JP4888048B2 - Audio signal encoding / decoding method, apparatus and program for implementing the method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide audio signal encoding and decoding technology, in which encoding efficiency is high and encoding and decoding processing delay is small. <P>SOLUTION: A preceding frame is searched which is similar to a frame to be encoded of an audio signal which is converted to a frequency domain. When the similar frame is searched, information which indicates the similar frame, and information which indicates difference of a frequency component of the frame to be encoded and the similar frame, are encoded. When the similar frame is not searched, the frequency component of the framed to be encoded is encoded. <P>COPYRIGHT: (C)2008,JPO&amp;INPIT

Description

  The present invention relates to an audio signal encoding / decoding technique, and more particularly to an encoding / decoding technique using the periodicity of an audio signal.

  As digital music players such as iPod become popular, it is desired to reduce the file size per song and increase the number of songs that can be stored in the memory installed in the player without deteriorating the sound quality of hearing. . With this technology, it is desirable to reduce the amount of code per content (music) without degrading the sound quality of the music.

  As a technique that meets this requirement, a technique described in JP-A-2005-292640 (Patent Document 1) is known.

  In the technique described in Patent Document 1, the encoding side divides the audio signal into frames of about 2 milliseconds, for example, and encodes the audio signal in units of frames. As preprocessing of this encoding, the encoding side classifies all frames into independent frames and prediction frames. Here, the independent frame is a frame in which the audio signal in the frame is encoded without referring to the audio signal of another frame. A predicted frame is a frame in which an audio signal in the frame is encoded using predictive encoding based on the audio signal of an independent frame. When this classification is completed, the encoding side first encodes the independent frame, and then performs predictive encoding of the prediction frame.

  On the decoding side of Patent Document 1, first, the independent frame is decoded, and then the prediction error signal of the prediction frame is predictively decoded using the reproduced audio signal of the independent frame.

  With the technique of this patent document 1, it is possible to predictively encode an audio signal of a prediction frame using not only an audio signal in a past frame in time but also an audio signal in a future independent claim. Therefore, the technique of this patent document 1 provides high encoding efficiency.

JP-A-2005-292640

  However, the technique described in Patent Document 1 has a problem that the processing delay of the encoding process and the processing delay of the decoding process are large. The reason for this is as follows.

  In this prior art, the encoding apparatus first classifies all frames into independent frames and prediction frames, then encodes independent frames, and finally predictively encodes prediction frames. For this reason, the encoding side needs to rearrange the frames. These classification and frame rearrangement increase the processing delay of the encoding process.

  In this prior art, the decoding device first decodes an independent frame, then predictively decodes a predicted frame, and rearranges the decoded independent frame and the decoded predicted frame to perform audio decoding. Play the signal. For this reason, the decoding side must also rearrange the frames. This frame rearrangement increases the processing delay of the decoding process.

  An object of the present invention is to provide an audio signal encoding / decoding technique having high encoding efficiency and low encoding processing delay and decoding processing delay.

In one aspect of the present invention, the encoding side normalizes an audio signal converted to the frequency domain based on a peak value of a frame of the audio signal converted to the frequency domain, and the normalized one piece of music The preceding frame similar to the encoding target frame is searched from the beginning of the audio signal to the immediately preceding encoding target frame. When a similar frame is searched, the encoding side includes information indicating that the encoding target frame having the frame number is determined to be a similar frame, and frequency component difference data between the encoding target frame and the similar frame. Are encoded. When the similar frame is not searched, the encoding side encodes the frequency component of the encoding target frame.

Further, the decoding side decodes the encoded frequency component information for a frame to which similar frame information is not added, and performs inverse frequency conversion on the decoded frequency component. Also, the decoding side, for the frame similar frame information is added decodes the frequency component difference data coded, the decoded frequency components by adding the frequency components of similar frame number, the similarity The frequency component information of the frame to which the frame information is added is restored, and the restored frequency component is subjected to inverse frequency conversion.

  As described above, according to one aspect of the present invention, the previous frame similar to the encoding target frame is searched, and the frequency component difference data between the encoding target frame and the similar frame is encoded. A difference signal between similar frames is smaller than a normal signal. For this reason, this one side surface can suppress the whole code amount, without degrading the sound quality on hearing.

