JP5466618B2 - Encoding device, decoding device, encoding method, decoding method, and program thereof - Google Patents

Description

  The present invention relates to an encoding apparatus and encoding method for encoding an input signal and transmitting an output code, a decoding apparatus and decoding method for receiving an output code as an input code and generating an output signal, and a program thereof.

<Scalable coding>
G.711.1 described in Non-Patent Document 1 is known as a conventional technique for performing scalable coding and decoding of an audio signal. Non-Patent Document 1 describes G.I. Operations of the encoding device 10 and the decoding device 20 of 711.1 will be described with reference to FIG.

  A sequence X (1), X (2),..., X (N) of N digital signal samples sampled with a sampling period of 16 kHz and quantized with 16 bits is G.G. 711.1 is input to the encoding device 10. Note that N is the number of samples included in one frame, and N = 80.

  The G.711.1 encoding apparatus 10 performs scalable encoding on the input signals X (1), X (2),. A 711 bit string L0 and an 80-bit low-frequency reinforcement bit string L1 and an 80-bit high-frequency emphasis bit string L2 are output. G. The 711 bit string L0 is G. , S (1), S (2),..., S (N / 2), which are 320 bits (8 bits × 40 samples).

  G. In order to transmit the 711 bit string L0, the low band reinforcement bit string L1 and the high band emphasizing bit string L2 in real time, 64 kbit / s (8 bits × 8000 samples or 320 bits × 200 columns per second), 16 kbit / s ( Transmission bandwidths of 80 bits × 200 columns per second and 16 kbit / s (80 bits × 200 columns per second) are required. The transmission band means the amount of information that can be transmitted per unit time.

  G. When 711.1 is used, a relay device (router, gateway, etc.) used when transmitting a bit string, a transmission band such as a communication line (hereinafter referred to as “used transmission bandwidth”), and a decoding device performance (G.711. Depending on whether or not it corresponds to 1), some of the bit strings L0, L1, and L2 output from the encoding device 10 may be selected and transmitted and input to the decoding device.

G. The 711.1 decoding apparatus 20 performs one of the following modes A to D according to the input bit string.
(A) Mode A
G. G. 711 in which only the 711 bit string L0 is input. 711.1 decoding apparatus 20 decodes input bit string L0, and a sequence of telephone band voice samples Y (1), Y (2),..., Y (N / N) having a sampling frequency of 8 kHz obtained by decoding. 2) is output.
(B) Mode B
G. 711 bit string L0 and low band reinforcement bit string L1 are input. The 711.1 decoding apparatus 20 decodes the input bit strings L0 and L1, and although the sampling frequency obtained by decoding is 8 kHz, the sample string Y (1), Y (2),..., Y (N / 2 ), V (1), V (2),..., V (N / 2) of telephone band voice samples with better hearing quality than the above.
(C) Mode C
G. 711 bit string L0 and high frequency emphasis bit string L2 are input. 711.1 decoding apparatus 20 decodes input bit sequences L0 and L2, and a sequence of wideband speech samples W (1), W (2),..., W (N ) Is output.
(D) Mode D
G. 711 bit string L0, low band reinforcement bit string L1, and high band emphasizing bit string L2 are input. 711.1 decoding apparatus 20 decodes input bit strings L0, L1 and L2, and the sampling frequency obtained by the decoding is 16 kHz, but sample strings W (1), W (2),..., W (N ), U (1), U (2),..., U (N) are output.

  As shown in FIG. 2, when the used transmission band is less than 64 kbit / s, G.I. The 711.1 bit string cannot be transmitted in real time.

  When the used transmission band is 64 kbit / s or more and less than 80 kbit / s, the G. Only the 711 bit string L0 can be transmitted in real time. In this case, G. In the 711.1 decoding device 20, only mode A can be used.

  When the used transmission band is 80 kbit / s or more and less than 96 kbit / s, 711 bit string L0 only, or G. 711 bit string L0 and low band reinforcement bit string L1 are transmitted. A bit string for obtaining a desired auditory quality can be selected and transmitted in real time from three ways of transmitting the 711 bit string L0 and the high-frequency emphasized bit string L2. In this case, G. The 711.1 decoding apparatus 20 can obtain, as an output, a sequence of speech samples obtained by any of the operations of mode A, mode B, and mode C according to the input bit sequence.

  When the used transmission band is 96 kbit / s or more, G. 711 bit string L0 or G. 711 bit string L0 and low band reinforcement bit string L1 are transmitted. 711 bit string L0 and high-frequency emphasis bit string L2 are transmitted. A bit string for obtaining desired auditory quality can be selected and transmitted in real time from the four types of transmission of the 711 bit string L0, the low band reinforcement bit string L1, and the high band emphasizing bit string L2. In this case, G. The 711.1 decoding apparatus 20 can obtain, as an output, a sequence of speech samples obtained by any of the operations of mode A, mode B, mode C, and mode D according to the input bit sequence.

In this way, scalable coding is performed by transmitting only a transmittable bit string determined by the use transmission band, or by selecting a desired combination from the combinations of transmittable bit strings determined by the use transmission band. Quality can be obtained.
<Lossless coding>
G.711.0 described in Non-Patent Document 2 is known as a conventional technique for lossless encoding and decoding of an audio signal. Non-patent document 2 describes G.I. Operations of 711.0 encoding apparatus 30 and decoding apparatus 40 will be described with reference to FIG.

  G. , S (I) is a G.711-compliant sample string S (1), S (2),. It is input to the 711.0 encoding device 30. Note that I is the number of samples included in one frame and is any of 40, 80, 160, 240, and 320. When I = N / 2, the input signal is the above S (1), S (2),... S (N / 2), that is, G.I. It is a 711 bit string L0. G. The 711.0 encoding device 30 performs lossless encoding on the input signal L0 and outputs a lossless bit string LL having a variable length.

  In addition, G. The lossless bit string LL output from the 711.0 encoding device 30 requires a prefix code of 1 byte (8 bits) for each frame (for each I). Therefore, the required transmission bandwidth is 65.6 kbit / s (payload 64 kbit + prefix code 1.6 kbit) or less when I = 40, and 64.8 kbit / s (payload 64 kbit + prefix code 0.8 kbit) when I = 80. ) Hereinafter, when I = 160, 64.4 kbit / s (payload 64 kbit + prefix code 0.4 kbit) or less, and when I = 240, 64.267 kbit / s (payload 64 kbit + prefix code 0.247 kbit) or less, I = 320 Is 64.2 kbit / s (payload 64 kbit + prefix code 0.2 kbit) or less. The prefix code includes information such as the frame length (the number of bytes included in the frame) and what kind of lossless encoding has been performed.

  G. to which the lossless bit string LL is input. The 71.10 decoding device 40 decodes the input lossless bit string LL, and the G.71. A sample string L0 represented by 8 bits conforming to 711 is output.

ITU-T Recommendation G.711.1, "Wideband embedded extension for G.711 pulsecode modulation", ITU-T, 2008. ITU-T Recommendation G.711.0, "Lossless compression of G.711 pulse code modulation", ITU-T, 2009.

  The scalable coding has a problem that a larger transmission bandwidth is required to improve the quality of the output signal. In addition, lossless encoding has a problem that a larger transmission bandwidth is required than in the case where lossless encoding is not used in order to cope with the worst value.

  An object of the present invention is to provide an encoding technique that can effectively use the bit reduction effect of lossless encoding and transmit more enhancement layer codes of scalable encoding. It is another object of the present invention to provide a decoding technique for decoding an input code that has been encoded.

  In order to solve the above problems, the coding according to the present invention performs scalable coding of a digital input signal, generates a base layer code and one or more extension layer codes, performs lossless coding of the base layer code, and generates a lossless code. And select a combination having a code amount per unit time that is less than or equal to the use transmission band and having the largest code amount per unit time from among a plurality of combinations of lossless codes and enhancement layer codes. .

  The decoding according to the present invention uses the mode information to determine the omission rule in the omission code or the combination of the enhancement layer code and the lossless code. If the omission code is included in the input code, the omission code is omitted. If the input layer includes a combination of an extended layer code and a lossless code, the lossless code is decoded to obtain a base layer code. An output signal is obtained by decoding a base layer code obtained by inserting a value or a base layer code and an enhancement layer code obtained by decoding a lossless code.

  The encoding technique according to the present invention has an effect that more scalable enhancement layer codes can be transmitted. In addition, the decoding apparatus according to the present invention has an effect that it can decode an input code including a larger number of enhancement layer codes and output a high-quality signal.

