JPH08274726A - Method and device for encoding and decoding sound signal - Google Patents

Abstract

PURPOSE: To match the internal states of an encoding part and a decoding part with each other if a frame deficit occurs. CONSTITUTION: An encoding side performs ADPCM encoding, frames the encoded series (a), and sends the series, frame by frame at a high speed (b). A decoding side once detecting a deficit of a frame #3 informs the encoding side of that (c). Once receiving the information, the encoding side initializes the internal state of the encoding part at predetermined time T1 (a). The decoding side, after sending the information, initializes the internal state of the decoding part the sum of times T0 and T1 of transmission to the encoding side after the end of the deficit frame #3 of encoded frames (e).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is applied to, for example, a mobile communication system and obtains at least a part of a locally decoded signal from a sound signal such as a voice or a musical sound by a process equivalent to the decoding. It encodes using the internal state based on the decoded signal, transmits the encoded code for each frame, and the decoding side creates the same internal state as the encoding side from the encoded signal,
The present invention relates to a coding / decoding method for decoding using this, and more particularly, to a coding / decoding method and a coding device and a decoding device thereof when a frame that is not received on the decoding side occurs.

[0002]

2. Description of the Related Art A coding / decoding method of this kind can be used for general transmission of acoustic signals, but in the following description, a case will be described in which acoustic signals are ADPCM encoded and transmitted by mobile radio. In speech coding applied to a digital mobile radio system, generally, a code sequence is collected for a certain period of time to form a frame, and the frame is transmitted for each frame.

Mobile radio has a very poor transmission path condition, and reception and demodulation may become impossible. In this case, in voice coding applied to the digital mobile radio system, a state may occur in which the code sequence is lost in frame units due to fading, and in the digital mobile radio system, for control of the receiving device, For example, in order to monitor the reception level of the radio wave from another base station at the boundary of the radio wave arrival area of the base station, the audio frame cannot be decoded during that period, resulting in intermittent reception and frame loss.

When a frame loss occurs, nothing is input to the speech decoding apparatus. Therefore, unless a technique for compensating the frame loss is provided, a great deterioration in voice quality occurs. The quality deterioration when a frame loss occurs is the deterioration of the quality of the lost frame due to the lack of decoding of the lost frame and the internal state of the decoding at the start of decoding the frame next to the lost frame (next frame). However, it is due to the deterioration of quality after the next frame due to the difference from the encoded one.

Regarding the quality deterioration of the lost frame, by using the waveform interpolation processing technique and the code interpolation processing technique,
Attempts have been made to suppress auditory quality deterioration of decoded speech signals. The above-mentioned technique called waveform interpolation processing is a technique for predictively synthesizing a speech waveform of a lost frame from a decoded waveform before frame loss. For example, a decoded waveform of a previous frame is used as a decoded speech waveform of a missing frame. Is used. By using this technique, the decoded speech of the defective frame is reproduced in a pseudo manner, and the auditory quality deterioration is reduced.

The above-mentioned technique called code interpolation processing is a technique for performing speech decoding using a code sequence for a defective frame prepared in advance when a frame loss occurs. For example, when a frame loss occurs, the waveform of the previous frame is lost. A method is used in which a code sequence of a defective frame is obtained by repeating a past code sequence in a cycle. By using this technique, the decoded speech of the defective frame is reproduced in a pseudo manner, and the auditory quality deterioration is reduced.

After the next frame, noise occurs due to the fact that the internal state of decoding at the start of decoding of the next frame is different from the internal state of encoding of the same portion, and the quality of the decoded signal deteriorates. This quality deterioration occurs regardless of whether or not the above-described waveform interpolation processing technology or code interpolation processing technology is applied. FIG. 8 shows a conventional A to which the code interpolation processing technique is applied.
The structure of a DPCM voice encoding / decoding method is shown. On the encoding side 11, the input acoustic signal is input to the input terminal 12 and AD
The PCM encoder 13 converts the ADPCM code.