  In addition, according to one aspect of the present invention, since similar frames are searched only from frames preceding the encoding target frame, it is not necessary to rearrange audio data in units of frames. This reduces not only the encoding process delay but also the decoding process delay.

  Next, an encoding apparatus according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is a block diagram illustrating a configuration of the encoding apparatus 100.

  As shown in FIG. 1, the encoding apparatus 100 includes an input buffer 101, a frequency converter 102, a normalization unit 103, a differential signal extraction processing unit 104, a differential signal threshold processing unit 105, a quantization / The encoding processing unit 106 and an output memory 107 are included.

  In the following description, it is assumed that an audio signal (PCM signal) for one song is stored in the input buffer 102 at the start of encoding.

  The frequency converter 102 reads an audio signal from the input buffer 201 in a predetermined time unit, for example, 2 milliseconds, and converts the frequency of the audio signal. This predetermined unit is a frame of the audio signal. A frame number is added to the frequency-converted data, and the frequency-converted audio data is generated for each frame and supplied to the normalization unit 103. As this frequency transform, Fourier transform, discrete cosine transform (DCT), modified discrete cosine transform (MDCT) or the like can be used. Needless to say, when the frequency-converted audio signal, that is, the frequency component data is stored in the input buffer, the frequency converter 102 becomes unnecessary.

  The normalization unit 103 normalizes the supplied frequency component data in units of frames. This normalization is performed according to the level of the peak frequency component in the frame. The normalization coefficient used for this normalization is supplied to the quantization / encoding processing unit 106. The normalized frequency conversion frequency component data X (k) (k = 0, 1,..., K−1) is supplied to the differential signal extraction processing unit 104.

  This normalization process contributes to particularly improving the prediction efficiency when, for example, the audio signals are of the same scale and the same tone color and differ only in amplitude between different frames.

  Needless to say, when the normalized frequency component data is already stored in the input buffer 101, the frequency converter 102 and the normalizing unit 103 are not necessary.

  The difference signal extraction processing unit 104 searches for frequency component data similar to the normalized frequency component data of the current encoding target frame from the normalized frequency component data in the frame before the previous frame.

  This search can be performed, for example, by calculating the following equation (1) for an appropriate number of preceding frames.

Here, X _diff indicates the degree of difference, and the smaller the value, the higher the degree of similarity. X (k) represents the kth normalized frequency component data. X _reference (k) indicates the k-th normalized frequency component data of the preceding frame. N indicates the number of frequency components per frame.

  The difference signal extraction processing unit 104 determines that there is no similar frame when a preceding frame in which the calculation result of the expression (1) is equal to or less than the threshold (T) cannot be searched. In addition, when the difference signal extraction processing unit 104 can detect at least one preceding frame in which the calculation result of equation (1) is T or less, one of the detected preceding frames is determined as a similar frame. select.

  When the difference signal extraction processing unit 104 cannot search for a preceding frame similar to the current encoding target frame, the differential signal extraction processing unit 104 determines the encoding target frame as a general frame (G frame). Then, the differential signal extraction processing unit 104 supplies a flag indicating that the encoding target frame is a G frame and the normalized frequency component data of the encoding target frame to the difference threshold processing unit 105.

  When the differential signal extraction processing unit 104 can search for a preceding frame similar to the current encoding target frame, the differential signal extraction processing unit 104 determines the encoding target frame as a differential frame (D frame). Then, the difference signal extraction processing unit 104 supplies a flag indicating that the encoding target frame is a D frame to the difference threshold processing unit 105. Further, the difference signal extraction processing unit 104 supplies the frame number of the preceding frame similar to the encoding target frame, that is, the similar frame number, to the difference threshold processing unit 105. Further, the difference signal extraction processing unit 104 converts the frequency component difference data, which is a difference signal between the normalized frequency component data of the encoding target frame and the normalized frequency component data of the similar frame number, to the difference threshold value. Supply to the processing unit 105. Although the number of preceding frames similar to the encoding target frame is not necessarily one, the similar frame number supplied to the difference threshold processing unit 105 is preferably the frame number of the preceding frame most similar to the encoding target frame.

  The flag indicating the type of G frame or D frame is also supplied to the quantization / encoding processing unit 106 via the difference threshold processing unit 105.