G. 711.1 is a diagram for explaining the scalable encoding device 10 and the decoding device 20. FIG. G. 711.1 is a diagram showing a necessary transmission band required for decoding each of modes A to D by the decoding device 20. FIG. G. The figure for demonstrating the 711.0 lossless encoding apparatus 30 and the decoding apparatus 40. FIG. The figure which shows the utilization transmission band required when the code amount converted per unit time by lossless encoding can be reduced by 16 kbit / s or more. The figure which shows the utilization transmission band required when the code amount converted per unit time by lossless encoding can be reduced by 32 kbit / s or more. The figure which shows the structural example of the encoding apparatus 100 and the decoding apparatus 200. FIG. The figure which shows the processing flow of the encoding apparatus. The figure which shows the processing flow of the decoding apparatus 200. FIG. 3 is a diagram illustrating a configuration example of an encoding apparatus 300. The figure which shows the processing flow of the encoding apparatus 300. The figure for demonstrating the method of selecting an output bit sequence using the value of MDCT gain. FIG. FIG. 12B is a diagram illustrating a data example when a scalable code of 711.1 is transmitted as an IP packet, and FIG. 12B illustrates a data example when the scalable code of the encoding device 100 or 300 is transmitted as an IP packet. FIG. 3 is a diagram illustrating a configuration example of an encoding apparatus 500. The figure which shows the processing flow of the encoding apparatus 500. The figure which shows the example of data of the output bit sequence in the case of mode (delta). The figure which shows the data example of the output bit sequence in the case of mode (gamma). The figure which shows the example of data of the output bit sequence in the case of mode (beta). The figure which shows the example of data of the output bit sequence in the case of mode (alpha). The figure which shows the structural example of the decoding apparatus 600. FIG. The figure which shows the processing flow of the decoding apparatus 600. FIG. 3 is a diagram illustrating a configuration example of an encoding apparatus 700. The figure which shows the processing flow of the encoding apparatus 700. The figure which shows the data example of the output bit sequence in the case of modes 1-4, and the data example at the time of insertion. The figure which shows the example of data of the output bit sequence in the case of modes 5-8, and the example of data at the time of insertion. The figure which shows the structural example of the decoding apparatus 800. FIG. The figure which shows the processing flow of the decoding apparatus 800. The figure which shows the example of data when mode information (modes 1-3, (alpha) -delta) is made into a variable-length code. The figure which shows the example of data when mode information (modes 1-8, (alpha) -delta) is made into a variable-length code. The figure which shows the data example of the synchronous word which does not contain the synchronous word which contains mode information, and mode information. The figure which shows the example of data of the synchronous word containing mode information (mode 1, (alpha) -delta). The figure which shows the example of data of the synchronous word containing mode information (Modes 1-3, (alpha) -delta). The figure which shows a simulation result. The figure which shows the example of data in case the same IP header is used by all the output bit strings.

<Points of invention>
The basic layer code obtained by scalable coding is lossless coded, the code amount is reduced, and the code amount per unit time of the code composed of the lossless code and the extended layer code obtained by scalable coding is less than the use transmission band. If possible, a high-quality signal can be transmitted with a small code amount.

  For example, G. 711.0 lossless coding allows G. If the code amount converted per unit time of the 711 bit string L0 can be reduced by 16 kbit / s or more, even if the low-frequency reinforcement bit string L1 or the high-frequency emphasis bit string L2 is transmitted, it is within 64 kbit / s (see FIG. 4). G. 711.0 lossless coding allows G. If the code amount converted per unit time of the 711 bit string L0 can be reduced by 32 kbit / s or more, the low band reinforcing bit string L1 and the high band emphasizing bit string L2 can be transmitted within 64 kbit / s (see FIG. 5).

  Hereinafter, embodiments of the present invention will be described in detail.

<Encoder 100>
The encoding apparatus 100 according to the first embodiment will be described with reference to FIGS. 6 and 7. The encoding apparatus 100 includes a scalable encoding unit 110 and a lossless encoding unit 120. In this embodiment, G.I. 711.1 scalable coding; A case where lossless coding of 711.0 is used will be described. 711 bit string L0, the enhancement layer code is a low-frequency reinforcement bit string L1 and a high-frequency emphasis bit string L2, and the lossless code is a lossless bit string. However, other scalable coding and lossless coding may be used.

  The encoding apparatus 100 encodes input signals X (1), X (2),..., X (N) and outputs an output code (lossless encoded G.711 bit string LL, low-frequency reinforcement bit string L1, and high-frequency emphasized bit string). L2) is transmitted.

<Scalable Encoding Unit 110>
The scalable encoding unit 110 converts the digital input signals X (1), X (2),. 71.1.1, and G. A 711 bit string L0, a low band reinforcement bit string L1, and a high band emphasizing bit string L2 are generated (s110). The scalable encoding unit 110 sends L0 to the lossless encoding unit 120 and transmits L1 and L2 to the decoding device 20 (s122). The scalable encoding unit 110 is, for example, G. It operates in the same manner as the 711.1 encoding apparatus 10 (see Non-Patent Document 1).

<Lossless Encoding Unit 120>
The lossless encoding unit 120 is a G. 711 bit string L0 to G. The data is encoded by 711.0 to generate a lossless bit string LL (s120) and transmitted to the decoding device 20 (s122). The lossless encoding unit 120 is, for example, G. It operates similarly to the 711.0 encoding apparatus 30 (see Non-Patent Document 2).

  Here, G. A value obtained by subtracting the code amount of the lossless bit string LL from the code amount of the 711 bit string L0, that is, G. (D) kbit / s is obtained by converting the amount of code that can be reduced by the lossless encoding of 711.0 per unit time. For example, if one frame is 5 ms, G. A value obtained by multiplying the value obtained by subtracting the code amount of the lossless bit string LL from the code amount of the 711 bit string L0 by 200 is defined as d. At this time, G. Information corresponding to 711 can be transmitted in real time in the transmission transmission band of (64-d) kbit / s.

When the used transmission band is less than (64-d) kbit / s, the bit string output by the encoding device 100 cannot be transmitted in real time.
When the used transmission band is (64-d) kbit / s or more and less than (80-d) kbit / s, only the lossless bit string LL among the bit strings output by the encoding device 100 can be transmitted in real time.

  When the used transmission band is not less than (80-d) kbit / s and less than (96-d) kbit / s, only the lossless bit string LL is transmitted among the bit strings output by the encoding device 100 or the lossless bit string LL is low. A bit string for obtaining a desired auditory quality can be selected and transmitted in real time from the three types of transmission of the area reinforcement bit string L1 or the lossless bit string LL and the high-frequency emphasis bit string L2.

  When the used transmission band is equal to or greater than (96-d) kbit / s, whether only the lossless bit string LL is transmitted among the bit strings output by the encoding device 100, or the lossless bit string LL and the low-band reinforcement bit string L1 are transmitted. A bit string for obtaining a desired auditory quality is selected from four ways of transmitting the lossless bit string LL and the high-frequency emphasis bit string L2, or transmitting the lossless bit string LL, the low-frequency reinforcement bit string L1, and the high-frequency emphasis bit string L2. It can be transmitted in real time.

<Decoding device 200>
The decoding device 200 according to the first embodiment will be described with reference to FIGS. 6 and 8. The decoding device 200 includes a lossless decoding unit 230 and a scalable decoding unit 250. The decoding apparatus 200 receives and decodes the output bit string of the encoding apparatus 100 as an input bit string (for example, only LL, LL and L1, LL and L2, or LL, L1, and L2), generates an output signal, and outputs it.

<Lossless Decoding Unit 230>
The lossless decoding unit 230 uses a combination of an extension layer code (low band reinforcement bit string L1 and high band emphasizing bit string L2) and a lossless bit string LL (LL only, LL and L1, LL and L2, or LL and L1 and L2) as an input bit string. ) Is included, the lossless bit string LL is decoded and G.G. A 711-bit string L0 is obtained (s230). Note that only LL is one of the combinations of the enhancement layer code and the lossless bit string. G. obtained by decoding. The 711-bit string L0 is output to the scalable decoding unit 250. The lossless decoding unit 230 is, for example, the G. It operates in the same manner as the 711.0 decoding device 40 (see Non-Patent Document 2).

<Scalable Decoding Unit 250>
The scalable decoding unit 250 is a G. The 711 bit string L0, the low band reinforcement bit string L1, and the high band emphasizing bit string L2 are decoded to obtain an output signal (s255), which is output. The scalable decoding unit 250 is, for example, the G. It operates in the same manner as the 711.1 decoding apparatus 20 (see Non-Patent Document 1).

