JP5235168B2 - Encoding method, decoding method, encoding device, decoding device, encoding program, decoding program - Google Patents

Description

  The present invention relates to a technique for encoding an input signal such as a speech signal by vector quantization and a technique for decoding a compressed code.

  When an acoustic signal whose level varies rapidly in time, such as a musical tone of castanets, is encoded by vector quantization in the frequency domain, the encoding distortion to the input signal is reduced in a time interval where the level in the input acoustic signal is small. This is perceived as noise that significantly impairs hearing. Such noise is called pre-echo. The encoding method described in Patent Literature 1 includes a first encoding method that performs encoding in the time domain and a second encoding method that performs encoding in the frequency domain. Either one encoding method or the second encoding method is selected and encoded (for example, refer to Patent Document 1).

  The occurrence of pre-echo can be reduced by switching between the first encoding method for encoding in the time domain and the second encoding method for encoding in the frequency domain, depending on the nature of the input signal. As a result, it was possible to obtain an efficient and high-quality decoded signal according to the characteristics and properties of the input signal.

Japanese Patent No. 3317470

  According to the encoder described in Patent Document 1, encoding can be performed with an optimal encoding method according to the nature of the input signal, but the quantization method used in the first encoding method and the second encoding method is On the other hand, there is a problem that it is necessary to hold a plurality of codebooks, and a lot of storage areas are required.

  In order to solve the above problem, in encoding, an input acoustic signal is processed in units of frames configured by collecting samples corresponding to a predetermined time length. Of the time domain encoding method and the frequency domain encoding method, an encoding method capable of obtaining an acoustic signal with higher quality when decoded is selected, and signal identification information as a selection result is output. When the time domain encoding method is selected, an acoustic signal configured in units of frames is encoded by vector quantization using the first codebook, and a signal quantization code index is output. When a frequency domain encoding method is selected, an acoustic signal composed of frames is time-frequency converted into a frequency domain signal, encoded by vector quantization using the first codebook, and the encoding result The signal quantization code index is output.

  In decoding, the signal quantization code index and signal identification information output from the code separation unit are used. When the signal identification information indicates a time domain encoding method, the signal quantization code index is decoded by inverse vector quantization using the first codebook to obtain a time domain decoded acoustic signal. If the signal identification information indicates a frequency domain encoding method, the signal quantization code index is decoded by inverse vector quantization using the first codebook, and then frequency-time converted to convert it to a time domain decoded signal. . Finally, the decoded signal is output by superimposing and adding the time domain decoded signal.

  By using the same codebook for the time domain encoding method and the frequency domain encoding method, the storage area can be made smaller than before.

The functional block diagram of the example of the encoding apparatus of 1st embodiment. The flowchart of the example of the encoding method of 1st embodiment. The functional block diagram of the example of the decoding apparatus of 1st embodiment. The flowchart of the example of the decoding method of 1st embodiment. The functional block diagram of the example of the encoding apparatus of 2nd embodiment. The flowchart of the example of the encoding method of 2nd embodiment. The functional block diagram of the example of the decoding apparatus of 2nd embodiment. The flowchart of the example of the decoding method of 2nd embodiment. The functional block diagram of the example of the encoding apparatus of 3rd embodiment. The flowchart of the example of the encoding method of 3rd embodiment. The functional block diagram of the example of the decoding apparatus of 3rd embodiment. The flowchart of the example of the decoding method of 3rd embodiment. The schematic diagram of subband division and subband synthesis. 6 is a flowchart of an example of dynamic bit allocation. Schematic diagram of dynamic bit allocation. The figure of the bit allocation table used for dynamic bit allocation.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[First embodiment]
<< Encoding >>
FIG. 1 illustrates functional blocks of the encoding apparatus according to the first embodiment. FIG. 2 illustrates a flowchart of the encoding method of the first embodiment.

