from the module 12 switch the coding scheme using the second coding scheme, and outputs the encoded target frame (coded speech/music signal) through the device 10E of the connection (step S18). If the module 12 switch coding scheme determines in step S11 that the target frame coding is a speech frame and the target frame encoding must be encoded using the first coding scheme (step S11: the first encryption module), the module 12 switch coding scheme accesses the content storage unit 10d determines whether the encoded frame immediately preceding the target frame coding (immediately preceding frame), using the first encoding module 13 or whether it was encoded using the second encoding module 14 (step S12). All coded the results of a predetermined number of coded frames (including the immediately preceding frame and the frames prior to the target frame coding) and yet to be encoded frames stored in the storage device 10d.

If the module 12 switch coding scheme determines at step S12 that the immediately preceding frame has been encoded using the first encoding module 13 (step S12; Yes), the module 12 switch coding scheme sends the target frame encoding in the first encoding module 14, and the first module 13 coding encodes the target frame coding, sent from the module 12 switch the coding scheme by using the first coding scheme, and outputs the encoded result of the target frame coding (encoded speech/music signal) through the device 10E of the connection (step S17). If the module 12 switch coding scheme determines at step S12 that the immediately preceding frame has been encoded using the second encoding module 14 (step S12; no), the module 16 calculate the internal state decodes the encoded result of the immediately preceding frame stored in the storage device 10d, and receives the decoded result of the immediately preceding frame (step S13). The decoded result is used by the device 10 coding gain by using a decoder (not illustrated)included in the device 10 of the encoding or decoding device of 20, described later. This decoding operation may be optional, if used directly preceding frame, still subject to encoding by the second encoding module 14, instead of the decoded result obtained by decoding the encoded financial p is Tata immediately preceding frame. This still encode the immediately preceding frame stored in the storage device 10d.

After step S13, the calculation module 16 internal state calculates the internal state of the first encoding module 13 using the decoded result of the immediately preceding frame (step S14). As illustrative of the process of calculating the internal state using the decoded result of the immediately preceding frame, the process of calculating the internal state of the first encoding module 13, which is performed using the module 16 calculate the internal state includes a process of calculating the coefficients of the linear prediction using a method such as the method of covariance on the basis of the decoded result of the immediately preceding frame (or immediately preceding frame, still subject to encoding by the second encoding module 14), and then obtain the residual signal by applying the inverse filter with linear prediction to the decoded result by using the calculated coefficients of linear prediction.

Since the process of calculating the coefficients of the linear prediction from the decoded result of the immediately preceding frame requires a lot about what Yama calculations instead of calculating the coefficients of the linear prediction from the decoded result of the preceding frame module 16 calculation of the internal state can use the linear prediction coefficients (stored in a storage device 10d) frames adjacent to the immediately preceding frame (previous frame directly preceding frame), which encode using the first coding scheme, instead of the linear prediction coefficients used in the above process (the process of calculating the internal state of the first encoding module 13), or may use the values obtained by interpolation of these coefficients linear prediction between frames, instead of the linear prediction coefficients used in the above process (the process of calculating the internal state of the first encoding module 13). Module 16 calculation of the internal state can use the values obtained by extrapolation of the linear prediction coefficients of the frame, adjacent to the immediately preceding frame, which encode according to the first encoding scheme, or the values obtained by extrapolation of the linear prediction coefficients between frames, instead of the linear prediction coefficients, used in the x in the above process (the process of calculating the internal state of the first encoding module 13). Module 16 calculation of the internal state can convert linear prediction coefficients to line spectral frequencies to extrapolate the linear spectral frequency and re-extrapolated to convert the result back in the coefficients of linear prediction. If the coefficients of linear prediction of the immediately preceding frame is included in the codes of the target frame encoding module 16 calculation of the internal state can use the linear prediction coefficients included in the codes of the target frame encoding, instead of the linear prediction coefficients used in the above process (the process of calculating the internal state of the first encoding module 13). Module 16 calculation of the internal state may use the decoded result of the immediately preceding frame, as he is as a replacement for the residual signal, without calculating the coefficients of the linear prediction. The internal state of the first encoding module 13 can be initialized through the use of internal state information indicating the internal state, remain in the memory device 10d), obtained during the encoding process of the frame adjacent to the immediately preceding frame (and prior to direct the directly previous frame), which encode according to the first encoding scheme. The process of applying the inverse filter with linear prediction to the decoded result of the immediately preceding frame may not be performed relative to the frame, and can be performed only with respect to part of the frame.