  For example, since scalable decoding section 250 of decoding apparatus 200 receives only L0 (s251, s253), it performs scalable decoding in mode A (s255A), and an output signal obtained by decoding (sampling frequency is 8 kHz) A sequence of telephone band voice samples Y (1), Y (2),..., Y (N / 2)) is output.

  Since scalable decoding section 250 of decoding apparatus 200 receives L0 and L1 (s251, s252), performs scalable decoding in mode A or B (s255B), and outputs an output signal (described above). Y (1), Y (2),..., Y (N / 2), or a sequence V (1), V (2),. N / 2)) is output.

  Since scalable decoding section 250 of decoding apparatus 200 receives L0 and L2 (s251, s253), performs scalable decoding in mode A or C (s255C), and outputs an output signal (described above). Y (1), Y (2),..., Y (N / 2), or a sequence of wideband speech samples having a sampling frequency of 16 kHz W (1), W (2),. Output.

  Since the scalable decoding unit 250 of the decoding device 200 receives L0, L1, and L2 (s251, s252), performs scalable decoding of any of modes A to D (s255D), and outputs obtained by decoding Signal (Y (1), Y (2),..., Y (N / 2) described above, or V (1), V (2),..., V (N / 2), or W (1) , W (2),..., W (N), or a sequence of wideband speech samples U (1), U (2),.

<Effect>
By adopting such a configuration, it is possible to transmit more enhancement layer codes of scalable coding. In addition, there is an effect that a high quality signal can be output by decoding an input code including a larger number of enhancement layer codes.

<Encoding device 300>
A coding apparatus 300 according to the second embodiment will be described with reference to FIGS. 9 to 11. Only parts different from the encoding apparatus 100 will be described. The encoding device 300 includes a bit string selection unit 330 in addition to the scalable encoding unit 110 and the lossless encoding unit 120.
After obtaining the lossless bit string LL, the low band reinforcement bit string L1, and the high band emphasizing bit string L2, the encoding apparatus 300 selects a bit string to be output according to the used transmission band.

<Bit string selector 330>
The bit string selection unit 330 includes a plurality of types of combinations (LL only, LL and L1, LL and L2, or LL and L1) of the lossless bit string LL and the enhancement layer code (low band reinforcement bit string L1 and high band emphasis bit string L2). A combination is selected from L2) where the code amount converted per unit time is equal to or less than the used transmission band and the code amount converted per unit time is the largest (s330).

  The lossless bit sequence LL is input from the lossless encoding unit 120 to the bit sequence selection unit 330, and the low-frequency reinforcement bit sequence L1 and the high-frequency emphasis bit sequence L2 are input from the scalable encoding unit 110. In addition, the use transmission band is input before the output code is transmitted. When the used transmission band is known, the used transmission band may be stored in the storage unit 331 in the bit string selection unit 330.

  When the code amount converted per unit time of the lossless bit string LL is larger than (used transmission band−16 kbit / s) (s332), the bit string selection unit 330 outputs only the lossless bit string LL as the output bit string L (s333). . Further, when the code amount converted per unit time of the lossless bit string LL is larger than (used transmission band−32 kbit / s) (s332) and less than (used transmission band−16 kbit / s) (s334), The lossless bit string LL and the low-frequency reinforcement bit string L1 or the lossless bit string LL and the high-frequency emphasis bit string L2 are output as the output bit string L (s336). Further, when the code amount converted per unit time of the lossless bit string LL is equal to or less than (utilization transmission band−32 kbit / s) (s332, s334), the lossless bit string LL, the low band reinforcement bit string L1, and the high band emphasizing bit string L2 Is output as an output bit string L (s335).

  In the above description, the comparison and determination are performed using the code amount converted per unit time of the lossless bit string LL and (used transmission bandwidth−xkbit / s) (for example, x is 16 or 32). The comparison and determination may be performed so that (the code amount converted per unit time of the lossless bit string LL + xkbit / s) and the used transmission band are equivalent to the above. In addition, a comparison equivalent to the above is made using the code amount of the lossless bit string LL and the code amount obtained by converting each of the used transmission band and xkbit / s into a value at the same time length as the lossless bit string LL. You may make judgments. When performing comparison or determination with a time length of 1 second, it is possible to use the code amount of the lossless bit string LL for 1 second, the used transmission band for 1 second, and xkbit / s.

  When the transmission bit rate is 64 kbit / s, the bit string selection unit 330 functions as shown in FIG. 7, for example. G. When the bit rate d reduced by 711.0 is 16 kbit / s or more, mode B or mode C is changed to G. When the bit rate d reduced by 711.0 is 32 or more kbit / s, the mode D is selected.

In s336, when the code amount converted per unit time of the lossless bit string LL is larger than (used transmission band−32 kbit / s) and equal to or smaller than (used transmission band−16 kbit / s), it is combined with the lossless bit string LL. The extended hierarchical code included in the output bit string can be selected from the low band reinforcement bit string L1 or the high band emphasizing bit string L2. As a selection method, for example, there are the following two methods.
(1) Selection method 1
Selection rules (for example, “always select the low-frequency reinforcement bit string L1”, “always select the high-frequency emphasis bit string L2”, etc.) are stored in the storage unit 331 in advance, and the extension hierarchy is determined according to the selection rules. Select a sign.
(2) Selection method 2
G. The high frequency emphasis bit string L2 defined by 711.1 includes information (8 bits) corresponding to a gain (MDCT gain) for emphasizing the high frequency.

When the MDCT gain is equal to or less than the threshold value T ₁ (for example, T ₁ = 1) (s336a), the bit string selection unit 330 selects a combination of the lossless bit string LL and the low-frequency reinforcement bit string L1 and outputs it as the output bit string L ( s336b). On the other hand, when MDCT gain is greater than the threshold value _{T 1 (s336a),} selects a combination of lossless bitstream LL and the high frequency emphasis bit string L2, as the output bit sequence L (s336c).

This is because when MDCT gain is small, the effect of improving the auditory quality is small even if wideband expansion is performed. Note that information corresponding to MDCT gain and MDCT gain correspond one-to-one. Therefore, even if the information corresponding to the MDCT gain is not decoded and the value of the MDCT gain is not obtained, the information corresponding to the MDCT gain larger than the threshold T ₁ is stored in the storage unit 331 or the like and acquired in the storage unit 331. and whether MDCT gain there is a corresponding information to determine the, MDCT gain may determine whether a thresholds _{T 1} below.

With such a configuration, it is possible to select an appropriate enhancement layer code according to the input signal, and to improve the auditory quality.
In addition to the output bit string L, the bit string selection unit 330 may output information indicating which bit string combination is the output bit string L as mode information.
The bit string L output from the encoding device 300 can be decoded by the decoding device 200 described in the first embodiment.

<Effect>
By adopting such a configuration, the same effect as in the first embodiment can be obtained. Furthermore, by selecting a bit string to be output according to the used transmission band and the code amount of the lossless bit string LL, it is possible to obtain high sound quality utilizing the used transmission band to the maximum.

  For example, G. 711 bit string L0 to G. If the code amount converted per unit time can be reduced by 16 kbit / s or more by making the lossless bit string LL by 711.0, the low-frequency reinforcement bit string L1 or the high-frequency emphasized bit string L2 is transmitted in addition to the lossless bit string LL. Is within 64 kbit / s (see FIG. 4). G. 711 bit string L0 to G. If the code amount converted per unit time can be reduced by 32 kbit / s or more by making the lossless bit string LL by 711.0, all of the lossless bit string LL, the low-frequency reinforcement bit string L1, and the high-frequency emphasis bit string L2 are 64 kbit / s. (See FIG. 5).

<Modification>
In the present embodiment, the case where the use transmission band is known in the bit string selection unit 330 is described, but it may be unknown. For example, prior to transmission of the output bit string L, a used transmission band measuring unit (not shown) may check the used transmission band and store it in the storage unit 331. Further, the used transmission band may be acquired by inquiring an existing SIP server or the like.

  G. When the scalable code of 711.1 is transmitted through the IP transmission path, as shown in FIG. Each of the 711 bit string L0, the low-frequency reinforcement bit string L1, and the high-frequency emphasis bit string L2 may be an IP payload, and each may be transmitted with an IP header. When the output bit string L of the first or second embodiment is transmitted through the IP transmission path, as shown in FIG. 12B, each of the lossless bit string LL, the low band reinforcement bit string L1, and the high band emphasizing bit string L2 is transferred to the IP payload. And each may be transmitted with an IP header.