In the ADPCM coding unit 13, the difference between the input acoustic signal and the prediction signal from the adaptive prediction unit 15 is calculated by the difference circuit 14, and the difference signal is coded by the adaptive quantization unit 16 and output to the output terminal 17. It is output and the adaptive inverse quantizer 1
8 and is inversely quantized, and the inversely quantized differential signal is added to the prediction signal by the adder circuit 19 to locally obtain a decoded signal. The decoded signal and the inverse quantized output are adaptively predicted. It is input to the unit 15 and temporarily held in the internal state holding unit 21. In the adaptive prediction unit 15, the holding unit 2
While predicting and outputting the next prediction signal from the decoded signal and the dequantized signal (difference signal) before the plurality of samples held in 1 and these newly input signals,
A preferred quantization width is estimated, and the estimated quantization width is adjusted by the adaptive control unit 22 to the adaptive quantization unit 16 and the adaptive dequantization unit 1.
Set to 8.

In this way, the input acoustic signal is ADPCM.
After being encoded, the encoded code sequence is framed by the framing unit 23 as shown in FIG. 9a, and is sent to the transmission line 24 at high speed for each frame as shown in FIG. 9b.
On the decoding side 26, the coded code sequence received through the transmission path 24 passes through the frame loss detection section 27, and is further input to the adaptive inverse quantization section 29 of the ADPCM decoding section 36 through the changeover switch 28. In the adaptive dequantization unit 29, each input code is dequantized to decode the difference signal, and the difference signal is added to the prediction signal from the adaptive prediction unit 32 in the adder circuit 31 to obtain the decoded signal, That is, a synthetic acoustic signal is obtained and output to the output terminal 33. The decoded signal and the dequantized differential signal are input to the adaptive prediction unit 32,
While being temporarily held in the internal state holding unit 34,
The next prediction signal is predicted and output by a method similar to that on the encoding side 11 using the previous plurality of samples and the current input, and a preferable quantization width is estimated. This estimated quantization width is set in the adaptive inverse quantization unit 29 by the adaptive control unit 35.

At the decoding side 26, in the received code sequence,
For example, when the third frame # 3 is lost as shown in FIG. 9c, the received coded code sequence becomes the missing third frame # 3 as shown in FIG. 9d, and this is detected by the frame loss detection unit 27. Therefore, in the conventional code interpolation processing technique, in the portion where the third frame # 3 is to be received, the coded sequence switching control unit 38 causes the changeover switch 28
Is switched to the side of the lost frame coded code sequence generation unit 39, and the third frame from the lost frame coded code sequence generation unit 39
The frame # 3 'is inserted as a third frame as shown in FIG. 9e, and each frame is provided with an encoding code and is supplied to the adaptive inverse quantizer 29. In the adaptive dequantization unit 29, the coded code of each frame is expanded to fill the frame as shown in FIG. 9f and then dequantized to obtain a decoded code sequence.

For example, the reception coded code sequence is input to the lost frame coded code sequence generation unit 39, and the coded codes of a plurality of past frames are always stored therein. For example, the second frame # immediately before the lost frame # 3 is stored. 2 is used as interpolation code # 3 '. Alternatively, a silent coded code string may be stored in advance and output as the interpolation code # 3 '.

[0012]

As described above, when a frame loss occurs, the audible quality deterioration of the decoded acoustic signal is considerably suppressed by the code interpolation process and the waveform interpolation process in the lost frame. But you can
In the encoding / decoding method using the internal state, the internal state of the internal state holding unit 34 on the decoding side at the time of starting the decoding of the frame following the lost frame is # 3 ′, # 4 as shown in FIG. 9g. , # 5 '... and the internal state of the internal state holding unit 21 of the encoding side 11 are different from each other, and the problem that the quality of the decoded acoustic signal deteriorates after the next frame has not been solved.