  As illustrated in FIG. 1, the encoding apparatus according to the first embodiment includes a frame configuration unit 1, a signal identification unit 2, a switch 31, a time domain encoding unit 4, a frequency domain encoding unit 5, and a codebook storage unit 6. The code multiplexer 7 is included.

  The input acoustic signal is input to the frame configuration unit 1. The input acoustic signal is an acoustic signal sample acquired in a digital format from a sensor such as a microphone. The frame construction unit 1 collects the input acoustic signals by the number of samples corresponding to a certain time length to construct an acoustic signal for each frame, and sends the acoustic signal for each frame to the signal identification unit 2 (step C1). For example, when the sampling frequency is 32 kHz, a frame corresponding to 5 msec is composed of 160 samples.

  The signal identification unit 2 selects an encoding method capable of obtaining a higher quality acoustic signal when decoded from among the time domain encoding method and the frequency domain encoding method (step C2), and the selection result. When the signal identification information represents the time domain encoding method, the switch 31 is switched so that the input acoustic signal in units of frames is sent to the time domain encoding unit 4, and then the signal identification information is sent to the code multiplexing unit 7 ( Step C3). On the other hand, when the signal identification information represents the frequency domain encoding method, the switch 31 is switched so that the input acoustic signal in units of frames is sent to the frequency domain encoding unit 5, and then the signal identification information is encoded by the code multiplexing unit 7. (Step C4).

  In order to select an encoding method capable of obtaining an acoustic signal with higher quality when decoded, the signal identification unit 2 calculates a feature amount from the acoustic signal for each frame, for example, and calculates the feature amount and a predetermined threshold value. And the time domain coding method or the frequency domain coding method is selected depending on the size. As the feature amount used for identification, for example, the difference between the maximum absolute value of the sample values in the frame and the average absolute value of the sample values in the frame can be used. In this case, when the feature amount is larger than the predetermined threshold, it is determined that the acoustic signal undergoes a rapid time change in the frame, and the time domain encoding method is selected. On the other hand, if the feature amount is equal to or less than a predetermined threshold, a frequency domain encoding method is selected. The predetermined threshold is appropriately determined based on the required performance and specifications.

  When the time domain coding method is selected in the signal identification unit 2, the vector quantization unit 41 of the time domain coding unit 4 uses the first codebook to generate an acoustic signal for each frame by vector quantization. Encoding is performed (step C5), and a signal quantization code index as a result of the encoding is output to the code multiplexing unit 7.

  When the frequency domain encoding method is selected in the signal identification unit 2, the time-frequency conversion unit 51 of the frequency domain encoding unit 5 converts the acoustic signal for each frame into a frequency domain signal (step C7). For example, the signal is converted into a frequency domain signal by modified discrete cosine transform (hereinafter referred to as MDCT) or discrete cosine transform (hereinafter referred to as DCT).

  The vector quantization unit 52 of the frequency domain encoding unit 5 regards the converted frequency domain signal as an input vector, encodes it by vector quantization using the first codebook (step C8), A certain signal quantization code index is output to the code multiplexer 7.

  Both the vector quantization unit 41 of the time domain encoding unit 4 and the vector quantization unit 52 of the frequency domain encoding unit 5 perform vector quantization using the same first codebook stored in the codebook storage unit 6. Do. In this way, by using the same code book (in this embodiment, the first code book) regardless of the encoding method, the necessary storage area can be made smaller than before.

  The code multiplexing unit 7 arranges the signal identification information and the signal quantization code index in a predetermined order and outputs them as one data set (step C10). When the packet is transmitted to an IP communication network or the like, necessary header information is added to the data set and transmitted.

<< Decryption >>
FIG. 3 illustrates functional blocks of the decoding device according to the first embodiment. FIG. 4 illustrates a flowchart of the decoding method of the first embodiment.