  When the supplied data is a G frame, the difference threshold processing unit 105 transfers the supplied frequency component data and a flag indicating the G frame to the quantization / encoding processing unit 106 as they are.

  When the supplied data is a D frame, the difference threshold processing unit 105 compares the supplied frequency component difference data with a difference threshold (DT). Then, the frequency component difference data having a value smaller than DT is replaced with the value zero. The frequency component difference data that has been subjected to the difference threshold processing is supplied to the quantization / encoding processing unit 106. Further, the difference threshold processing unit 105 supplies the flag (indicating D frame) and the similar frame number supplied from the difference signal extraction processing unit to the quantization / encoding processing unit 106 as they are. In the present embodiment, it goes without saying that the difference threshold processing unit 105 is not necessary when the difference threshold processing is not performed.

  The quantization / encoding processing unit 106 encodes the frequency component data or frequency component difference data supplied from the difference threshold processing unit 105, and generates encoded frequency component data or encoded frequency component difference data. To do.

  When this frame corresponds to the G frame, the quantization / encoding processing unit 106 encodes the frequency component data supplied from the difference threshold processing unit 105 and generates encoded frequency component data. . Then, the quantization / encoding processing unit 106 creates the G frame illustrated in FIG. 2 from the frame number, the flag, the normalization coefficient, and the m-encoded frequency component data, and stores it in the output memory 107. Store.

  In FIG. 2, “frame number” indicates the frame number of this frame, “FT” indicates the type of frame (in this case, G frame), and “normalization coefficient” is supplied from the normalization unit 103. A normalization coefficient is indicated, and “frequency component data” indicates encoded frequency component data.

  Further, when this frame corresponds to the D frame, the quantization / encoding processing unit 106 encodes the frequency component difference data supplied from the difference threshold processing unit 105 and encodes the frequency component difference thus encoded. Generate data. Then, the quantization / encoding processing unit 106 creates the D frame illustrated in FIG. 3 from the frame number, the flag, the normalization coefficient, the similar frame number, and the encoded frequency component data. And stored in the output memory 107.

  In FIG. 3, “frame number” indicates the frame number of this frame, “FT” indicates the type of frame (in this case, D frame), and “similar frame number” is selected by the differential signal extraction processing unit. The “normalization coefficient” indicates the normalization coefficient supplied from the normalization unit 103, and the “frequency component difference data” indicates the encoded frequency component difference data.

  As described above, the encoding apparatus in FIG. 1 searches for a preceding frame similar to the encoding target frame, and encodes frequency component difference data between the encoding target frame and the similar frame. The difference signal between similar frames is smaller than the frequency component. For this reason, the encoding apparatus of FIG. 1 can suppress the entire code amount without degrading the audible sound quality.

  In addition, when the frequency component difference data of a certain frequency component is equal to or smaller than the difference threshold (DT), the encoding device in FIG. Can be suppressed. If this threshold is set to zero and all the difference components are encoded, lossless encoding can be realized. That is, lossy coding can be seamlessly handled only by changing the difference threshold (DT) from lossless.

  Further, since the encoding apparatus in FIG. 1 searches for similar frames only from frames preceding the encoding target frame, it is not necessary to rearrange audio data in units of frames. This reduces the encoding delay of the encoding device.

  Next, a decoding device according to an embodiment of the present invention will be described with reference to FIG. FIG. 4 is a block diagram illustrating a configuration of the decoding apparatus 200.

  As illustrated in FIG. 4, the decoding device 200 includes an input buffer 201, a decoding processing unit 202, an addition processing unit 203, a denormalization unit 204, a frequency inverse transformer 205, and an output memory 206. Consists of.

  At the start of decoding, it is assumed that the input buffer 201 stores a predetermined number of G frames and D frames generated by the encoding device of FIG.

  The decoding processing unit 202 reads G frames or D frames from the input buffer frame by frame in the order of frame numbers. First, the decoding processing unit 202 separates frame type information (“TYPE” in FIGS. 2 and 3) and a normalization coefficient from the read frame. This frame type information is supplied to the addition processing unit 204. The normalization coefficient is supplied to the denormalization unit 204. In addition, when the frame is a D frame, the decoding processing unit 202 supplies the similar frame number to the addition processing unit 204. Further, the decoding processing unit 202 decodes the encoded frequency component data or the encoded frequency component difference data, and supplies the decoding result to the addition processing unit 204.