  In this embodiment, a serial digital data transmission path can be synchronized only in byte units (for example, a communication path that performs real-time communication at 64 kbit / s or less like an ISDN network or a private wireless network). Assume application. Such a communication network transmits a sequence of samples of a digital voice signal (telephone band voice) that is originally sampled at a sampling period of 8 kHz and quantized at 8 bits at 64 kbit / s. By using a combination of coding and scalable coding, it is possible to transmit speech with a wider bandwidth as much as possible.

<Encoder 500>
A coding apparatus 500 according to the third embodiment will be described with reference to FIGS. 14 to 18. Only parts different from the encoding apparatus 300 will be described. The encoding device 500 further includes a synchronization word insertion unit 550 in addition to the scalable encoding unit 110 and the lossless encoding unit 120, and the processing content of the bit string selection unit 530 is different.

<Bit string selection unit 530>
The bit string selection unit 530 includes a mode information generation unit 533 and converts the unit per unit time from a plurality of types of combinations of the lossless bit string LL and the enhancement layer code (low band reinforcement bit string L1 and high band emphasis bit string L2). A combination whose code amount is equal to or less than the use transmission band and whose code amount converted per unit time is the largest is selected, and mode information generation unit 533 provided therein generates mode information indicating the combination to be selected. (S530). In FIG. 14, for s332 to s336, processing similar to that performed by the bit string selection unit 530 of the second embodiment is performed.

  For example, when the bit string selected by the bit string selection unit 530 is only the lossless bit string LL (s333), that is, when the decoding apparatus 600 described later is operated only in mode A, the mode information generation unit 533 outputs the lossless bit string. Mode information (mode α) indicating that only the bit string LL is formed is generated (s538).

  When the bit string selected by the bit string selection unit 530 is the lossless bit string LL and the low-frequency reinforcement bit string L1 (s336), that is, when the decoding apparatus 600 is operated in mode A or mode B, the mode information generation unit 533 Mode information (mode β) indicating that the output bit string is composed of the lossless bit string LL and the low-frequency reinforcement bit string L1 is generated (s539).

  When the bit string selected by the bit string selection unit 530 is the lossless bit string LL and the high frequency emphasis bit string L2 (s336), that is, when the decoding apparatus 600 is operated in mode A or mode C, the mode information generation unit 533 Mode information (mode γ) indicating that the output bit string is composed of the lossless bit string LL and the high-frequency emphasized bit string L2 is generated (s539).

When the bit strings selected by the bit string selection unit 530 are the lossless bit string LL, the low-frequency reinforcement bit string L1, and the high-frequency emphasis bit string L2 (s335), that is, the decoding apparatus 600 can be operated in all modes A to D. In this case, mode information (mode δ) indicating that the output bit string is composed of the lossless bit string LL, the low band reinforcement bit string L1, and the high band emphasizing bit string L2 is generated (s537).
For example, 2 bits are assigned to the mode information so that four modes can be distinguished. The bit string selection unit 530 outputs the selected bit string and mode information to the synchronization word insertion unit 550.

<Synchronous word insertion unit 550>
The synchronization word insertion unit 550 inserts a predetermined synchronization word at a predetermined position in the frame (s550). The synchronization word is a predetermined bit pattern composed of a combination of “0” and “1” bits.

  The synchronization word insertion unit 550 outputs K synchronization words for a fixed frame length of J bytes (for example, J = 40) and m output from the bit string selection unit 530 (where m is an integer satisfying 0 <m ≦ 8) ) Output including bit mode information and the selected bit string as an output bit string L of the encoding apparatus 500. The selected bit string is included in the remaining ((J−K) × 8−m) bits obtained by subtracting the synchronization word K bytes and the mode information m bits from the frame length J bytes. Therefore, the substantial available transmission band necessary for transmission of the output bit string L is (N ′ / N) × ((J−K) × 8−m) kbit / s. N ′ represents the number of samples per unit time, N represents the number of samples included in one frame as described above, and (N ′ / N) represents the number of frames per unit time. Therefore, a substantial use transmission band in the case of transmitting a digital audio signal sample sequence sampled at a sampling period of 8 kHz (in other words, a use transmission band for transmitting a signal representing the digital audio signal sample sequence itself). ), When the frame length J is 40 bytes, the synchronization word is 1 byte, the mode information is 2 bits, and one sample is 1 byte (in this case, J = N), 62 kbit / s Become.

  The bit string selected by the bit string selection unit 530 is always ((J−K) × 8−m) bits or less, but the selected bit string is exactly ((J−K) × 8−m)) bits. Is rare. Therefore, when the selected bit string is less than ((J−K) × 8−m)) bits, the synchronization word insertion unit 550 adds dummy bits to the remaining bits so that the output bit string L becomes J bytes. (For example, 0 or 1) is stored.

  As shown in FIGS. 15 to 18, the synchronization word insertion unit 550 outputs an output bit string L including the synchronization word, the mode information, and the bit string selected by the bit string selection unit 530. As described above, the output bit string L includes dummy bits as necessary. “Dummy” in each figure indicates the above “dummy bit”, and 0 or 1 is stored.

<Decoding device 600>
A decoding apparatus 600 according to the third embodiment will be described with reference to FIGS. 19 and 20. Only parts different from the decoding device 200 will be described. In addition to the lossless encoding unit 230, the decoding device 600 further includes a synchronization word determination unit 610 and a mode determination unit 620, and the processing contents of the scalable decoding unit 650 are different.

<Synchronous word determination unit 610>
The synchronization word determination unit 610 receives the output bit string L of the encoding device 500 as an input bit string, determines whether or not a predetermined synchronization word exists at a predetermined position in the frame (s610), and exists In this case, it is assumed that synchronization is established, the bit string portion for each frame is specified, and the input bit string L is output to the mode determination unit 620. At that time, the synchronization word and the dummy bit in the frame may be deleted and output to the mode determination unit 620. If there is no synchronization word, error processing is performed (s661). As error processing, communication may be terminated, or synchronization processing with the encoding apparatus 500 may be performed again.

<Mode determination unit 620>
The mode determination unit 620 determines a combination of the enhancement layer code (the low-frequency reinforcement bit string L1 and the high-frequency emphasis bit string L2) and the lossless bit string LL using the mode information (s620).

  When the mode information represents the mode δ (s620D), the mode determination unit 620 outputs the mode information, the lossless bit string LL, the low-frequency reinforcement bit string L1, and the high-frequency emphasized bit string L2. When the mode information represents the mode γ (s620C), the mode information, the lossless bit string LL, and the high frequency emphasized bit string L2 are output. When the mode information represents the mode β (s620B), the mode information, the lossless bit string LL, and the low-frequency reinforcement bit string L1 are output. When the mode information is not β, γ, or δ (s620B), that is, when it represents the mode α, the mode information and the lossless bit string LL are output.

The output lossless bit string LL is input to the lossless decoding unit 230, and other than the lossless bit string LL is input to the scalable decoding unit 650.
The lossless decoding unit 230 decodes the input lossless bit string LL to generate a G. A 711-bit string L0 is obtained (s230A to s230D) and output to the scalable decoding unit 650.

<Scalable Decoding Unit 650>
Scalable decoding section 650 receives mode information and a bit string (L0, L0 and L1, L0 and L2, or L0, L1 and L2) corresponding to each mode information, and the mode indicated by the input mode information Decoding is performed by selecting a desired mode from among the decodable modes specified from α to δ (at least one of modes A to D including mode A) (s655A to s655D), and obtained by decoding. Output a sequence of audio samples. The output signal obtained by decoding is the same as that of the scalable decoding unit 250 of the first embodiment.

<Effect>
By adopting such a configuration, the same effect as in the second embodiment can be obtained. Furthermore, in the transmission band (64 kbit / s) of the currently popular telephone line, G. It is possible to transmit higher quality voice than 711 in real time.

  G. 711.1 is a real-time transmission and G.G. In order to obtain a sound quality higher than 711, it is necessary to ensure a transmission band of 80 kbit / s or more. A sound quality higher than 711 could not be obtained. On the other hand, G. In 711.0, the average data rate can be reduced, so that as many bits as possible are transmitted and the decoding side uses only the obtained bits to decode, or the decoding side temporarily accumulates on the decoding side and then uses the decoding. That is, the best effort type has advantages. However, in the case of real-time transmission at a fixed rate, a line that secures a transmission band exceeding the worst value (65.6 kbit / s) of the data rate is required, and the prevailing transmission band is used with a line of 64 kbit / s. I couldn't. Such a problem can be solved by the configuration of this embodiment.