An object of the present invention is to provide an acoustic signal coding / decoding method and apparatus capable of reducing the occurrence of waveform distortion occurring after a frame when a frame loss occurs.

[0014]

According to the encoding / decoding method of the invention of claim 1, when a frame loss is detected, the detection is notified to the encoding side, and after the notification is received by the encoding side, On the encoding side and the decoding side, respective internal states are initialized at the time points when the transmission encoding code sequence and the receiving decoding code sequence correspond.

According to the encoding / decoding method of the second aspect of the present invention, when a frame loss is detected, the lost frame is notified to the encoding side, and the encoding side and the decoding side each temporarily store the internal state thereof. After receiving the lost frame on the encoding side, the accumulated internal states of a predetermined frame before the lost frame are replaced with the respective internal states.

According to the encoding / decoding method of the third aspect of the present invention, this is because the frame in which the frame is lost in advance is
When it is known in advance that the coding side will stop sending the coded code sequence and the decoding side will stop receiving the received coded code sequence, and if this is known, the other side is notified and the coding is performed. Both the decoding side and the decoding side start temporary storage of the internal state, and at the same predetermined time point after the lost frame, the internal state used for encoding and decoding is stored in the accumulated internal state. Replace state.

In the invention of claim 4, the encoding and decoding of the inventions of claims 1 to 3 are ADPCM encoding and ADP, respectively.
7 illustrates CM decoding. In the invention of claim 5, claim 2 or 3
In the above invention, the internal state is accumulated at the end of each frame. In the encoding device according to the sixth aspect of the present invention, the loss detection notification is received, and the internal state holding unit is initialized at a predetermined timing after the reception.

In the encoding device according to the seventh aspect of the present invention, when the internal state is stored in the storage unit and the notification of the lost frame is received, the internal state holding unit is replaced with the stored internal state of a predetermined portion from the lost frame. To be done. In the decoding device according to the invention of claim 8, a frame loss is detected, the detection is notified to the encoding side, and the internal state holding unit is initialized at a predetermined time point after the notification.

In the decoding device of the ninth aspect of the present invention, when a missing frame is detected, the missing frame is notified to the coding side, the internal state is accumulated in the accumulating section, and the missing frame is detected after the missing frame is detected. The internal state holding unit is replaced with the accumulated internal state of a portion that is predetermined before the frame.

[0020]

FIG. 1 shows each embodiment of a coding / decoding method according to the invention of claim 1, an encoding device according to the invention of claim 6 and a decoding device according to the invention of claim 8, and these embodiments are When these inventions are applied to ADPCM encoding and decoding, the same reference numerals are given to the portions corresponding to FIG.

In this embodiment, the encoding side 11 is provided with a loss detection receiving section 41 for extracting a loss detection notification from the received signal from the decoding side 26, and the extracted loss detection notification is initialized by the initialization control section 42. Is supplied to the initialization control unit 4
2 indicates that the internal state of the internal state holding unit 21, the adaptive quantization unit 16, the adaptive dequantization unit 18, that is, the encoding unit 13 is initialized by the adaptive control unit 22 at a predetermined time point after the loss detection notification, that is, the code. It will be in the state it was in when the conversion started.

On the decoding side 26, when the frame loss detection section 27 detects a frame loss, the loss detection is notified to the encoding side 11 by the loss detection notification section 44.
The loss detection is also supplied to the initialization control unit 45, and at a predetermined time point after the loss detection notification, the internal state holding unit 34 and the adaptive dequantization unit 29, that is, the decoding unit, are transmitted via the adaptive control unit 35. The internal state of 36 is initialized to the state at the start of decoding.

Similar to the time chart shown in FIG.
An operation time chart when frame # 3 is lost is shown in correspondence with FIGS. 2 and 9. When the frame loss is detected, the loss detection notification unit 44 notifies the encoding side 11 of the loss detection, as indicated by an arrow 47 in FIG. 2, and this notification is sent by the loss detection receiving unit 41 to the encoding code. It is taken out in the middle of the fourth frame # 4 of the stream. Although the internal state may be initialized by the initialization control unit 42 immediately at the time of taking out, in this example, the time t ₁ (the time on the encoded code sequence) when the time T _{1 has} elapsed from the time of taking out the notification. ), The internal state is initialized.