  When the frame type information indicates a G frame, the addition processing unit 203 supplies the frequency component data supplied from the decoding processing unit 203 to the inverse normalization unit 204 as it is. The addition processing unit 203 stores the frequency component data together with the frame number of the frame.

  In addition, when the frame type information indicates a D frame, the addition processing unit 203 adds the frequency component difference data of the decoding target frame supplied from the decoding processing unit 203 to the frequency data of the similar frame. The frequency component data of the decoded reference frame is reproduced. The addition processing unit 203 supplies the reproduced frequency component data to the inverse normalization unit 204. The addition processing unit 203 stores the frequency component data together with the frame number of the frame.

  The inverse normalization unit 204 multiplies the frequency component data supplied from the addition processing unit by the normalization coefficient supplied from the decoding processing unit 202. The multiplication result is supplied to the frequency inverse converter 205.

  The frequency inverse transformer 205 inversely transforms the denormalized frequency component data supplied from the denormalization unit, reproduces the time domain audio signal, and stores it in the output memory 208. The time-domain audio signal stored in the output memory 206 is read in response to a request from an audio playback device or the like.

  In this decoding apparatus, it is obvious that the denormalization unit 204 may be provided between the decoding processing unit 202 and the addition processing unit 203.

  As described above, since the decoding apparatus in FIG. 4 performs the decoding process in the order of the frame numbers, the frame rearrangement process is not required in the decoding apparatus, so that the decoding delay is small. Therefore, the decoding apparatus according to the present embodiment solves the problem that cannot be avoided in Patent Document 1 in cooperation with the encoding apparatus of FIG.

  Next, a program for realizing the encoding apparatus of FIG. 1 and the decoding apparatus of FIG. 4 by a computer will be described with reference to the flowcharts of FIGS. FIG. 5 is a flowchart for explaining a program for realizing the encoding apparatus of FIG. 1 by a computer. FIG. 6 is a flowchart for explaining a program for realizing the decoding apparatus of FIG. 4 by a computer.

  First, FIG. 5 will be described. In FIG. 5, it is assumed that a predetermined number of frames of audio data are stored in the input buffer of FIG. 1 before the steps after step S11 are executed. The central processing unit (CPU) of the computer executes processing equivalent to the processing from the frequency converter 102 to the quantization / encoding processing unit 106 in FIG. 1 in steps S11 to S19 in FIG.

  In step S11, the CPU reads audio data for one frame from the input buffer of FIG. At this time, the CPU gives a frame number to the read data.

  Next, in step S12, the CPU converts the audio data of one frame into the frequency domain. In step S13, the CPU normalizes this frequency domain data by the method as described above, and generates frequency component data. The CPU stores the normalization coefficient used for the normalization in a register or the like. Further, the CPU stores the normalized frequency component data together with the frame number in an internal memory or the like.

  In step S14, the CPU searches whether there is a frame having data similar to the normalized frequency component data of the current frame in the normalized frequency component data of the past frame. The CPU performs this search by calculating the above equation (1) for the past frame. The similarity calculation with the past frame may limit the range of the similarity calculation for the past frame from the frame immediately before the encoding target frame to about 100 frames, for example. Further, by separately obtaining the pitch frequency of the audio signal and determining a past frame to be searched based on this pitch frequency, it is possible to reduce the amount of calculation required for the similar frame search.

  In step S15, the CPU determines whether there is a similar frame based on the execution result of step S14. If there is no similar frame, a flag indicating the G frame is set in the current frame. Then, the processing of the CPU proceeds to step S18. If there is a similar frame, the CPU proceeds to step S16.

  In step S16, the CPU selects one similar frame, preferably the most similar frame. Then, the CPU stores the frame number of the selected similar frame in the register. Further, the CPU calculates frequency component difference data, which is a difference signal between the normalized frequency component data of the current frame and the normalized frequency component data of the similar frame. Thereafter, the processing of the CPU proceeds to step S17.