  In the third embodiment, when the code amount for one frame of the lossless bit string LL exceeds ((J−K) × 8−m) bits, a bit string for performing real-time communication at 64 kbit / s or less is generated. Can not do it. In this embodiment, when the code amount for one frame of the lossless bit string LL exceeds ((J−K) × 8−m) bits, such as when there is no effect of lossless encoding, the bits of some samples are converted. By omitting, the information is transmitted in the specified use transmission band.

<Encoding device 700>
A coding apparatus 700 according to the fourth embodiment will be described with reference to FIGS. 21 and 22. Only parts different from the encoding apparatus 500 will be described. The encoding device 700 further includes an abbreviated code generation unit 760 in addition to the scalable encoding unit 110 and the lossless encoding unit 120, and the processing contents of the bit string selection unit 730 and the synchronization word insertion unit 750 are different.

<Bit string selection unit 730>
The mode information generation unit 740 of the bit string selection unit 730 generates mode information indicating a combination selected by the bit string selection unit 730 or mode information indicating a generation rule when an omitted code generation unit 760 described later generates an omitted bit string. (S730).

  The bit string selection unit 730 determines whether the code amount converted per unit time of the lossless bit string LL is 64 kbit / s or less (s731). If the code amount is 64 kbit / s or less, the same processing as the bit string selection unit 530 is performed. (S332 to s539), the bit string selection unit 530 outputs mode information indicating the combination of the selected bit string and the bit string selection unit 730 to the synchronization word insertion unit 550.

  When the bit rate is greater than 64 kbit / s (that is, when the code amount converted per unit time of the lossless bit sequence LL is larger than the used transmission band), the bit sequence selection unit 730 generates an abbreviated bit sequence at the abbreviated code generation unit 760. The mode information indicating the generation rule is generated (s733). The bit string selection unit 530 outputs mode information indicating a generation rule for generating an abbreviated bit string to the synchronization word insertion unit 550.

  Note that the mode information is information indicating a combination selected by the bit string selection unit 730 or a generation rule when generating an abbreviated bit string by an abbreviated code generation unit 760 described later. The number of bits will increase.

<Synchronous word insertion unit 750>
(In the case of modes α to δ)
When mode information indicating the selected bit string and the combination selected by the bit string selection unit 730 is input, the synchronization word insertion unit 750 inserts a predetermined synchronization word at a predetermined position in the frame (s550). Furthermore, the synchronization word insertion unit 750 outputs the output bit string L including the synchronization word, the mode information output by the bit string selection unit 730, and the selected bit string. When the selected bit string is less than ((J−K) × 8−m)) bits, the synchronization word inserting unit 750 adds dummy bits to the remaining bits so that the output bit string L becomes J bytes. (For example, 0 or 1) is stored.
(Modes 1 to 8)
When mode information indicating a generation rule for generating an abbreviated bit string is input, the synchronization word insertion unit 750 creates a frame and inserts a predetermined synchronization word at a predetermined position in the frame (s750). ). For the frame in which the synchronization word is inserted, mode information indicating a generation rule for generating an abbreviated bit string is stored and output to the abbreviated code generation unit 760.

<Omitted Code Generation Unit 760>
The abbreviated code generation unit 760 includes a synchronization word, mode information indicating a generation rule when generating the abbreviated bit string L00, and an output from the scalable encoding unit 110. The 711 bit string L0 is input.

  The abbreviated code generation unit 760 uses G. A portion having little auditory influence is omitted from the 711 bit string L0, and an abbreviated bit string L00 in which the code amount converted per unit time is equal to or less than the use transmission band is generated (s760). For example, the part with a little auditory influence is G.I. It is the least significant bit in 1 byte (8 bits) indicating each sample in the 711-bit string L0.

  The abbreviated code generation unit 760 uses G. It is obtained by omitting the least significant bit out of the bits corresponding to each sample of the 711 bit string L0 by the number of bits exceeding ((J−K) × 8−m) bits ((J−K) × 8−m). An abbreviated bit string L00 of bits is generated (s760). Then, the abbreviated code generation unit 760 outputs the abbreviated bit string L00 and the mode information indicating the generation rule when generating the abbreviated bit string L00 as the output bit string L00.

  Note that the abbreviated bit string L00 to G.I. When generating the 711 bit string L0 ', 0 or 1 is inserted into the omitted bits.

  G. When the 711 bit string L0 cannot be compressed to ((J−K) × 8−m) bits or less, G. The acoustic signal corresponding to the 711-bit string L0 has a large amplitude and is a noisy signal that cannot be predicted. Since the least significant bit of the code representing the amplitude of such a signal has very little influence on the voice quality, the sound quality degradation caused by using the omitted bit string L00 is extremely small.

  G. When generating the abbreviated bit string L00 from the 711 bit string L0, it may be determined in advance which position in the frame the least significant bit of the sample is omitted, or the position of the sample in the frame. Whether to omit the least significant bit may be prepared by selecting a plurality of options and selecting one with less deterioration.

  When selecting from the choices, the G.D. 711 which is the amplitude value of the 711 bit string L0 'and the output of the scalable encoding unit 110. It is desirable to select the one having the smallest difference in amplitude value of the 711 bit string L0. Therefore, prior to transmission, the abbreviated bit string L00 generated in the abbreviated code generation unit 760 is reconstructed. The smallest option may be obtained by obtaining the 711-bit string L0 'and calculating the difference from L0. Note that the amplitude value at this time may be a difference in numerical values with logarithm. Since the difference in the waveform that is actually reproduced is the difference in the region converted from logarithm to linearity, the deformation of the lower 1 bit increases the difference in the sample having a large amplitude. However, auditory deterioration can be dealt with by comparing the numerical values in the logarithmic region because the deterioration is relatively small even if the difference is large for samples with large amplitude.

  As a method for generating the abbreviated bit string L00 (hereinafter referred to as “generation rule”), as a method for determining in advance or including it in a plurality of options, for example, the following modes 1 to 8 may be mentioned. That is, one of these is determined in advance, or some or all of these are included in the options.

  Modes 1-4 are G. This is a method of omitting the least significant bit of consecutive samples in the 711-bit string L0, and a schematic diagram is shown in FIG. In mode 1, G. The least significant 1 bit of (K × 8−m) samples from the beginning of the 711-bit string L0 is omitted. The decoding device 800 inserts 0 into the omitted bits. In mode 2, G. The least significant bit of (K × 8-m) samples from the end of the 711-bit string L0 is omitted. The decoding device 800 inserts 0 into the omitted bits. In mode 3, G. The least significant 1 bit of (K × 8−m) samples from the beginning of the 711-bit string L0 is omitted. The decoding device 800 inserts 1 into the omitted bits. In mode 4, G. The least significant bit of (K × 8-m) samples from the end of the 711-bit string L0 is omitted. The decoding device 800 inserts 1 into the omitted bits.

  Modes 5 to 8 are G. This is a method of omitting the least significant bit of the skipped sample in the 711-bit string L0, and FIG. 24 shows a schematic diagram. In mode 5, G. In the 711-bit string L0, the least significant bit of (K × 8-m) samples that are even-numbered from the beginning is omitted. The decoding device 800 inserts 0 into the omitted bits. In mode 6, G. The least significant 1 bit of the (K × 8−m) samples that are odd-numbered from the beginning of the 711-bit string L0 is omitted. The decoding device 800 inserts 0 into the omitted bits. In mode 7, G. In the 711-bit string L0, the least significant bit of (K × 8-m) samples that are even-numbered from the beginning is omitted. The decoding device 800 inserts 1 into the omitted bits. In mode 8, G. The least significant 1 bit of the (K × 8−m) samples that are odd-numbered from the beginning of the 711-bit string L0 is omitted. The decoding device 800 inserts 1 into the omitted bits.

<Decoding device 800>
A decoding apparatus 800 according to the fourth embodiment will be described with reference to FIGS. 25 and 26. Only parts different from the decoding device 600 will be described. Decoding apparatus 800 further includes an insertion unit 840 in addition to synchronization word determination unit 610 and lossless decoding unit 230, and processing contents of mode determination unit 820 and scalable decoding unit 850 are different.