On the other hand, on the decoding side 26, the frame loss detection is notified to the initialization control section 45.
The total of the time T ₀ from the end of the lost frame # 3 in 1 to the extraction of the loss detection notification (this is a fixed value that can be known in advance) and the time T ₁ from extraction to initialization, the decoded code string Time point t ₁ (time point on the decoding code sequence) after the start of the fourth frame # 4 in FIG. 2E, that is, time point t on the coding code sequence on the coding side 11.
_1, the initialization of the internal state of the decoding side 26 at time t ₁ on the corresponding decoding code sequence is performed.

In this embodiment, the interpolation for the missing frame is performed by code interpolation as in the case of FIG. In addition to this interpolation, the coding side 1
The encoding side 11 and the decoding side 26 are both initialized at the time point t ₁ on the encoded code string of ₁ and the same time point t ₁ on the decoding code sequence of the decoding side 26, and their internal states are forced. Since they are matched, the distortion of the reproduced acoustic signal due to the internal state mismatch between the encoding side 11 and the decoding side 26 is reduced. The initialization associated with the loss detection may be performed immediately after the encoding side 11 takes out the loss detection notification. The point is that after the lost frame, the time point on the encoding code sequence for initialization on the encoding side 11 and the time point on the decoding code sequence for initialization on the decoding side 26 may match.

Next, each embodiment of the encoding / decoding method according to the invention of claim 2, the encoding device according to the invention of claim 7 and the decoding device according to the invention of claim 9 is subjected to ADPCM encoding,
FIG. 3 shows a case where the present invention is applied to decoding, and the portions corresponding to those in FIG. On the encoding side 11, a lossy frame receiving section 49 is provided in place of the loss detection receiving section 41 in FIG. 1, the initialization control section 42 in FIG. 1 is omitted, and instead, the internal state holding section 21 An internal state storage unit 51 for continuously storing states (including the quantization width of the adaptive quantization unit 16) is provided, and when a notification of a lost frame is taken out, the internal state replacement control unit 52 causes The state accumulating unit 51 is controlled to read the internal state corresponding to the previously determined portion from the lost frame, and the internal state of the internal state holding unit 21, the adaptive quantizing unit 16, and the adaptive dequantizing unit 18 are read. Of each quantization width (also internal state), that is, the internal state of the encoding unit 13 is replaced.

On the decoding side 26, the loss detection notification unit 4
4 is provided with a missing frame notifying section 53 for notifying the detected missing frame, the initialization control section 45 is omitted, and instead, the internal state accumulating section 54 like the encoding side 11 is provided.
And an internal state replacement control unit 55 are provided, the internal state of the internal state holding unit 34 and the quantization width of the adaptive dequantization unit 29 are continuously accumulated at all times, and when the frame loss is detected, the internal state is detected. The replacement control unit 55 reads the internal state of the storage unit 54 in a predetermined portion before the lost frame, and the internal state of the internal state holding unit 34 and the quantization width of the adaptive dequantization unit 29, that is, the decoding unit 36. The internal state of is replaced.

An operation example in this case is shown in FIG. 4 corresponding to FIG. 2 and FIG. 4 in a time chart in the case where the third frame # 3 is lost, as in the case of FIG.
Only the parts different from FIG. 9 will be described. Encoding side 11
Each frame in the coded code sequence at (Fig. 4a)
4h shows the internal state (state of the holding unit 21 and quantization width) at the end of the encoding of the internal state # 1,
It is stored in the internal state storage unit 51 as # 2, .... Also on the decoding side 26, the internal state (state of the holding unit 34 and dequantization width) at the end of the decoding of each frame (FIG. 4f) in the decoding code sequence is internally as shown in FIG. The states # 1, # 2, ... Are stored in the internal state storage unit 54.