After step S14 module 17 specify how the initialization of the internal state specifies, on the basis of the target frame coding or the decoded result of the immediately preceding frame, one of the predefined methods of initialization, including initialization of the internal state of the first encoding module 13, using the internal state calculated by the module 16 calculation of the internal state, the method of initializing the internal state with “0” and the like (step S15). Then the module 17 job initialization method initializes the internal state the internal state of the first encoding module 13 by executing the initialization method specified in step S15 (step S16). Initializing the internal state of the first encoding module 13, which is done using the module 17 specify how the initialization of the internal state is the initialization of the internal state of the first encoding module 13 using the internal with the standing, calculated by the calculation module 16 internal state, and may include the initialization of the internal state (indicating the value stored by means of delay elements) of filter synthesis by linear prediction of the first encoding module 13 for use in the calculation of the residual signal according to the first encoding scheme. When setting method of initializing the internal state of the first encoding module 13 module 17 specify how the initialization of the internal state may, for example, to encode the target frame encoding using the first coding scheme, in accordance with each of the many ways to initialize, including the aforementioned two ways to initialize and select the initialization method that minimizes the quadratic error or perceptual weighted error.

After the module 17 job initialization method initializes the internal state the internal state of the first encryption module at step S16, the first module 13 coding encodes the target frame coding according to the first encoding scheme, and outputs the encoded result of the target frame coding (encoded speech/music signal) through the device 10E of the connection (step S17).

The above process can be configured so that the module 15 m is liplicious code multiplexes information about the initialization method, the selected module 17 specify how the initialization of the internal state at step S15, as additional information, the encoded result obtained according to the first encoding scheme. It can also be configured to specify how to initialize the internal state of the first encoding module 13 on the basis of information (described above), obtained jointly by the first module 13 encoding and the second encoding module 14, and a decoder (decoder included in the device 10 of the encoding or decoding device of 20). In this case, the multiplexing module 15 code no additional multiplexes information indicating the specified initialization method to initialize the internal state of the first encoding module 13, the encoded result. For example, when the gain of the adaptive codebook target frame coding according to the first encoding scheme is large, or when the frequency of the decoded result of the immediately preceding frame is high, or in similar cases, the module 17 specify how the initialization of the internal state can initialize the internal state of the first encoding module 13 using the internal state calculated by the calculation module 16 internal SOS is sustainability.

Alternatively, you can dispense module 17 specify how the initialization of the internal state, if the first module 13 coding always initializes its internal state using the internal state calculated by the module 16 calculation of the internal state. Despite the fact that the module 16 calculation of the internal state and the module 17 specify how the initialization of the internal state is configured to perform the aforementioned process (the first stage of initialization) relative to the target frame coding directly after the module 12 switch the encoding scheme is switched from the second encoding scheme to the first encoding scheme (after the first phase switching), it is not necessarily limited thereby, if the module 16 calculation of the internal state and the module 17 specify how the initialization of the internal state perform the above process, when the immediately previous frame (just before the target frame coding) encoded immediately before module 12 switch coding scheme switches from the second encoding scheme to the first encoding scheme. Although it has been argued that switching is performed between the two encoding schemes, that is, the first encoding scheme (the first module is 13 encoding and the second encoding scheme (the second module 14 encoding) switching can be performed between three or more encoding schemes, including many coding schemes, different from the first coding scheme.