  As illustrated in FIG. 1, the decoding device according to the first embodiment includes a code separation unit 8, a time domain decoding unit 9, a frequency domain decoding unit 10, a codebook storage unit 11, a superposition addition unit 12, switches 31 and 32, A signal identification unit 14 is included.

  The code separation unit 8 reads a predetermined number of bits from a predetermined position from the input data set, and separates the signal identification information and the signal quantization code index (step D1). The signal identification information is sent to the signal identification unit 14. The signal quantization code index is sent to the time domain decoding unit 9 or the frequency domain decoding unit 10 according to the switching state of the switch 32. The code separation unit 8 corresponds to the input unit in the claims, and step D1 corresponds to the input step in the claims.

  The signal identification unit 14 identifies whether the signal identification information indicates a time domain decoding method or a frequency domain decoding method, and switches the switch 32 (step D2).

  When the signal identification information indicates a time domain encoding method, the signal identification unit 14 switches the switch 32 so as to connect the code separation unit 8 and the time domain decoding unit 9, and overlaps the time domain decoding unit 9. The switch 33 is switched so as to connect the unit 12. Next, the time domain decoding unit 9 decodes the signal quantization code index (step D3).

  Specifically, the inverse vector quantization unit 91 of the time domain decoding unit 9 decodes the signal quantization code index by inverse vector quantization using the first codebook stored in the codebook storage unit 11 ( Step D3), outputting a time domain signal for each frame. The time domain signal in frame units is sent to the superposition addition unit 12.

  On the other hand, when the signal identification information indicates a frequency domain encoding method, the signal identification unit 14 switches the switch 32 so as to connect the code separation unit 8 and the frequency domain decoding unit 10, The switch 33 is switched so as to connect the superposition addition unit 12. Next, the frequency domain decoding unit 10 decodes the signal quantization code index (steps D4 and D5).

  Specifically, the inverse vector quantization unit 101 of the frequency domain decoding unit 10 decodes the signal quantization code index by inverse vector quantization using the first codebook stored in the codebook storage unit 11 (step D4), a frequency domain signal for each frame is sent to the frequency time conversion unit 102.

  The frequency time conversion unit 102 converts the frequency domain signal for each frame into a time domain signal for each frame by frequency time conversion corresponding to the time frequency conversion performed by the time frequency conversion unit 51 (FIG. 1) at the time of encoding. Conversion is performed (step D5). For example, when MDCT is used as time frequency conversion, inverse MDCT is executed. The time domain signal for each frame output by the frequency time conversion is sent to the superposition addition unit 12.

  Both the inverse vector quantization unit 91 of the time domain decoding unit 9 and the inverse vector quantization unit 101 of the frequency domain decoding unit 10 use the same first codebook stored in the codebook storage unit 11 to perform inverse vector quantization. I do. In this way, by using the same code book (in this embodiment, the first code book) regardless of the decoding method, the necessary storage area can be made smaller than before.

  The first code book stored in the code book storage unit 11 is the same code book as the first code book stored in the code book storage unit 6 of the encoding device.

  The superposition adding unit 12 superimposes and adds time domain signals for each frame to generate a continuous time domain signal (step D6). That is, a continuous time-domain signal is generated by adding a signal obtained by multiplying the time-domain signal for each frame by a window function such as a Hamming window, for example, by overlapping half the frame length.

[Second Embodiment]
The second embodiment is a part that performs vector quantization and inverse vector quantization in units of subframes obtained by dividing a frame, and is different from the first embodiment. Other parts are the same as in the first embodiment. Hereinafter, parts different from the first embodiment will be described with emphasis, and parts that are the same as those of the first embodiment will be denoted by the same reference numerals, and redundant description will be omitted.

<< Encoding >>
FIG. 5 illustrates functional blocks of the encoding apparatus according to the second embodiment. FIG. 6 illustrates a flowchart of the encoding method of the second embodiment.