When the decoding side 26 detects a frame loss as shown in FIG. 4c, the lossy frame # 3 is notified to the encoding side 11 from the lossy frame notifying section 53. This is prior to the notified third frame # 3, which is taken out by the lost frame receiving unit 49 during the encoding of the fourth frame # 4,
In this example, the internal state of the immediately preceding frame # 2 is extracted from the internal state storage section 51 under the control of the internal state replacement control section 52, and the internal state of the internal state holding section 21 and the adaptive quantization section 16 and the adaptive dequantization section Each quantization width of the unit 18 is replaced with the beginning of the fifth frame of the coded code sequence, as indicated by the arrow 57 in FIG.

Also on the decoding side 26, the lost frame # 3
Is supplied to the internal state replacement control unit 55, and the internal state replacement control unit 55 reads the internal state of the frame # 2 before the lost frame # 3, in this example, the immediately preceding frame # 2, from the internal state storage unit 54, State of holding unit 34, inverse quantization unit 29
Is replaced with the beginning of the fifth frame # 5 of the decoded code sequence, as indicated by the arrow 58 in FIG.

As described above, the position on the coding code sequence of the internal state used for the replacement on the coding side 11, the time point on the coding code sequence of the replacement, and the internal used for the replacement on the decoding side 26. The position on the decoding code sequence of the state and the time point on the decoding code sequence of the replacement are made to coincide with each other, so that the internal state of the encoding side 11 and the internal state of the decoding side 26 after the frame loss. And match. In this case as well, since the ADPCM encodes each sample, it is not necessary to perform the internal state storage unit or replacement at the boundary of each frame. Therefore, the internal state storage cycle can be arbitrarily determined, and even in that case. As described above, the coded code sequence and the decoded code sequence may be matched.
However, it is preferable to accumulate at the end of the frame. Because it detects the presence or absence of a loss for each frame,
This is because if there is a frame loss, the internal state immediately before the lost frame can be used.

FIG. 5 shows a case where the embodiment of the coding / decoding method according to the invention of claim 3 is applied to ADPCM coding and decoding, and the same code is attached to the portion corresponding to FIG. For example, in a PHS (simple mobile phone), since the in-cell (zone) is searched after the reception state has deteriorated, the call disconnection time becomes relatively long. From this point, it is conceivable to periodically switch to the control channel and monitor the reception level to know the existing cell. In such a case, the frame that is lost on the mobile device side can be known before the arrival time of the frame. The present invention is applied when the defective frame can be known in advance.

The functional configuration of this embodiment is almost the same as that shown in FIG. 3, but when the notification of the missing frame is taken out at the encoding side 11, the internal state replacement control section 52 causes the internal state storage section 52 to carry out. The accumulation of the internal state for 51 is started, and the internal state accumulated in a predetermined portion of the encoded code sequence after the notification loss frame is encoded by the encoding unit 1.
Replace with the internal state in 3. On the decoding side 26, a frame which is lost due to disconnection of communication is sent from the control unit 61 for controlling the whole of the wireless telephone of the simple type portable telephone, which is provided with this decoding device, before the point of the frame. The missing frame notification unit 53 notifies the coding side, and the notification frame is also supplied to the internal state replacement control unit 55. The internal state replacement control unit 55 receives this notification and the decoding unit 3
The accumulation of the internal state of No. 6 in the internal state accumulation unit 54 is started,
After the notification frame, the accumulated internal state of the part of the predetermined decoding code sequence is taken out from the internal state accumulation section 54 and replaced with the internal state of the decoding section 36.