1 and 4 is a diagram depicting the configuration of the decoding device 20, in accordance with one embodiment. The decoding device 20 is physically includes a computer device comprising a CPU 20a, a ROM 20b, RAM 20c, the storage device 20d, the device 20E connection and the like. The CPU 20a, a ROM 20b, RAM 20c, the storage device 20d and the device link 20E is connected to the bus 20f. The CPU 20A performs Central control device 20 decoding by performing a pre-installed computer programs (for example, decoding of an audio signal for performing the process depicted in the block diagram of the sequence of steps of the method of figure 5), which is stored in an external memory such as a ROM 20b, and loads it into the RAM 20C. Storage device 20d is a memory that is writable and readable, and stores a set of computer programs, many of the data required to execute a computer program (including, for example, the adaptive code book and the linear prediction coefficients used in the decoding according to the first encoding scheme, and, in addition, various parameters required to perform the Dec is tiravanija according to the first encoding scheme and the second coding scheme, and the prescribed number of decoded frames and frames before decoding, etc). Storage device 20d stores, at least, speech/music signal, the decoded most recently (the last decoded frame).

The device 20 decoding functionally includes the module 22 to determine the encoding scheme (first means for determining the decoding, the second means for determining the decoding), the module 23 division code, the first module 24 decoding (first decoding means), the second decoding module 25 (second decoding means), and the module 26 specify how the initialization of the internal state (the tool initialization decoding) and module 27 calculation of the internal state (the means of calculating the internal state of the decoding). Module 22 schema definition decoding module 23 division code, the first module 24 decoding, the second decoding module 25 module 26 specify how the initialization of the internal state and the module 27 calculation of the internal state are functions implemented using the CPU 20A executes computer programs stored in the external memory device 20 decoding, such as a ROM 20b to control each component of the device 20 coding is depicted in figure 1. The CPU 20A executes the process image the config on the block diagram of the sequence of steps of the method of figure 5, using program execution decoding audio (using module 22 schema definition encoding module 23 division code of the first module 24 decoding, the second decoding module 25 module 26 specify how the initialization of the internal state and module 27 calculation of the internal state).

Next, referring to figure 5, describes the operation of the device 20 encoding. Module 22 schema definition decoding determines whether the first coding scheme or a second coding scheme to encode the target frame decoding encoded speech/music signal entered through the device 20E of communication, and on the basis of the determination, sends the target frame decoding either the first module 24 of the decoding, to apply decoding according to the first decoding scheme or the second decoding module 25, for use decoding by the second decoding scheme (step S21; the second stage switch). At step S21 module 22 schema definition decoding determines that the decoding should be performed using the first module 24 decoding, if the target frame decoding has been encoded according to the first encoding scheme, and that the decoding should be performed using the second decoding module 25, if the target frame is decoded what I have been encoded by the second encoding scheme. Then, after the second stage switch, perform the second stage of initialization (steps S22 through S27), which initializes the internal state of the first module 24 decoding (hereinafter in this application referred to as includes the contents of the adaptive code book or the value stored by means of the delay elements of the filter synthesis with linear prediction, or the like).

If the module 22 schema definition decoding determines in step S21 that the target frame decoding has been encoded by the second encoding scheme, that is, the target frame decoding must be decoded by the second decoding module 25 (step S21: the SECOND DECODING MODULE), the module 22 schema definition decoding sends the target frame is decoded in the second decoding module 25, and a second decoding module 25 decodes the target frame decoding, sent from the module 22 to determine the coding scheme according to the second coding scheme, and outputs the decoded result of the decoding target frame to be decoded (decoded speech/music frame) through the device 20E communication (step S27). If the module 22 to determine the encoding scheme determines in step S21 that the target frame decoding has been encoded according to the first encoding scheme, that is, the target frame decoding up to the wives to be decoded using the first module 24 decoding (step S21: the first encryption module), module 22 schema definition decode accesses the content storage device 20d determines whether the encoded frame, immediately before the target frame decoding (immediately preceding frame), according to the first coding scheme (i.e. the immediately preceding frame has been decoded using the first module 24 decoding), or whether it was coded by the second coding scheme (i.e. the immediately preceding frame has been decoded using the second device 25 decoding (step S22). All the decoded results of a predetermined number of decoded frames (including the immediately preceding frame and the frames prior to the target frame decoding) and frames that are still subject to decoding, maintain in the storage device 20d.