The time domain encoding unit 4 of the second embodiment includes a subframe dividing unit 42 in addition to the vector quantization unit 41. Further, the frequency domain encoding unit 5 of the second embodiment includes a subband division unit 53 in addition to the time frequency conversion unit 51 and the vector quantization unit 52.
The input acoustic signal is converted into an acoustic signal for each frame by the frame construction unit 1, and signal identification information is determined and output by the signal identification unit 2 (steps C1, C2, C3, and C4). The above is the same as in the first embodiment.

  When the signal identification information represents a time domain encoding method, the subframe division unit 42 of the time domain encoding unit 4 performs frame unit output from the frame configuration unit 1 as illustrated in FIG. Are divided into subframes (step C11). For example, when the number of samples in one frame is 160 and the number of subframes or subbands is 10, the number of samples in one subframe is 16.

  The acoustic signal configured in units of subframes is sent to the vector quantization unit 41. The vector quantization unit 41 encodes an acoustic signal configured in units of subframes by vector quantization using a second codebook for each subframe, and outputs a signal quantization code index as a result of the encoding. (Step C5).

  When the signal identification information represents frequency domain coding, the time-frequency conversion unit 51 performs time-frequency conversion on the input acoustic signal in units of frames and outputs a frequency unit signal in units of frames, and the subband division unit 53 The frequency domain signal in frame units is divided into subband units (step C12) and sent to the vector quantization unit 52. The vector quantization unit 52 encodes the frequency domain signal divided by the subbands for each subband by vector quantization using the second codebook, and outputs a signal quantization code index.

  Both the vector quantization unit 41 of the time domain encoding unit 4 and the vector quantization unit 52 of the frequency domain encoding unit 5 perform vector quantization using the same second codebook stored in the codebook storage unit 15. Do.

<< Decryption >>
FIG. 7 illustrates functional blocks of the decoding device according to the second embodiment. FIG. 8 illustrates a flowchart of the decoding method according to the second embodiment.

  The time domain decoding unit 9 of the second embodiment includes a subframe synthesis unit 92 in addition to the inverse vector quantization unit 91. The frequency domain decoding unit 10 of the second embodiment includes a subband synthesis unit 103 in addition to the inverse vector quantization unit 101 and the frequency time conversion unit 102.

  When the signal identification information indicates a time domain encoding method, the inverse vector quantization unit 91 of the time domain decoding unit 9 performs inverse vector quantization using the second codebook stored in the codebook storage unit 16. The signal quantization code index is decoded to generate a time domain signal in subframe units (step D3).

  Then, as illustrated in FIG. 13B, the subframe synthesis unit 92 arranges the time domain signals for each subframe in time series order to generate a time domain signal in units of frames (step D7), and the superposition addition unit 12 Send to.

  When the signal identification information indicates a frequency domain coding method, the inverse vector quantization unit 101 of the frequency domain decoding unit 10 performs inverse vector quantization using the second codebook stored in the codebook storage unit 16. The signal quantization index is decoded to generate a frequency domain signal in subframe units (step D4).

  Then, the subband synthesizing unit 103 synthesizes the frequency domain signals in units of subframes, generates a frequency domain signal in units of frames (step D8), and sends it to the time frequency conversion unit 102.

  Both the inverse vector quantization unit 91 of the time domain decoding unit 9 and the inverse vector quantization unit 101 of the frequency domain decoding unit 10 use the same second codebook stored in the codebook storage unit 16 to perform inverse vector quantization. I do.

  Note that the second code book stored in the code book storage unit 16 is the same code book as the second code book stored in the code book storage unit 15 of the encoding device.

  By performing vector quantization for each subframe and subband shorter than the frame, the dimension of the code vector constituting the second codebook stored in the codebook storage units 15 and 16 can be reduced, and the storage area can be reduced. It can be made smaller.