  In step S17, the CPU performs the above-described difference threshold process. At this time, the difference threshold value (DT) may be a fixed value, but the difference threshold value may be variable according to the quality required for the reproduced audio signal. For example, in this embodiment, when the difference threshold value (DT) is set to zero, all frequency component difference data are encoded, and encoding close to lossless encoding can be realized. That is, the present embodiment can seamlessly cover from lossless to lossy coding only by changing the difference threshold. Thereafter, the processing of the CPU proceeds to step S18.

  In step S18, if the current frame is a G frame, the CPU performs quantization and encoding on the normalized frequency component data, and the bit stream of the audio signal signal itself (“frequency component data in FIG. 2). )). Further, the CPU creates the G frame shown in FIG. 2 and writes it in the output memory 107 of FIG.

  In step S18, if the current frame is a D frame, the CPU performs quantization and encoding on the normalized frequency component difference data, and a bit stream of the audio signal signal itself (see “ Frequency component difference data ") is generated. Further, the CPU creates the D frame shown in FIG. 3 and writes it into the output memory 107 of FIG.

  In step S19, the CPU determines whether or not the processing for the last frame has been completed. When the determination result is “Yes”, the CPU ends the encoding process. Otherwise, the processing of the CPU returns to step S11 and performs the encoding processing of the audio signal of the next frame.

  Next, FIG. 6 will be described. FIG. 6 is a flowchart for explaining a program for realizing the decoding apparatus of FIG. 4 by a computer.

  In FIG. 6, it is assumed that the encoded data created by the encoding apparatus of FIG. 1 for a predetermined number of frames is stored in the input buffer 201 of FIG. 4 before step S21 and subsequent steps are executed. The CPU executes processing equivalent to the processing from the decoding processing unit 202 to the frequency inverse transformer 206 in FIG. 4 in steps S21 to S26 in FIG.

  In step S21, the CPU reads out one frame of encoded data from the input buffer 201 in FIG. This encoded data is in the format shown in FIG.

  The CPU reads the frame number, frame type (TYPE), and normalization coefficient from the read data, and sets each value in a register or the like. If this frame is a D frame, a similar frame number is extracted and this value is set in a register.

  When the frame is a G frame, the CPU restores normalized frequency component data and stores it in an internal memory or the like in association with the frame number. If the frame is a D frame, the normalized frequency component difference data is restored and stored in an internal memory or the like in association with the frame number.

  In step S22, the CPU determines whether the current frame is a G frame or a D frame. If it is the G frame, the CPU proceeds to step S24. If it is a D frame, the CPU proceeds to step S23.

  In step S23, the CPU adds the normalized frequency component difference data of the current frame and the normalized frequency component difference data corresponding to the similar frame number to obtain the normalized frequency component data of the current frame. Restore. The restored frequency component data is stored in the internal memory in association with the frame number.

  In step S24, the CPU denormalizes the normalized frequency component data restored in step S21 or step S23, and restores the frequency component data of the current frame.

  Note that the denormalization process in step S24 may be performed in step S21. In this case, in step S21, the CPU restores the frequency component data after denormalization in the G frame, and restores the frequency component difference data after denormalization in the D frame.

  In step S25, the CPU performs inverse frequency conversion processing on the restored frequency component data of the current frame to reproduce the audio signal.

  In step S26, the CPU determines whether or not the decoding process has been completed for all frames. If this determination is “NO”, the CPU returns to step S 21 to perform decoding processing for the subsequent frame. If the determination result is “YES”, the CPU ends the decoding process.

It is a block diagram which shows the structural example of the encoding apparatus concerning embodiment of this invention. It is a figure which shows the example of the G flame | frame in this invention. It is a figure which shows the example of D frame in this invention. It is a block diagram which shows the structural example of the decoding apparatus concerning embodiment of this invention. It is a flowchart for demonstrating the program for implement | achieving the encoding apparatus of FIG. 1 with a computer. 6 is a flowchart for explaining a program for realizing the decoding apparatus of FIG. 4 by a computer.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Encoding apparatus 101 Input buffer 102 Frequency converter 103 Normalization part 104 Difference signal extraction process part 105 Difference signal threshold value process part 106 Quantization / coding process part 107 Output memory 200 Decoding apparatus 201 Input buffer 202 Decoding process part 203 Addition Processing Unit 204 Inverse Normalization Unit 205 Frequency Inverse Transformer 206 Output Memory

Claims (23)