If the module 22 to determine the encoding scheme determines in step S22 that the immediately preceding frame has been encoded using the first coding scheme (i.e., the immediately previous frame was encoded using the first module 24 coding) (step S22; Yes), the module 22 to determine the encoding scheme sends the target frame decoding in the first module 24 decoding, and the first module 24 decoding decodes the target frame decoding, sent from the module 22 definition and therefore the s encoding, according to the first encoding scheme, and outputs the decoded result to the target frame to be decoded (decoded speech/music signal) through the device 20E of the connection (step S26).

If the module 22 to determine the encoding scheme determines in step S22 that the immediately preceding frame has been encoded according to the second coding scheme (i.e., the immediately preceding frame has been decoded by the second decoding module 25) (step S22; no), the module 22 to determine the encoding scheme sends the immediately preceding frame in the module 23 separation of code and module 23 division code divides the multiplexed codes immediately preceding frame for codes first coding scheme and additional information indicating the method of initializing the internal state of the first module 24 decoding (for example, information indicating a method of initializing the internal state of the first encoding module 13 that set module 17 specify how the initialization of the internal state and is used when encoding the immediately preceding frame). Then the module 27 calculate the internal state calculates the internal state of the first module 24 decoded using the decoded result of the immediately preceding frame (step S23). As ill the administrative process of calculating the internal state of the decoded result of the immediately preceding frame, the process of calculating the internal state of the first module 24 decoding, which is performed using the module 27 calculate the internal state includes a process of calculating the coefficients of the linear prediction using a method such as the method of covariance on the basis of the decoded result of the immediately preceding frame, and then calculate the residual signal by applying the inverse linear prediction filter to the decoded result by using the calculated coefficients of linear prediction.

Since the process of calculating the coefficients of the linear prediction from the decoded result of the immediately preceding frame requires a large amount of computation, instead of calculating the coefficients of the linear prediction from the decoded result of the immediately preceding frame module 27 calculation of the internal state can use the linear prediction coefficients (which are the coefficients of the linear prediction used at the time of decoding by using the first module 24 decoding, and stored in the storage device 20d) of the frame adjacent to the immediately preceding frame (and prior to the immediate previous frame), which is encoded according to the PE the new encoding scheme, instead of the linear prediction coefficients used in the above process (the process of calculating the internal state of the first module 24 decoding), or may use the values obtained by interpolation of these coefficients linear prediction between frames, instead of the linear prediction coefficients used in the above process (the process of calculating the internal state of the first module 24 decoding). Module 27 calculation of the internal state can use the values obtained by extrapolation of the linear prediction coefficients of the frame, adjacent to the immediately preceding frame, which is encoded according to the first encoding scheme, or the values obtained by extrapolation of the linear prediction coefficients between frames, instead of the linear prediction coefficients used in the above process (the process of calculating the internal state of the first module 24 decoding). Module 27 calculation of the internal state can convert linear prediction coefficients to line spectral frequencies to extrapolate the linear spectral frequency and re-extrapolated to convert the result back in the coefficients of linear prediction. If the coefficients of linear p is escasany immediately preceding frame included in the codes of the target frame to be decoded, module 27 calculation of the internal state can use the linear prediction coefficients included in the codes of the target frame decoding, instead of the linear prediction coefficients used in the above process (the process of calculating the internal state of the first module 24 decoding). Alternatively, you can do without calculating the coefficients of the linear prediction using the passes of applying the inverse filter with linear prediction. In addition, the internal state of the first module 24 decoding can be initialized through the use of internal state information indicating the internal state, remain in the memory device 10d), obtained during the decoding process of the frame adjacent to the immediately preceding frame (and prior to the immediately preceding frame), which encode according to the first encoding scheme. The process of applying the inverse filter with linear prediction to the decoded result of the immediately preceding frame may not be performed relative to the frame, and can be performed only with respect to part of the frame.