[Third embodiment]
The third embodiment is different from the second embodiment in that the number of bits allocated for each subframe or subband is dynamically changed. Other parts are the same as in the second embodiment. Hereinafter, parts different from those of the second embodiment will be described with emphasis, and parts that are the same as those of the second embodiment will be denoted by the same reference numerals and redundant description thereof will be omitted.

<< Encoding >>
FIG. 9 illustrates functional blocks of the encoding apparatus according to the third embodiment. FIG. 10 illustrates a flowchart of the encoding method of the third embodiment.

  The encoding device of the third embodiment includes a dynamic bit allocation unit 13.

  The subframe or subband unit acoustic signal divided by the subframe division unit 42 or the subband division unit 53 is sent to the dynamic bit allocation unit 13.

  The dynamic bit allocation unit 13 allocates a different number of bits for each subframe or subband to each subframe or subband (steps C13 and C14), and allocates bit information as a result of the allocation to the vector quantization unit 41 or The data is sent to the vector quantization unit 52 and the code multiplexing unit 7. For example, dynamic bit allocation is performed such that more bits are allocated to subframes and subbands having higher auditory importance. Alternatively, more bits are assigned to subframes and subbands having a larger average amplitude. Here, the average amplitude is an average value of values of samples constituting the subframe and the subband.

Hereinafter, an example of dynamic bit allocation will be described.
A frequency domain signal composed of subbands is denoted as S ^(w) [k] (w = 0,..., W−1, k = 0,..., L′−1), and the wth subband frequency domain signal sequence. S ^(w) [k] average amplitude index A ^to [w] (w = 0,..., W−1) (hereinafter referred to as “wth subband average amplitude index”) is calculated according to the following equation. . Here, W is the number of subbands, and L ′ is the number of samples constituting the subband. round (x) is a function that rounds off the decimal part of x to make it an integer.

Next, allocated bit information B [w] (w = 0,..., W−1) of the w-th subband is calculated in the following procedure. First, w-th subband auditory importance ip [w] (w = 0,..., W−1) determined by the following equation is calculated from the w-th subband average amplitude index A ^to [w].
ip [w] = A ^to [w] / 2

Next, the assigned bit information B [w] of the wth subband is output by the binary search method using the wth subband auditory importance degree ip [w] and the bit assignment table R. In dynamic bit allocation, “water Level” is selected by a binary search based on the following equation, and “water Level λ” and the w th subband auditory importance degree ip [w] are used. Band allocation bit information B [w] is calculated. Bit allocation table R is different R-number of natural numbers _{b 0,} b 1 together, _{..., b R-1} is a table registered. b ₀ <b ₁ <... <b _R−1 .

Specifically, for example, the calculation may be performed by the method shown in FIG. First, parameters (maxIP, minIP, λ, i) are set to initial values (S50351). Bt [w] (w = 1,..., W−1) is calculated as a temporary B [w] value, and the sum of the calculated Bt [w] Sum_Bt = Σ ₀ ^W−1 Bt [w] Is obtained (S50352). It is confirmed whether Sum_Bt exceeds the allocatable total number of bits (total_bit_budget) (S50353).

If step S50353 is Yes, the parameter (minIP, λ, i) is changed (S50354). When step S50353 is No, B ⁱ [w] is set to Bt [w], and the parameters (maxIP, λ, i) are changed (S50355).

It is confirmed whether i is less than a predetermined constant (S50356). If step S50356 is Yes, the process returns to step S50352. When Step S50356 is No, B ⁱ [w] is output as allocated bit information B [w] of the w-th subband. In addition, when the iterative search of a predetermined number of times is completed, the above expression B [w] is evaluated.

  The process may be ended by separately defining a convergence condition for ending the iterative process. For example, a method of ending the process when the total number of allocated bits is equal to the total number of bits that can be allocated (total_bit_budget) can be considered. If the final total number of bits exceeds the allocatable total number of bits, for example, by assigning bits that are smaller by one table than the number of bits selected in the above equation, in order from the subband with the smallest ip [w]. The final w-th subband allocation bit information is determined by adjusting so that the total number of allocated bits is smaller than the total number of total bits.