FIG. 4 shows the time chart of the operation example of FIG. 5 in the case where the lost frame is the third frame # 3.
Corresponding to is shown in FIG. In this example, the notification of the missing frame # 3 is one frame before that, that is, the second frame # 3.
2 is performed at the beginning of frame 2 and immediately before the loss of frame # 3 on the encoding side 11, that is, the internal state # 2 of the encoding unit 13 at the end of encoding of the second frame # 2 is accumulated as shown in FIG. Immediately after the lost frame # 3 stored in the unit 51, that is, at the beginning of the fourth frame # 4, the internal state # 2 stored in the storage unit 51 replaces the internal state of the encoding unit 13. On the decoding side 26, similarly to the encoding side 11, to the second frame # 2 immediately before the notification frame # 3, the internal state # 2 of the decoding unit 36 at the end of decoding is stored in the storage unit 54 as shown in FIG. 6i. Immediately after the notification (missing) frame # 3 (the beginning of the fourth frame # 4), the accumulated internal state # of the accumulation unit 54 is accumulated.
At 2, the internal state of the decoding unit 36 is replaced. In this way, the internal states of the encoding side 11 and the decoding side 12 are matched. In this example as well, the example shown in FIG. 6 is preferable, but the accumulation of the internal state may be performed at an appropriate time before the defective frame, and the replacement with the internal state may be performed at an appropriate time after the defective frame. The point is that the accumulation time point and the replacement time point may be matched with the position on the encoded code sequence and the position on the decoded code sequence, respectively.

Next, the method of the invention according to claim 2 for solving the problem of internal state inconsistency due to frame loss is CELP.
FIG. 7 shows the functional configuration of the embodiment applied to the encoding / decoding method.
The parts corresponding to those in FIG. In the CELP coding unit 71, the input acoustic signal from the input terminal 12 is sent to the LPC analysis filter (inverse filter) 72 and the LPC.
It is supplied to the analysis unit 73, the latter is subjected to linear prediction analysis, and the filter coefficient of the inverse filter 72 is set by the analysis coefficient,
The change in the spectral envelope of the input acoustic signal is suppressed, that is, the linear prediction residual, so-called residual signal, is extracted from the inverse filter 72. This residual signal is input to the codebook selection unit 74, the decoded residual signal of the previous coded frame is compared from the adaptive codebook 75 with the vector extracted at various pitch periods, and the closest one is selected, The selected vector is subtracted from the residual signal from the inverse filter 72 in the difference circuit 76, and the difference signal is vector quantized in the vector quantization unit 77 with reference to the fixed codebook 78. The dequantized output and the vector selected by the codebook selection unit 74 are added by the addition circuit 79 to decode (combine) the residual signals, which are supplied to the adaptive codebook 75. The quantized code of the LPC coefficient as the analysis result of the LPC analysis unit 73, the code indicating the adaptive vector selected by the codebook selection unit 74, and the code indicating the fixed vector selected by the vector quantization unit 77 are combined by the synthesizing unit 8
They are combined by 1 and output as an encoded code. In this case, the synthesizing unit 81 converts the signal into a high-speed signal and outputs it to the transmission line 24. This transmission frame is the same as or several times as large as the encoded frame.

In the CELP decoding section 83, the input coded code is separated into a vector quantization code, an adaptive vector code and an LPC coefficient quantization code in a separation section 84, and the coding section 71 is divided by the vector quantization code. Fixed codebook 85, which is the same as the fixed codebook 78, is taken out by the dequantization unit 86, and the adaptive vector is synthesized by the adaptive vector synthesis unit 87 from the adaptive codebook 88 by the adaptive vector code. The dequantized vector of the conversion unit 86 is added by the addition circuit 89, the residual signal is decoded, and the adaptive codebook 8
8 is input. On the other hand, the LPC coefficient quantization code is LPC
The coefficient is inversely quantized by the coefficient inverse quantization unit 91 and set as a filter coefficient in the LPC synthesis filter 92. The residual signal is passed from the adding circuit 89 to the synthesizing filter 92, and the original acoustic signals are synthesized and output to the output terminal 33.