After step S23 module 26 specify how the initialization of the internal state specifies, on the basis of the additional information included multiplexion codes immediately preceding frame, and specify how the initialization of the internal state of the first module 24 decoding, one of the predefined methods of initialization, including initialization of the internal state of the first module 24 decoding, using the internal state calculated by the module 27 calculate the internal state initialization method using “0” and the like (step S24). Then the module 26 job initialization method initializes the internal state the internal state of the first module 24 decoding, in accordance with the initialization method specified in step S24 (step S25). Initializing the internal state of the first module 24 decoding, which is done using the module 26 specify how the initialization of the internal state is the initialization of the internal state of the first module 24 decoding using the internal state calculated by the module 27 calculation of the internal state, and may include the initialization of the internal state (the values stored by using delay elements) of filter synthesis by linear prediction of the first module 24 decoding, which calculates the output signal from the residual signal according to the first encoding scheme.

After the module 26 specify how ISI is alizatio internal state initializes the internal state of the first module 24 of the decoding at step S25, the first module 24 decoding decodes the target frame decoding, in accordance with the first encoding scheme, and outputs the decoded result to the target frame to be decoded (decoded speech/music signal) through the device 20E of the connection (step S26).

If additional information indicating how the initialization, the initialization of the internal state of the first module 24 decoding, not multiplexed in codes immediately preceding frame, the initialization method for initializing the internal state of the first module 24 decoding, can be specified using a fixed gain codebook target frame coding according to the first encoding scheme, or the analysis of periodic decoded result of the immediately previous frame, or the like (using information obtained jointly from the first module 24 decoding and the second decoding module 25, and the encoder (encoder included in the device 20 of the decoding or the first module 13 coding). It can also be configured to dispense module 26 specify how the initialization of the internal state, if the first module 24 decoding always initializes its internal state using the internal state, wycis the military module 27 calculation of the internal state. In this case there is no need to use additional information that indicates the initialization method, which is multiplexed in codes immediately preceding frame. Despite the fact that the work of the module 27 calculation of the internal state and the operation module 26 specify how the initialization of the internal state described above in connection with the case where the immediately preceding frame has been encoded according to the first encoding scheme, it is not limited to this. If it is determined in advance that the target frame decoding has been encoded by the second encoding scheme, and the frame that immediately follows the target frame decoding has been encoded according to the first encoding scheme, the module 27 calculation of the internal state and the module 26 specify how the initialization of the internal state can calculate the internal state of the first module 24 decoding and selecting a method of initializing the internal state on the basis of advance information. Although discussed a configuration in which switching is performed between the two encoding schemes, that is, the first encoding scheme and the second coding scheme may be configured so that switching is performed between three or more encoding schemes, including many coding schemes, different from the first schema code for the simulation.

Next will be described the work and the entity device 10 coding, in accordance with the embodiment. The device 10 coding includes the first module 13 coding, working according to the encoding scheme with linear prediction, and the second module 14 coding, working according to a different encoding scheme that is different from the encoding scheme with linear prediction, and encodes the audio signal using the first module 13 encoding and the second encoding module 14. The device 10 coding additionally includes a module 12 switch the coding scheme module 16 calculation of the internal state and the module 17 specify how the initialization of the internal state. Module 12 switch coding scheme determines whether to use the first encoding module 13 or the second module 14 of the encoder to encode the target frame coding, i.e. the target frame to be coded is included in the audio signal. If it is determined that the target frame encoding must be encoded using the first encoding module 13 module 12 switch coding scheme determines whether the encoded frame immediately preceding the target frame coding using the first encoding module 13 or the second encoding module 14. If module 12 switch the coding scheme is of the defined what immediately preceding frame has been encoded using the second encoding module 14, the module 16 calculate the internal state decodes the encoded result of the immediately preceding frame, and calculates the internal state of the first encoding module 13 using the decoded result. Module 17 job initialization initializes the internal state the internal state of the first encoding module 13 using the internal state calculated by the module 16 calculation of the internal state. Then the first module 13 coding encodes the target frame coding, after the internal state is initialized by the module 17 specify how the initialization of the internal state.