  When dynamic bit allocation is performed in an input signal in subframe units, a time domain signal configured in subframe units may be used instead of a frequency domain signal configured in subband units.

  When the signal identification information indicates a time domain encoding method, the vector quantization unit 41 stores the second data stored in the codebook storage unit 6 within a search range determined according to the allocated number of bits B [w]. The code book is searched, and each subframe w (w = 1,..., W) is vector quantized (see step C5, FIG. 15 and FIG. 16). For example, since the number that can be represented by the number of B [w] bits is 2 ^ B [w], 2 ^ 0th to 2 ^ B [w] -1th in the second codebook B [w] code vectors are searched to select a vector closest to a vector composed of acoustic signals in subframe units, and the selected code vector is expressed by B [w] bits. , The signal quantization code index.

  When the signal identification information indicates a frequency domain encoding method, the vector quantization unit 52, like the vector quantization unit 41, uses a codebook within a search range determined according to the allocated number of bits B [w]. The second codebook stored in the storage unit 6 is searched, and each subband w (w = 1,..., W) is vector quantized (see step C8, FIG. 15 and FIG. 16).

  The assigned bit information is sent to the code multiplexing unit 7. The code multiplexing unit 7 arranges the signal identification information, the signal quantization code index, and the assigned bit information in a predetermined order to form a data set (step C9). . When transmitting on the IP network, necessary header information is added to form a packet and output.

<< Decryption >>
FIG. 11 illustrates functional blocks of the decoding device according to the third embodiment. FIG. 12 illustrates a flowchart of the decoding method according to the third embodiment.

  The code separation unit 8 reads a predetermined number of bits from a predetermined position of the input data set, and generates signal identification information, a signal quantization code index, and assigned bit information (step D1).

  The signal identification unit 14 switches the switches 32, 33, and 34 according to the signal identification information. When the signal identification information indicates a time domain encoding method, the signal identification unit 14 switches the switch 34 so as to connect the code separation unit 8 and the time domain decoding unit 9. On the other hand, when the signal identification information indicates a frequency domain encoding method, the signal identification unit 14 switches the switch 34 so as to connect the code separation unit 8 and the frequency domain decoding unit 10.

  When the signal identification information indicates a time domain encoding method, the inverse vector quantization unit 91 of the time domain decoding unit 9 obtains a signal quantization index for each subframe with reference to the assigned bit information (step D9). ). Then, the signal quantization code index is decoded by inverse vector quantization using the second codebook stored in the codebook storage unit 16 to generate a time domain signal for each subframe (step D3).

  When the signal identification information indicates a frequency domain encoding method, the inverse vector quantization unit 101 of the frequency domain decoding unit 10 refers to the assigned bit information and obtains a signal quantization index for each subframe (step D9). ). Then, the signal quantization code index is decoded by inverse vector quantization using the second codebook stored in the codebook storage unit 16 to generate a frequency domain signal for each subband (step D4).

  Both the inverse vector quantization unit 91 of the time domain decoding unit 9 and the inverse vector quantization unit 101 of the frequency domain decoding unit 10 use the same second codebook stored in the codebook storage unit 16 to perform inverse vector quantization. I do. In this way, by using the same code book (in this embodiment, the second code book) regardless of the decoding method, the necessary storage area can be made smaller than before.

  The operations of the subframe synthesis unit 92, the subband synthesis unit 103, the frequency time conversion unit 102, and the superposition addition unit 12 are the same as those in the second embodiment.

  In this way, by assigning more bits to subframes or subbands with a large average amplitude, that is, auditory important subframes or subbands, encoding is performed as accurately as possible within a limited code length. be able to.