In this embodiment, the coding side 11 is provided with a lost frame receiving section 49, an internal state accumulating section 51, and an internal state replacement control section 52 as in FIG. The internal state in the CELP encoder 71 is a past sample value, that is, LP.
Internal states of the C inverse filter 72 and the LPC analysis unit 73 (depending on the LPC coefficient quantization method, the past quantized LPC
Coefficients, that is, the internal state of the LPC analysis unit 73), and these are stored in the internal state storage unit 51. Encoding unit 71
The internal state of is replaced with the accumulated internal state in the same manner as in the case of FIG. 3 except that it must be performed at the timing of the encoded frame and in frame units. The decoding side 26 is also provided with a missing frame notification unit 53, an internal state storage unit 54, and an internal state replacement control unit 55, as in FIG. CE
The internal state of the LP decoding unit 83 is a past decoded sample value,
That is, the internal states of the LPC synthesis filter 92 and the LPC coefficient dequantization unit 91 (the past quantized LPC coefficients depending on the LPC coefficient quantization method, that is, the internal state of the LPC dequantization unit 91), and this is the internal state accumulation unit. It is stored in 54. The operation in this case is similar to that on the encoding side 11, except that the replacement of the internal state is at the boundary of the decoded frame at the timing and is obtained in the decoded frame unit.
Is the same as

For the CELP coding method and the decoding method, the method, the coding apparatus and the decoding apparatus shown in FIGS. 1 and 5 can be similarly applied. Also in other encoding methods and decoding methods, the method and apparatus of the present invention may be similarly applied to those that need to retain the past state as an internal state for the encoding and decoding. it can. Further, the present invention can be applied to the communication between a base station and a mobile station in mobile communication by detecting a frame loss in any of them, and can be applied not only to mobile communication but also to other communication. .

[0039]

As described above, when a frame loss occurs, the internal state of the encoding unit and the internal state of the decoding unit are made to coincide with each other, and the distortion of the decoded acoustic signal due to the disagreement of the internal states occurs. It can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing an example of a part of a system to which the invention of claim 1 is applied and each embodiment of the invention of claim 6 and the invention of claim 8;

FIG. 2 is a time chart for explaining the operation example of FIG.

FIG. 3 is a block diagram showing an example of a part of a system to which the invention of claim 2 is applied and each embodiment of the invention of claim 7 and the invention of claim 9;

FIG. 4 is a time chart for explaining the operation example of FIG.

FIG. 5 is a block diagram showing an example of a part of a system to which the invention of claim 3 is applied.

FIG. 6 is a time chart for explaining the operation example of FIG.

FIG. 7 is a block diagram showing another example of a part of the system to which the invention of claim 2 is applied, and other embodiments of the invention of claim 7 and the invention of claim 9;

FIG. 8 is a block diagram showing a part of a system to which a conventional encoding / decoding method is applied.

9 is a time chart for explaining an operation example of FIG.

Claims (9)