In the encoding device, even when the target frame encoding must be encoded using the first encoding module 13 according to the encoding scheme with linear prediction, while the immediately preceding frame has been encoded using the second encoding module 14 according to the encoding scheme that is different from the encoding scheme with linear prediction, the target frame encoding may be encoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first encoding module 13. AFL is stvie this can be implemented processing coding made according to the number of encoding schemes, including coding scheme with linear prediction and the other encoding scheme that is different from the encoding scheme with linear prediction.

Next will be described the operation and the result of the decoding device 20, in accordance with the embodiment. The device 20 decoding includes the first module 24 decoding, working according to the encoding scheme with linear prediction, and the second decoding module 25 operating according to a different encoding scheme that is different from the encoding scheme with linear prediction, and decodes the encoded audio signal using a first module 24 decoding and the second decoding module 25. The device 20 decodes additionally includes the module 22 to determine the coding scheme module 27 calculation of the internal state and the module 26 specify how the initialization of the internal state. The module 22 to determine the encoding scheme determines whether to use the first module 24 of the decoding or the second module 25 decoding to decode the target frame decoding, i.e. the target frame to be decoded included in the audio signal. If using the module 22 to determine the encoding scheme is determined that the target frame decoding must be decoded using first the th module 24 of the decoding, the module 22 to determine the encoding scheme determines whether the decoded frame immediately preceding the target frame decoding, using the first module 24 of the decoding or the second decoding module 25. If using the module 22 to determine the encoding scheme is determined that the immediately preceding frame has been encoded using a second decoding module 25, the internal state of the first module 24 decoding is calculated using the decoded result of the immediately preceding frame. The internal state of the first module 24 decoding initialize using the internal state calculated by the module 27 calculation of the internal state. Then the first module 24 decoding decodes the target frame decoding, once initialized internal state, in accordance with module 27 specify how the initialization of the internal state.

In the device 20 of the decoding of the present invention, even when the target frame decoding must be decoded using the first module 24 decoding according to the encoding scheme with linear prediction, while the immediately preceding frame has been decoded by the second decoding module 25 according to the decoding scheme other than the scheme to which tiravanija with linear prediction, the target frame decoding can be decoded according to the encoding scheme with a linear prediction using the initialization of the internal state of the first module 24 decoding. This can be implemented in the decoding processing performed according to the number of encoding schemes, including coding scheme with linear prediction and the other encoding scheme that is different from the encoding scheme with linear prediction.

Industrial applicability

When switching from an encoding scheme that does not use linear prediction, coding scheme, using coding with linear prediction, the internal state of the means of coding or the decoding means operating according to the encoding scheme that uses coding with linear prediction, is installed in a suitable initial value, thus, can be improved speech quality, reproducible from shot coming directly after switching.