<Mode determination unit 820>
The mode determination unit 820 receives the input bit string L or L00 that is the output of the synchronization word determination unit 610, and uses the mode information included in the input bit string to omit the omission rule or the enhancement layer code (low band code) in the omitted bit string L00. The combination of the reinforcement bit string L1, the high frequency emphasis bit string L2), and the lossless bit string LL is determined (s820). Here, the omission rule for the omission bit string L00 is a rule that has a one-to-one correspondence with the generation rule when the omission bit string L00 is generated in the encoding apparatus 700.

When the mode information indicates the modes α to δ (s820), the same processing as that of the decoding device 600 is performed (s620 to s655).
When the mode information indicates an omission rule in the omitted bit string L00 (s820), that is, when the mode information indicates modes 1 to 8, the mode information and the omitted bit string L00 are output to the insertion unit 840.

<Insert part 840>
The insertion unit 840 receives the mode information and the abbreviated bit string L00, and inserts a predetermined value in the omitted part as described in the abbreviated code generation unit 760 (see FIGS. 23 and 24). 711 bit string L0 ′ is set (s840).

<Scalable Decoding Unit 850>
When the mode information indicates the modes α to δ (s820), the processing content of the scalable decoding unit 850 is the same as that of the scalable decoding unit 650 of the third embodiment.
When the mode information is modes 1 to 8, the G.G. The output signal is obtained by decoding the 711 bit string L0 ′ (mode A decoding). Then, scalable decoding section 850 outputs an output signal obtained by decoding (telephone band voice sample sequence Y ′ (1), Y ′ (2),..., Y ′ (N / 2) having a sampling frequency of 8 kHz). Is output.

<Effect>
By adopting such a configuration, the same effect as in the third embodiment can be obtained. Furthermore, when there is an effect of lossless compression, an output signal can be obtained using any of modes α to δ, and when there is no effect of lossless compression, an output signal can be obtained. There is an effect that information can be transmitted in the set transmission band.

[Modification 1]
<Encoder 700A>
An encoding apparatus 700A according to the first modification of the fourth embodiment will be described with reference to FIGS. 21 and 22. Only parts different from the encoding apparatus 700 will be described. The encoding apparatus 700A is different in processing contents of the mode information generation unit 740A in the bit string selection unit 730.

<Mode information generation unit 740A>
The mode information generation unit 740A sets the mode information indicating the combinations (modes α to δ) selected by the bit string selection unit 730 or the generation rules (modes 1 to 8) when the omitted code generation unit generates the omitted bit string L00. The mode information is generated, and the mode information indicating the generation rules (modes 1 to 8) is generated so that the code amount is smaller than the mode information indicating the combination (modes α to δ) (s537 to s539, s733). That is, the mode information is a variable length code as shown in FIGS.

  In the modes α to δ, the G. In this mode, an auditory quality of 711 or higher can be obtained. On the other hand, in modes 1 to 8, the more bits allocated to mode information, the more lower bits are omitted. Compared to H.711, hearing quality is worse. Therefore, G. The number of bits of mode information representing modes 1 to 8, which is a mode in which the lower bits of the 711 bit string are omitted, is G.711. The mode information is variable-length encoded so as to be smaller than the number of bits of the mode information representing the modes α to δ which are modes in which the lower bits of the 711-bit string are not omitted. Thereby, the influence on the auditory quality of omitting the lower bits can be reduced.

  FIG. Three modes 1 to 3 are prepared as modes in which the lower bits of the 711 bit string L0 are omitted. This is an example of mode and mode information when a mode is selected from seven types including four types of modes α to δ as modes in which the lower bits of the 711 bit string L0 are not omitted. G. 2-bit codes “00”, “01”, and “10” are assigned as mode information to modes 1, 2, and 3, respectively, in which the lower bits of the 711-bit string L0 are omitted. On the other hand, G. 4-bit codes “1100”, “1101”, “1110”, and “1111” are assigned as mode information to the modes α, β, γ, and δ, which are modes in which the lower bits of the 711-bit string are not omitted.

  FIG. Eight modes 1 to 8 are prepared as modes in which the lower bit L0 of the 711-bit string is omitted. This is an example of mode and mode information when a mode is selected from 12 types including four types of modes α to δ as modes in which the lower bit L0 of the 711-bit string is not omitted.

  G. In modes 1 to 7, which are modes in which the lower bits of the 711-bit string L0 are omitted, 3-bit codes “000”, “001”, “010”, “011”, “100”, “101”, “110” "Is assigned as mode information, and a 4-bit code" 1110 "is assigned to mode 8 as mode information. On the other hand, G. 6-bit codes “111100”, “111101”, “111110”, and “111111” are assigned as mode information to the modes α, β, γ, and δ, which are modes in which the lower bits of the 711-bit string are not omitted.

<Decoding device 800>
The decoding device 800 according to the first modification of the fourth embodiment performs the same process as the first embodiment with reference to FIGS. 25 and 26. However, the mode information indicating the omission rule (modes 1 to 8) in the abbreviated bit string is different in that the code amount is smaller than the mode information indicating the combination of the enhancement layer code and the lossless bit string (modes α to δ). This will be specifically described below.

  In the mode determination unit 820 in the decoding apparatus 800, the G.P. The number of bits of mode information representing modes 1 to 8, which is a mode in which the lower bits of the 711 bit string are omitted, is G.711. The mode information represented by a variable length code smaller than the number of bits of the mode information representing the modes α to δ, which is a mode in which the lower bits of the 711 bit string are not omitted, is decoded.

  In the case of FIG. 27, the mode determination unit 820 determines that each of the mode information is 2-bit code “00”, “01”, “10”, which is mode 1, 2, 3, respectively. Are 4-bit codes “1100”, “1101”, “1110”, and “1111”, it is determined that the modes are α, β, γ, and δ, respectively.

  In the case of the example of FIG. 28, the mode determination unit 820 has the mode information of 3-bit codes “000”, “001”, “010”, “011”, “100”, “101”, “110”. Are determined to be mode 1, 2, 3, 4, 5, 6 respectively, and if the mode information is a 4-bit code “1110”, it is determined to be mode 8 and the mode information is 6 bits. When the codes are “111100”, “111101”, “111110”, and “111111”, it is determined that the modes are α, β, γ, and δ, respectively.

<Effect>
By adopting such a configuration, the same effect as in the fourth embodiment can be obtained. Further, the mode information is variable length encoded so that the number of bits of the mode information representing the modes 1 to 8 is smaller than the number of bits of the mode information representing the modes α to δ, thereby reducing the auditory quality without the lower bits. There is an effect that the influence of can be reduced.

[Modification 2]
<Encoder 700B>
An encoding apparatus 700B according to the second modification of the fourth embodiment will be described with reference to FIGS. 21 and 22. Only parts different from the encoding apparatus 700 will be described. In the encoding device 700B, the processing contents of the synchronization word insertion unit 750B and the omitted code generation unit 760B are different.

<Synchronous word insertion unit 750B>
When the synchronization word insertion unit 750B receives mode information indicating a generation rule (modes 1 to 8) when the abbreviated code generation unit generates the abbreviated bit string L00 (that is, a code obtained by converting the lossless bit string LL per unit time) When the amount is larger than the used transmission band), a specific synchronization word is inserted at a predetermined position in the frame (s750B). The specific synchronization word means a synchronization word indicating a generation rule when the omitted code generation unit 760 generates the omitted bit string L00. That is, mode information indicating a generation rule when generating the abbreviated bit string L00 is included in a specific synchronization word. The synchronization word insertion unit 750B inserts a specific synchronization word into an empty frame and outputs it to the abbreviated code generation unit 760.

  When the synchronization word insertion unit 750B receives mode information indicating the combination (modes α to δ) selected by the bit sequence selection unit 730 from the mode information generation unit 740 (that is, the code amount obtained by converting the lossless bit sequence LL per unit time) Is a predetermined synchronization word (hereinafter simply referred to as “a”) when the code amount obtained by converting the lossless bit string LL per unit time is equal to or less than the use transmission band at a predetermined position in the frame. Is inserted (s550). Note that the predetermined synchronization word is one synchronization word corresponding to all of a plurality of types of combinations of the extended layer code (low band reinforcement bit string L1 and high band emphasis bit string L2) and lossless bit string LL. The synchronization word insertion unit 750B outputs the output bit string L of the coding apparatus 700 including the synchronization word, the mode information, and the selected bit string.