[Claims] 1. An encoding side obtains at least a part of a locally decoded signal from an acoustic signal by a process equivalent to decoding,
Encode using the internal state based on the decoded signal, transmit the encoded code sequence for each frame, and on the decoding side, encode the transmitted encoded code sequence from the encoded code to the encoded In the encoding / decoding method that creates an internal state similar to the internal state of the side and obtains an acoustic signal by decoding using the internal state, it is possible that the decoding side does not receive the frame in the encoded code sequence. A step of detecting, a step of notifying the encoding side of the detection, and a step of notifying the encoding side after receiving the notification of the detection on the encoding side, and on the encoding side and the decoding side, An acoustic signal coding / decoding method comprising: initializing each of the internal states at a time corresponding to a decoding code sequence. 2. The encoding side obtains at least a part of a locally decoded signal from an acoustic signal by a process equivalent to decoding,
Encode using the internal state based on the decoded signal, transmit the encoded code sequence for each frame, and on the decoding side, encode the transmitted encoded code sequence from the encoded code to the encoded In the coding / decoding method, in which an internal state similar to that of the above-mentioned side is created, and decoding is performed using the internal state to obtain an acoustic signal, in the decoding side, an unreceived defective frame in the above-mentioned encoded code sequence is detected. Process, the process of notifying the coding side of the lost frame, the process of accumulating the internal states of the coding side and the decoding side, respectively, and the process of notifying the missing frame of the coding side. After reception, the internal states on the encoding side and the decoding side are stored as the encoding code sequence before the lost frame and the decoding code sequence corresponding thereto. Acoustic signal coding and decoding method characterized by having the steps of changing place in the same internal state. 3. On the encoding side, an acoustic signal is obtained by a process equivalent to decoding to obtain at least a part of a locally decoded signal,
Encode using the internal state based on the decoded signal, transmit the encoded code sequence for each frame, and on the decoding side, encode the transmitted encoded code sequence from the encoded code to the encoded It is known in advance that an internal state similar to the internal state of the side is created, decoding is performed using the internal state to obtain an acoustic signal, and transmission of the encoded code sequence is temporarily interrupted on the encoding side. Or in the acoustic signal coding / decoding method in which it is known in advance that the reception of the coded code sequence transmitted by the decoding side is temporarily interrupted, before the interruption, A step of notifying the frame to be interrupted, a step of accumulating the internal state of a predetermined part before the notification frame on the encoding side and the decoding side, respectively, In predetermined portion after the notification frame at the serial decoding end, the acoustic signal coding and decoding method characterized by comprising the steps of changing places the internal state in each internal state of the storage, the. 4. The encoding on the encoding side is ADPC.
M decoding, and the decoding on the decoding side is ADP
4. The acoustic signal coding / decoding method according to claim 1, wherein the acoustic signal coding / decoding method is CM decoding. 5. The acoustic signal coding / decoding method according to claim 2, wherein the accumulation of the internal state is performed at the end of each frame. 6. An acoustic signal is obtained by at least a part of a local decoded signal by a process equivalent to decoding, and an internal state based on the decoded signal is held in an internal state holding unit, and the internal state is held. In an acoustic signal coding apparatus that encodes in a coding unit using, and sends the coded code sequence for each frame, a means for receiving a loss detection notification and a predetermined timing after receiving the loss detection notification. A means for initializing the internal state of the internal state holding section, and the audio signal encoding device. 7. An acoustic signal is obtained by at least a part of a locally decoded signal by a process equivalent to decoding, and an internal state based on the decoded signal is held in an internal state holding unit, and the internal state is held. In an acoustic signal encoding device that encodes in the encoding unit using, and transmits the encoded code sequence for each frame, a storage unit that accumulates the internal state, a unit that receives a missing frame notification, and the notification An audio signal coding apparatus, comprising: means for extracting an internal state of a predetermined portion past a lost frame from the storage section and setting the internal state as the internal state of the internal state holding section. 8. An internal state holding unit holds an internal state based on at least a part of a decoded signal of the encoded code, and decodes the received encoded code sequence using the internal state to obtain an acoustic signal. In the obtained audio signal decoding apparatus, loss detection means for detecting that the frame in the coded code sequence is not received, means for notifying the coding side of the detection, and at a predetermined time point after the notification. A means for initializing the internal state holding unit, and an audio signal decoding apparatus comprising: 9. An internal state holding unit holds an internal state based on at least a part of a decoded signal of the encoded code, and decodes the received encoded code sequence using the internal state to obtain an acoustic signal. In the obtained acoustic signal decoding device, loss detection means for detecting a lost frame in which a frame in the encoded code sequence is not received, a storage unit for storing the internal state of the internal state holding unit, and encoding the lost frame Means for notifying the user of the internal state, and means for replacing the internal state of the internal state holding unit with the internal state corresponding to a time point previously determined before the lost frame from the accumulating unit. A characteristic audio signal decoding device.