A list of reference designations

10 - device encoding

10A, 20b - CPU

10b, 20b - ROM

10C, 20C - RAM

10d, 20d - storage device

10e, 20e - device communication

10f, 20f - bus

12 - switch module coding scheme

13 - the first encryption module

14 - second encryption module

15 module code multiplexing

6, 27 module calculating the internal state

17, 26 module specify how the initialization of the internal state

20 - decoder

22 - determining module coding scheme

23 - separation module code

24 - the first module of the decoding

25 - the second decoding module

1. The method of audio encoding for encoding an audio signal includes multiple frames, using the first encoding means working according to the encoding scheme with linear prediction, and second encoding means working according to the encoding scheme that is different from the encoding scheme with linear prediction, and a method of coding an audio signal contains
the phase switch, which switch the encoding means for encoding the second frame immediately following the first frame from the second encoding means on the first means of coding, after the first frame of an audio signal encoded using a second encoding means, and
the stage of initialization, which initializes the internal state of the first encoding means, in accordance with a predefined method, after step switch,
moreover, during initialization of the internal state of the first encoding means initialize by us is anouki residual signal as the content of the adaptive codebook of the first encoding means, moreover, the residual signal is obtained by applying the inverse filter with linear prediction or to the first frame, still subject to encoding by the second encoding means, or a signal obtained by decoding the encoded first frame generated by the second encoding means, and
moreover, the linear prediction coefficients of the first frame is included in the codes of the second frame, and in the initialization phase, the coefficients of linear prediction is used to inverse filter with linear prediction.

2. The method of decoding an audio signal for decoding the encoded audio signal includes a set of encoded frames is generated from the audio signal includes multiple frames, using the first decoding means operating according to the encoding scheme with linear prediction, and second decoding means operating according to the encoding scheme that is different from the encoding scheme with linear prediction, and the method of decoding an audio signal includes
the phase switch, which switches the decoding means for decoding the second encoded frame immediately following the first coded frame from the second decoding means to the first decoding means, after the first encoded frame of the encoded audio signal decoded by the second decoding means, and the first encoded frame is formed from the first frame of the audio signal and the second encoded frame is formed from a second frame immediately following the first frame in the audio signal, and
the stage of initialization, which initializes the internal state of the first decoding means in accordance with a predefined method, after step switch,
moreover, during initialization of the internal state of the first decoding means initialize by setting the residual signal as the content of the adaptive codebook of the first decoding means, and the residual signal is obtained by applying the inverse filter with linear prediction to the signal obtained by decoding the first encoded frame using the second decoding means, and
moreover, the linear prediction coefficients of the first frame is included in the codes of the second encoded frame, and in the initialization phase, the coefficients of linear prediction is used to inverse filter with linear prediction.

3. The encoding device includes first encoding means working according to the encoding scheme with linear prediction, and second encoding means working according to the encoding scheme that is different from the encoding scheme with the line is the prediction and encodes the audio signal using the first encoding means and the second encoding means, and encoding device contains
the first tool definition that determines whether the first or the second encoding means to encode the target frame, which is included in the audio signal and is used as the target encoding,
the second definition, which determines if the first means for determining determines that the target frame must be encoded using the first encoding means, was coded immediately preceding frame which immediately precedes the target frame by using the first encoding means or the second means of encoding,
the means of calculating the internal state, which evaluates, if the second means for determining determines that the immediately preceding frame has been encoded using a second encoding means, the internal state of the first encoding means by applying the inverse filter with linear prediction to the signal obtained by decoding the encoded result of the immediately preceding frame to calculate the residual signal, and
tool initialization, which initializes the internal state of the first encoding means by setting the residual is signal calculated using the calculation tools internal state, as the content of the adaptive codebook of the first means of encoding,
moreover, the coefficients of the linear prediction immediately preceding frame codes included in the target frame, and a means of calculating the internal state uses the linear prediction coefficients for the inverse filter with linear prediction, and
first encoding means encodes the target frame after initialization of the internal state by using the initialization.