<Omission Code Generation Unit 760B>
The abbreviated code generation unit 760 </ b> B receives the G.D. signal received from the scalable encoding unit 110 in the remaining portion of the frame in which the specific synchronization word is inserted. 711 bit string L0 is stored; An audible bit sequence in which the code amount converted per unit time is equal to or less than the used transmission band is generated by omitting a portion having a small acoustic influence from the 711 bit sequence (s760B).

  In the second modification, for the same purpose as in the first modification, a synchronization word indicating a generation rule (modes 1 to 8) when the omitted code generation unit generates the omitted bit string L00 is generated.

  A synchronization word used when mode information indicating the combination (modes α to δ) selected by the bit string selection unit 730 is received from the mode information generation unit 740, and a generation rule when the abbreviation code generation unit generates the abbreviated bit string L00 ( The synchronization word used when the mode information indicating the modes 1 to 8) is received is expressed using different bit patterns.

  As shown in FIG. 29, FIG. 30 and FIG. 31, when the mode information indicating the generation rule (modes 1 to 8) is received, the number of bits omitted for the mode information can be substantially 0 bits. The effect on the auditory quality of omitting the lower bits can be reduced. That is, when mode information indicating a generation rule (modes 1 to 8) is received, K = 1 and m = 0 are set, and the mode information is embedded in the synchronization word to be 8 bits including the synchronization word, G. A bit for designating a mode in which the lower bits of the 711 bit string L0 are omitted is substantially 0 bits. That is, “predetermined sync word” (mode information is after the sync word) and “specific sync word” (there is no mode information after the sync word. In this case, the abbreviated code generation unit 760B lowers by a predetermined 8 samples 2 types of synchronization words are used.

  FIG. As a mode in which the lower bits of the 711 bit string are omitted, only one mode 1 is prepared. This is an example of mode and mode information when a mode is selected from five types including four types of modes α to δ as modes in which the lower bits of the 711-bit string are not omitted. G. Each of the modes α, β, γ, and δ, which is a mode in which the lower bits of the 711-bit sequence are not omitted, is assigned “11111111” that is an 8-bit “predetermined synchronization word” as a synchronization word, and a 2-bit code “ Any one of “00”, “01”, “10”, and “11” is assigned as mode information. G. In mode 1, which is a mode in which the lower bits of the 711-bit string are omitted, 8-bit “specific synchronization word” “11111110” is assigned as a synchronization word, and mode information is not assigned.

  FIG. Three modes 1 to 3 are prepared as modes in which the lower bits of the 711 bit string are omitted. This is an example of mode and mode information when a mode is selected from seven types including four types of modes α to δ as modes in which the lower bits of the 711-bit string are not omitted. G. Each of the modes α, β, γ, and δ, which is a mode in which the lower bits of the 711-bit sequence are not omitted, is assigned “11111111” that is an 8-bit “predetermined synchronization word” as a synchronization word, and a 2-bit code “ Any one of “00”, “01”, “10”, and “11” is assigned as mode information. G. In each of modes 1 to 3, which are modes in which the lower bits of the 711-bit string are omitted, 8-bit “specific synchronization words” “11111100”, “11111101”, and “1111110” are assigned as synchronization words, and the mode Information is not assigned.

<Decoding device 800B>
A decoding apparatus 800B according to the second modification of the fourth embodiment will be described with reference to FIGS. 25 and 26. Only parts different from the decoding device 800 will be described. In the decoding device 800B, the processing contents of the synchronization word determination unit 810B and the mode determination unit 820B are different.

<Synchronous word determination unit 810B>
The synchronization word determination unit 810B determines whether or not there is a synchronization word at a predetermined position in the frame (s810), and whether the synchronization word is a specific synchronization word indicating an abbreviated rule or an extended hierarchy It is determined whether the code (low band reinforcement bit string L1 and high band emphasizing bit string L2) or one predetermined synchronization word corresponding to all of a plurality of combinations of the lossless bit string LL is determined. When the synchronization word is a predetermined synchronization word, it is considered that there is mode information indicating a combination of the extended hierarchical code and the lossless bit string LL (modes α to δ), and therefore the input bit string is output to the mode determination unit 820. (S820B).

  The synchronization determination unit 800B stores a “predetermined synchronization word” and a “specific synchronization word” indicating an abbreviation rule in advance in a storage unit (not shown), and the synchronization word included in the input bit string is a specific synchronization word Is determined which omission rule the specific synchronization word indicates, and outputs the corresponding mode information and input bit string to the mode determination unit 820.

  When the mode determination unit 820 receives an input bit string including a predetermined synchronization word, the mode information indicates the modes α to δ (s820), and thus performs the same processing as that of the decoding device 600 (s620 to s655).

  When the mode determination unit 820 receives an input bit string including a specific synchronization word, the mode information indicates the omission rule in the omitted bit string L00 (s820B), and outputs the mode information and the omitted bit string L00 to the inserting unit 840. Note that the synchronization word determination unit 810B and the mode determination unit 820 may be integrated. That is, when the synchronization word determination unit 810B is provided inside the mode determination unit 820 and the synchronization word determination unit 810B receives an input bit string including a specific synchronization word, the mode information and the omitted bit string L00 are directly input to the insertion unit 840. It may be output.

  In the case of the example of FIG. 30, it is determined whether or not the synchronization word is “11111111”. If the synchronization word is “11111111”, the mode information is “00”, “01”, “10”, It is determined which mode is α to δ depending on which one is “11”, and when the synchronization word is “11111110”, it is determined that the mode 1 is selected.

  In the case of the example of FIG. 31, it is determined whether or not the synchronization word is “11111111”. If the synchronization word is “11111111”, the mode information is “00”, “01”, “10”, It is determined which of the modes α to δ depending on which of “11”, mode 1 when the synchronization word is “11111100”, mode 2 when the synchronization word is “11111101”, When the synchronization word is “11111110”, it is determined that the mode 3 is set.

<Effect>
By adopting such a configuration, the same effect as in the fourth embodiment can be obtained. Furthermore, since the mode information indicating the generation rule (modes 1 to 8) when generating the omitted bit string in the abbreviated code generation unit is included in the synchronization word, the number of bits of the mode information representing modes 1 to 8 is substantially 0 bits. And the effect of omitting the lower bits on the auditory quality can be reduced.

<Simulation results>
FIG. 32 shows the number of bytes per frame after compression when encoding is performed using the encoding apparatus 100 according to the first embodiment. Since the code amount of 16 kbit / s or more can be reduced in most frames (when converted per unit time), the low-frequency reinforcement bit string L1 or the high-frequency emphasis bit string L2 can be transmitted together. It is possible to transmit a higher quality signal than when only the 711 bit string L0 is transmitted.

<Other variations>
In the case of transmission using IP packets, an IP packet configuration as shown in FIG. 33 may be used. That is, the bit information LL, L1, and L2 may be included as an IP payload for one IP header, and the mode information may be included in the IP header for transmission. In this case, a fixed-length code representing the mode information of all modes with the same number of bits may be used, or a variable-length code as shown in FIGS. 27 and 28 may be used. Also, as shown in FIG. 12B, the bit strings LL, L1, and L2 may be different IP packets, and the mode information may be included in the header of the IP packet including the LL.

  Further, as scalable coding and lossless coding, G.I. 711.1 and G.I. The encoding may be other than 711.0, and MPEG-4SVC and MPEG-4ALS (acoustic encoding), which are encodings of different media (video encoding), may be combined.

<Program>
The computer may function as the above-described encoding device and decoding device. In this case, each process of a program for causing a computer to function as a target device (a device having the functional configuration shown in the drawings in various embodiments) or a processing procedure (shown in each embodiment) is processed by the computer. A program to be executed by the computer may be downloaded from a recording medium such as a CD-ROM, a magnetic disk, or a semiconductor storage device or via a communication line into the computer, and the program may be executed.