[Modifications, etc.]
The encoding method and the decoding method can be realized by a computer. Processing contents of functions that each device should have are described by a program. Then, by executing this program on a computer, each processing function in each apparatus is realized on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. In this embodiment, these apparatuses are configured by executing a predetermined program on a computer. However, at least a part of these processing contents may be realized by hardware.

  The present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit of the present invention.

DESCRIPTION OF SYMBOLS 1 Frame structure part 2 Signal identification part 4 Time domain encoding part 41 Vector quantization part 42 Subframe division | segmentation part 5 Frequency domain encoding part 51 Time frequency conversion part 52 Vector quantization part 53 Subframe division part 6 1st codebook Storage unit 9 Time domain decoding unit 91 Inverse vector quantization unit 92 Subframe synthesis unit 10 Frequency domain decoding unit 101 Inverse vector quantization unit 102 Time frequency conversion unit 103 Subframe synthesis unit 11 Second codebook storage unit 13 Dynamic bit Allocation unit 14 Signal identification unit 15 First codebook storage unit 16 Second codebook storage unit

Claims (11)

A frame configuration step for configuring a sound signal for each frame by collecting a predetermined number of samples constituting the input sound signal;
A signal identification method that selects a coding method capable of obtaining an acoustic signal with higher quality when decoded from among a time domain coding method and a frequency domain coding method, and outputs signal identification information as a result of the selection. Steps,
When the time domain encoding method is selected, the time for encoding the acoustic signal for each frame by vector quantization using the first codebook and outputting the signal quantization code index that is the encoding result A region encoding step;
When a frequency domain encoding method is selected, the acoustic signal for each frame is converted into a frequency domain signal, encoded by vector quantization using the first codebook, and a signal that is the encoding result A frequency domain encoding step for outputting a quantization code index;
An encoding method including: The encoding method according to claim 1,
The time domain encoding step includes a subframe division step for dividing the acoustic signal for each frame into subframes shorter than the predetermined time length, and vector quantization of the acoustic signal for each subframe using a second codebook. And a vector quantization step for outputting a signal quantization code index as a result of the encoding, and
The frequency domain encoding step includes a time frequency conversion step for converting the acoustic signal for each frame into a frequency domain signal, and a subband division step for dividing the converted frequency domain signal into subbands shorter than the predetermined time length. And a vector quantization step of encoding a frequency domain signal for each subband by vector quantization using the second codebook, and outputting a signal quantization code index that is a result of the encoding.
An encoding method characterized by the above. The encoding method according to claim 2, wherein
The time domain encoding step further includes a dynamic bit allocation step of allocating a different number of bits for each subframe to each subframe and outputting allocation bit information which is a result of the bit allocation, and the allocated bits are Searching for the code vector included in the second codebook, the more the number is, and representing the signal quantization code index with the assigned bits.
The frequency domain encoding step further includes a dynamic bit allocation step of allocating a different number of bits for each subband to each subband and outputting allocation bit information as a result of the bit allocation, and the allocated bits are Searching for the code vector included in the second codebook, the more the number is, and representing the signal quantization code index with the assigned bits.
An encoding method characterized by the above. An input step in which the signal quantization code index output by the encoding method according to claim 1 and the signal identification information are input;
A signal identification step for identifying the encoding method indicated by the signal identification information;
When the signal identification information indicates a time domain encoding method, a time domain decoding step of decoding the signal quantization code index by inverse vector quantization using the first codebook;
If the signal identification information indicates a frequency domain encoding method, a frequency domain decoding step of decoding the signal quantization code index by inverse vector quantization using the first codebook and converting it to a time domain signal; ,
A decoding method including: An input step in which the signal quantization code index output by the encoding method according to claim 2 or 3 and the signal identification information are input;
The time-domain decoding step includes: an inverse vector quantization step for generating a time-domain signal for each subframe by decoding the signal quantization code index by inverse vector quantization using the second codebook; Including a subframe synthesis step of generating time domain signals for each frame by arranging time domain signals for each subframe in chronological order;