4. The decoding device includes first decoding means operating according to the encoding scheme with linear prediction, and the second decoding means operating according to the encoding scheme that is different from the encoding scheme with linear prediction, and decoding the encoded audio signal that is generated from the audio signal, using the first decoding means and the second decoding means, and the decoding device includes
the first tool definition that determines whether the first or the second decoding means to decode the target encoded frame, which is included in the encoded audio signal and serves as a target of decoding,
the second tool defined who I am, which determines if the first means for determining determines that the target encoded frame should be decoded by the first decoding means, whether the decoded immediately preceding encoded frame that immediately precedes the target encoded frame using the first decoding means or the second means for decrypting,
the means of calculating the internal state, which evaluates, if the second means for determining determines that the immediately preceding encoded frame has been decoded by the second decoding means, the internal state of the first decoding means by applying the inverse filter with linear prediction to the signal obtained by decoding the immediately preceding encoded frame to calculate the residual signal, and
tool initialization, which initializes the internal state of the first decoding means by setting a residual signal calculated by the calculation tools internal state, as the content of the adaptive codebook of the first decoding means,
moreover, the immediately preceding encoded frame and the target encoded frame are formed from the first audio frame and the second frame, neposredno is but following the first frame in the audio signal, accordingly,
the linear prediction coefficients of the first frame is included in the codes of the target encoded frame,
the means of calculating the internal state uses the linear prediction coefficients for the inverse filter with linear prediction, and
first decoding means decodes the target encoded frame after initialization of the internal state by using the initialization.

5. System for processing an audio signal that contains
the encoding device according to claim 3 and
the decoding device according to claim 4,
moreover, the decoding device decodes the encoded audio signal encoded by the encoder.

6. A storage device containing the stored program audio encoding for encoding the audio signal using a first encoding means working according to the encoding scheme with linear prediction, and second encoding means working according to the encoding scheme that is different from the encoding scheme with linear prediction, and program for encoding an audio signal induces a computer device to function as a
the first means of encoding,
the second tool, encoding,
first means for determining for determining whether the first or the second tool Kodirov the tion, to encode the target frame, which is included in the audio signal and is used as the target encoding,
second means for determining for determining if the first means for determining determines that the target frame must be encoded using the first encoding means, was coded immediately preceding frame which immediately precedes the target frame by using the first encoding means or the second means of encoding,
the means of calculating the internal state, which evaluates, if the second means for determining determines that the immediately preceding frame has been encoded using a second encoding means, the internal state of the first encoding means by applying the inverse filter with linear prediction to the signal obtained by decoding the encoded result of the immediately preceding frame to calculate the residual signal, and
tool initialization, which initializes the internal state of the first encoding means by setting a residual signal calculated by the calculation tools internal state, as the content of the adaptive codebook of the first means of encoding,
moreover, the coefficients of linear prediction of the immediately preceding frame is included in to the s of the target frame, and the means of calculating the internal state uses the linear prediction coefficients for the inverse filter with linear prediction.

7. A memory device containing stored therein a program for decoding an audio signal for decoding the encoded audio signal that is generated from the audio signal, using the first decoding means operating according to the encoding scheme with linear prediction, and second decoding means operating according to the encoding scheme that is different from the encoding scheme with linear prediction, and program for decoding audio makes a computer device to function as a
first decoding means,
second decoding means,
first means for determining for determining whether the first or the second decoding means to decode the target encoded frame, which is included in the encoded audio signal and serves as a target of decoding,
second means for determining for determining if the first means for determining determines that the target encoded frame should be decoded by the first decoding means, whether the decoded immediately preceding encoded frame that immediately precedes Clevo is the encoded frame, using the first decoding means or the second means for decrypting,
the means of calculating the internal state, which evaluates, if the second means for determining determines that the immediately preceding encoded frame has been decoded by the second decoding means, the internal state of the first decoding means by applying the inverse filter with linear prediction to the signal obtained by decoding the immediately preceding encoded frame to calculate the residual signal, and
tool initialization, which initializes the internal state of the first decoding means by setting a residual signal calculated by the calculation tools internal state, as the content of the adaptive codebook of the first decoding means,
moreover, the immediately preceding encoded frame and the target encoded frame are formed from the first audio frame and the second frame immediately following the first frame in the audio signal, respectively,
the linear prediction coefficients of the first frame is included in the codes of the target encoded frame, and
the means of calculating the internal state uses the linear prediction coefficients for the inverse filter with linear forecast is to be neglected.