100, 300, 500, 700, 700A, 700B Encoder 200, 600, 800, 800B Decoder 110 Scalable encoder 120 Lossless encoder 230 Lossless decoder 250, 650, 850 Scalable decoders 330, 530, 730 Bit string selection unit 533, 740, 740A Mode information generation unit 550, 750, 750B Sync word insertion unit 610, 810B Sync word determination unit 620, 820 Mode determination unit 760, 760B Abbreviated code generation unit 840 Insertion unit

Claims (14)

A scalable encoding unit that performs scalable encoding of a digital input signal and generates a base layer code and one or more enhancement layer codes;
A lossless encoding unit for lossless encoding the base layer code to generate a lossless code;
Code selection for selecting a combination having a code amount per unit time that is equal to or less than a use transmission band and having the largest code amount per unit time from a plurality of types of combinations of the lossless code and the enhancement layer code And
When the amount of code per unit time of the lossless code is larger than the use transmission band, the portion of the base layer code that is not affected acoustically or visually is omitted, and the code amount per unit time is equal to or less than the use transmission band. An ellipsis generator for generating an ellipsis,
Mode information indicating a combination selected by the code selection unit, or a mode information generation unit that generates mode information indicating a generation rule when generating the abbreviated code by the abbreviated code generation unit,
The mode information generated by the mode information generation unit has less code amount in the mode information indicating the generation rule than in the mode information indicating the combination.
An encoding apparatus characterized by that. A scalable encoding unit that performs scalable encoding of a digital input signal and generates a base layer code and one or more enhancement layer codes;
A lossless encoding unit for lossless encoding the base layer code to generate a lossless code;
Code selection for selecting a combination having a code amount per unit time that is equal to or less than a use transmission band and having the largest code amount per unit time from a plurality of types of combinations of the lossless code and the enhancement layer code And
When the amount of code per unit time of the lossless code is larger than the use transmission band, the portion of the base layer code that is not affected acoustically or visually is omitted, and the code amount per unit time is equal to or less than the use transmission band. An abbreviated code generation unit that generates an abbreviated code
When the code amount per unit time of the lossless code is larger than the use transmission band, a specific synchronization word indicating a generation rule when generating the abbreviated code in the abbreviated code generation unit, otherwise, A synchronization word insertion unit that inserts a synchronization word common to cases other than the above at a predetermined position in the frame,
An encoding apparatus characterized by that. A scalable encoding unit that performs scalable encoding of a digital input signal and generates a base layer code and one or more enhancement layer codes;
A lossless encoding unit for lossless encoding the base layer code to generate a lossless code;
Code selection for selecting a combination having a code amount per unit time that is equal to or less than a use transmission band and having the largest code amount per unit time from a plurality of types of combinations of the lossless code and the enhancement layer code And
The enhancement layer code includes a low-frequency reinforcement code and a high-frequency emphasis code, and the high-frequency emphasis code includes information corresponding to a gain when emphasizing a high frequency,
In the code selection unit, when selecting a combination of the lossless code and the low-frequency reinforcement code, or a combination of the lossless code and the high-frequency emphasis code, if the gain is equal to or less than a threshold, the lossless code Selecting a combination of the low-frequency reinforcement codes;
An encoding apparatus characterized by that. The input code is a combination of one or more enhancement layer codes of a predetermined scalable coding and a lossless code obtained by lossless coding of the base layer code of the scalable coding, or auditory or visual from the base layer code. In addition, the input code includes mode information indicating a combination of the enhancement layer code and the lossless code, or mode information indicating an abbreviation rule in the abbreviated code. A mode determination unit that determines an abbreviation rule in the abbreviation code, or a combination of the enhancement layer code and the lossless code, using the mode information;
When the abbreviation code is included in the input code, an insertion unit having a base layer code that is obtained by inserting a predetermined value in a portion where the abbreviation code is omitted;
When the input code includes a combination of the enhancement layer code and the lossless code, a lossless decoding unit that decodes the lossless code to obtain a base layer code;
A scalable decoding unit that obtains an output signal by decoding the base layer code obtained by the insertion unit, or the base layer code obtained by the lossless decoding unit and the enhancement layer code;
A decoding device. The decoding device according to claim 4, wherein
The mode information indicating the omission rule in the abbreviation code has a smaller code amount than the mode information indicating the combination of the enhancement layer code and the lossless code.
A decoding device characterized by the above. The decoding device according to claim 4, wherein
The input code has a synchronization word at a predetermined position in a frame, the synchronization word is a specific synchronization word indicating an abbreviation rule in the abbreviation code, and a plurality of combinations of the enhancement layer code and the lossless code One of the predetermined synchronization words corresponding to all of
It is determined which sync word is a sync word existing at a predetermined position in a frame, and when the sync word is a specific sync word indicating the omission rule, which sync word is A synchronization word determination unit that determines whether the omission rule is shown.
A decoding device characterized by the above. A scalable encoding step of scalable encoding a digital input signal to generate a base layer code and one or more enhancement layer codes;
A lossless encoding step of lossless encoding the base layer code to generate a lossless code;
Code selection for selecting a combination having a code amount per unit time that is equal to or less than a use transmission band and having the largest code amount per unit time from a plurality of types of combinations of the lossless code and the enhancement layer code Steps,
When the amount of code per unit time of the lossless code is larger than the use transmission band, the portion of the base layer code that is not affected acoustically or visually is omitted, and the code amount per unit time is equal to or less than the use transmission band. An abbreviated code generation step for generating an abbreviated code
A mode information generation step for generating mode information indicating a combination selected in the code selection step, or mode information indicating a generation rule when generating the abbreviated code in the abbreviated code generation step;
The mode information indicating the generation rule has a smaller code amount than the mode information indicating the combination.
An encoding method characterized by the above. A scalable encoding step of scalable encoding a digital input signal to generate a base layer code and one or more enhancement layer codes;
A lossless encoding step of lossless encoding the base layer code to generate a lossless code;
Code selection for selecting a combination having a code amount per unit time that is equal to or less than a use transmission band and having the largest code amount per unit time from a plurality of types of combinations of the lossless code and the enhancement layer code Steps,
When the amount of code per unit time of the lossless code is larger than the use transmission band, the portion of the base layer code that is not affected acoustically or visually is omitted, and the code amount per unit time is equal to or less than the use transmission band. An abbreviated code generation step for generating an abbreviated code
When the code amount per unit time of the lossless code is larger than the use transmission band, a specific synchronization word indicating a generation rule when generating the abbreviated code in the abbreviated code generation step, otherwise, Including a synchronization word insertion step of inserting a synchronization word common to cases other than the above at a predetermined position in the frame,
An encoding method characterized by the above. A scalable encoding step of scalable encoding a digital input signal to generate a base layer code and one or more enhancement layer codes;
A lossless encoding step of lossless encoding the base layer code to generate a lossless code;
Code selection for selecting a combination having a code amount per unit time that is equal to or less than a use transmission band and having the largest code amount per unit time from a plurality of types of combinations of the lossless code and the enhancement layer code Including steps,
The enhancement layer code includes a low-frequency reinforcement code and a high-frequency emphasis code, and the high-frequency emphasis code includes information corresponding to a gain when emphasizing a high frequency,
In the code selection step, when the combination of the lossless code and the low-frequency reinforcement code, or the combination of the lossless code and the high-frequency emphasis code is selected, if the gain is equal to or less than a threshold, the lossless code Selecting a combination of the low-frequency reinforcement codes;
An encoding method characterized by the above. The input code is a combination of one or more enhancement layer codes of a predetermined scalable coding and a lossless code obtained by lossless coding of the base layer code of the scalable coding, or auditory or visual from the base layer code. In addition, the input code includes mode information indicating a combination of the enhancement layer code and the lossless code, or mode information indicating an abbreviation rule in the abbreviated code. Including
A mode determination step of determining an abbreviation rule in the abbreviation code or a combination of the enhancement layer code and the lossless code using the mode information;
When the input code includes the abbreviation code, an insertion step in which a predetermined value is inserted into a portion where the abbreviation code is omitted is a base layer code;
If the input code includes a combination of the enhancement layer code and the lossless code, a lossless decoding step of decoding the lossless code to obtain a base layer code;
A scalable decoding step of obtaining an output signal by decoding the base layer code obtained in the insertion step, or the base layer code obtained in the lossless decoding step and the enhancement layer code;
A decryption method. The decoding method according to claim 10, wherein
The mode information indicating the omission rule in the abbreviation code has a smaller code amount than the mode information indicating the combination of the enhancement layer code and the lossless code.
A decoding method characterized by the above. The decoding method according to claim 10, wherein
The input code has a synchronization word at a predetermined position in a frame, the synchronization word is a specific synchronization word indicating an abbreviation rule in the abbreviation code, and a plurality of combinations of the enhancement layer code and the lossless code One of the predetermined sync words corresponding to all,
It is determined which sync word is a sync word existing at a predetermined position in a frame, and when the sync word is a specific sync word indicating the omission rule, which sync word is A synchronization word determination step of determining whether the omission rule is
A decoding method characterized by the above. Program for causing a computer to function as the sign KaSo location according to any one of claims 1 to 3. The program for functioning a computer as a decoding apparatus of any one of Claims 4-6.

Images