An audio signal encoding method including the following steps (A) to (D):
(A) normalizing the audio signal converted to the frequency domain based on the peak value of the frame of the audio signal converted to the frequency domain ;
(B) Search for a preceding frame similar to the encoding target frame from the beginning of the normalized audio signal for one song to immediately before the encoding target frame;
(C) When a similar frame is searched, the frame number of the encoding target frame , information indicating that the encoding target frame of the frame number is determined to be a similar frame, and similar to the encoding target frame Encode the frequency component difference data with the frame,
(D) When a similar frame is not searched, the frame number of the encoding target frame and the frequency component of the encoding target frame are encoded.   The audio signal encoding method according to claim 1, further comprising a step of storing a normalization coefficient used for the normalization. The encoding method according to claim 1, wherein the step (B) includes:
When a preceding frame is searched for in which the sum of squares of the difference between the frequency component of the preceding frame and the frequency component of the encoding target frame is equal to or less than a predetermined threshold, it is determined that there is a similar frame. The encoding method according to claim 2, wherein the step (B) includes:
When a preceding frame in which the sum of squares of the difference between the normalized frequency component and the normalized frequency component of the encoding target frame is not more than a predetermined threshold is searched, it is determined that there is a similar frame. 5. The encoding method according to claim 1, wherein the step (C) includes:
The preceding frame having the frequency component most similar to the frequency component of the encoding target frame is set as the similar frame. General frame (G frames) containing the frequency component information of the frame the frame number and the frame number, the difference frame including a frequency component difference data between frames similar to the frame number of the frame numbers similar to the frame number and the frame number ( An audio signal decoding method for decoding an audio signal from a bit stream mixed with D frames),
The G frame and the D frame further include a normalization coefficient, and the G frame includes frequency component data normalized based on a normalization coefficient included in the G frame as the frequency component information, and the D frame includes An audio signal decoding method including difference frequency data normalized based on a normalization coefficient included in the D frame as the frequency component difference data and including the following steps (A) to (C).
(A) When the current frame is a G frame, the frequency component data is restored by decoding the frequency component information,
(B) If the current frame is a D frame,
Decoding the frequency component difference data, decoding the difference frequency component of the current frame to obtain frequency component data,
The difference frequency component of the decoded current frame is restored from the frequency component of the similar frame, and the frequency component of the current frame is restored.
(C) The frequency component restored in steps (A) and (B) is inversely transformed to generate an audio signal of the current frame. The audio signal decoding method according to claim 6, wherein
The method
The method further comprises the step of denormalizing the restored frequency component based on a normalization coefficient included in the G frame or D frame, and supplying the denormalized frequency component to the step (C). The audio signal decoding method according to claim 6, wherein
The step (A) restores the frequency component data by denormalizing the frequency component data based on a normalization coefficient included in the G frame ,
The step (B) denormalizes the frequency component difference data based on a normalization coefficient included in the D frame, and restores the frequency component based on the denormalized frequency component difference data. A normalization unit that normalizes the audio signal converted into the frequency domain using a normalization coefficient based on the peak value of the frame of the audio signal converted into the frequency domain ;
A search unit that searches for a preceding frame similar to the encoding target frame from the head of the normalized audio signal of one music piece to immediately before the encoding target frame;
When a similar frame is searched, the frame number of the encoding target frame , information indicating that the encoding target frame of the frame number is determined to be a similar frame, and the encoding target frame and the similar frame An encoding unit that encodes the frequency component difference data and encodes the frame number of the encoding target frame and the frequency component of the encoding target frame when a similar frame is not searched;
An audio signal encoding device comprising: The audio signal encoding device according to claim 9, wherein the normalization unit further stores the normalization coefficient . The encoding device according to claim 9, wherein
The search unit determines that there is a similar frame when a previous frame in which the sum of squares of the difference between the frequency component of the previous frame and the frequency component of the encoding target frame is found is equal to or less than a predetermined threshold. The encoding device according to claim 10, wherein
The search unit has a similar frame when a preceding frame in which the sum of squares of the difference between the normalized frequency component and the normalized frequency component of the encoding target frame is less than a predetermined threshold is searched. Is determined. The encoding device according to claim 9 to 12,
The encoding unit includes:
The preceding frame having the frequency component most similar to the frequency component of the encoding target frame is set as the similar frame. General frame (G frames) containing the frequency component information of the frame the frame number and the frame number, the difference frame including a frequency component difference data between frames similar to the frame number of the frame numbers similar to the frame number and the frame number ( An audio signal decoding device that decodes an audio signal from a bit stream mixed with D frame),
The G frame and the D frame further include a normalization coefficient, and the G frame includes frequency component data normalized based on a normalization coefficient included in the G frame as the frequency component information, and the D frame includes The difference frequency data normalized based on the normalization coefficient included in the D frame is included as the frequency component difference data,
When the current frame is a G frame, the frequency component information is decoded by restoring the frequency component information, and when the current frame is a D frame,
A decoding unit that decodes the frequency component difference data and decodes the difference frequency component of the frame to obtain frequency component data;
When the current frame is a G frame, an addition processing unit that restores the differential frequency component of the decoded current frame from the frequency component of the similar frame, and the decoding unit or the addition An inverse transform unit that inversely transforms the frequency component restored by the processing unit to generate an audio signal of the current frame;
An audio signal decoding device comprising: The audio signal decoding device according to claim 14, wherein
The decoding unit
The image processing apparatus further includes a denormalization unit that denormalizes the restored frequency component based on a normalization coefficient included in the G frame or the D frame . A program for causing a computer to execute an audio signal encoding process, and the program includes the following steps (A) to (D).
(A) normalizing the audio signal converted to the frequency domain based on the peak value of the frame of the audio signal converted to the frequency domain ;
(B) Search for a preceding frame similar to the encoding target frame from the beginning of the normalized audio signal for one song to immediately before the encoding target frame;
(C) When a similar frame is searched, the frame number of the encoding target frame , information indicating that the encoding target frame of the frame number is determined to be a similar frame, and similar to the encoding target frame Encode the frequency component difference data with the frame,
(D) When a similar frame is not searched, the frame number of the encoding target frame and the frequency component of the encoding target frame are encoded. The program according to claim 16, further comprising a step of storing a normalization coefficient used for the normalization. The program according to claim 16, wherein the step (B) includes:
When a preceding frame is searched for in which the sum of squares of the difference between the frequency component of the preceding frame and the frequency component of the encoding target frame is equal to or less than a predetermined threshold, it is determined that there is a similar frame. The program according to claim 17, wherein the step (B) includes:
When a preceding frame in which the sum of squares of the difference between the normalized frequency component and the normalized frequency component of the encoding target frame is not more than a predetermined threshold is searched, it is determined that there is a similar frame. The program according to any one of claims 16 to 19, wherein the step (C) includes:
The preceding frame having the frequency component most similar to the frequency component of the encoding target frame is set as the similar frame. A program for causing a computer to execute an audio signal decoding process,
General frame (G frames) containing the frequency component information of the frame the frame number and the frame number, the difference frame including a frequency component difference data between frames similar to the frame number of the frame numbers similar to the frame number and the frame number ( D frame) is a program for decoding an audio signal from a mixed bitstream,
The G frame and the D frame further include a normalization coefficient, and the G frame includes frequency component data normalized based on a normalization coefficient included in the G frame as the frequency component information, and the D frame includes A program including difference frequency data normalized based on a normalization coefficient included in the D frame as the frequency component difference data and including the following steps (A) to (C).
(A) When the current frame is a G frame, the frequency component data is restored by decoding the frequency component information,
(B) If the current frame is a D frame,
Decoding the frequency component difference data, decoding the difference frequency component of the current frame to obtain frequency component data,
The difference frequency component of the decoded current frame is restored from the frequency component of the similar frame, and the frequency component of the current frame is restored.
(C) The frequency component restored in steps (A) and (B) is inversely transformed to generate an audio signal of the current frame. The program according to claim 21, wherein
The program is
The method further comprises the step of denormalizing the restored frequency component based on a normalization coefficient included in the G frame or D frame, and supplying the denormalized frequency component to the step (C). The program according to claim 21, wherein
The step (A) restores the frequency component data by denormalizing the frequency component data based on a normalization coefficient included in the G frame ,
The step (B) denormalizes the frequency component difference data based on a normalization coefficient included in the D frame, and restores the frequency component based on the denormalized frequency component difference data.

Images