The frequency domain decoding step includes: an inverse vector quantization step for generating a frequency domain signal for each subband by decoding the signal quantization code index by inverse vector quantization using the second codebook; A subband combining step for generating a frequency domain signal for each frame by combining frequency domain signals for each subband; and a time domain signal conversion step for converting the generated frequency domain signal for each frame into a time domain signal. ,
A decoding method characterized by the above. A frame configuration unit that configures an acoustic signal for each frame by collecting a predetermined number of samples constituting the input acoustic signal,
A signal identification method that selects a coding method capable of obtaining an acoustic signal with higher quality when decoded from among a time domain coding method and a frequency domain coding method, and outputs signal identification information as a result of the selection. And
When the time domain encoding method is selected, the time for encoding the acoustic signal for each frame by vector quantization using the first codebook and outputting the signal quantization code index that is the encoding result A region encoding unit;
When a frequency domain encoding method is selected, the acoustic signal for each frame is converted into a frequency domain signal, encoded by vector quantization using the first codebook, and a signal that is the encoding result A frequency domain encoding unit for outputting a quantization code index;
An encoding device including: The encoding device according to claim 6, wherein
The time domain encoding unit includes a subframe dividing unit that divides the acoustic signal for each frame into subframes shorter than the predetermined time length, and vector quantization of the acoustic signal for each subframe using a second codebook. And a vector quantization unit that outputs a signal quantization code index that is a result of the encoding, and
The frequency domain encoding unit includes a time frequency conversion unit that converts the acoustic signal for each frame into a frequency domain signal, and a subband division unit that divides the converted frequency domain signal into subbands shorter than the predetermined time length. And a vector quantization unit that encodes a frequency domain signal for each subband by vector quantization using the second codebook, and outputs a signal quantization code index that is a result of the encoding.
An encoding apparatus characterized by that. The encoding device according to claim 7,
The time domain encoding unit further includes a dynamic bit allocation unit that allocates a different number of bits for each subframe to each subframe, and outputs allocation bit information that is a result of the bit allocation. The greater the number, the more code vectors included in the second codebook are searched, and the assigned bit represents the signal quantization code index.
The frequency domain encoding unit further includes a dynamic bit allocation unit that allocates a different number of bits for each subframe to each subframe and outputs allocation bit information that is a result of the bit allocation. The greater the number, the more code vectors included in the second codebook are searched, and the assigned bit represents the signal quantization code index.
An encoding apparatus characterized by that. An input unit to which the signal quantization code index and the signal identification information output by the encoding device according to claim 6 are input;
A signal identification unit for identifying an encoding method indicated by the signal identification information;
When the signal identification information indicates a time domain encoding method, a time domain decoding unit that decodes the signal quantization code index by inverse vector quantization using the first codebook;
When the signal identification information indicates a frequency domain encoding method, a frequency domain decoding unit that decodes the signal quantization code index by inverse vector quantization using the first codebook and converts it into a time domain signal; ,
A decoding device. An input unit to which the signal quantization code index and the signal identification information output by the encoding device according to claim 7 or 8 are input;
The time domain decoding unit is generated with an inverse vector quantization unit that generates a time domain signal for each subframe by decoding the signal quantization code index using the second codebook by inverse vector quantization. A time domain signal for each subframe is arranged in chronological order, and a subframe synthesis unit that generates a time domain signal for each frame, and
The frequency domain decoding unit is generated with an inverse vector quantization unit that generates a frequency domain signal for each subband by decoding the signal quantization code index by inverse vector quantization using the second codebook. A subband synthesis unit that synthesizes frequency domain signals for each subband to generate a frequency domain signal for each frame, and a time domain signal conversion unit that converts the generated frequency domain signal for each frame into a time domain signal ,
A decoding device characterized by the above.   A program for causing a computer to execute each step of the encoding method according to claim 1 or each step of the decoding method according to claim 4 or 5.

Images