JP2001154699A - Hiding for frame erasure and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the hiding method of erased frames by frame repetition that the performance is improved. SOLUTION: This method uses a decoder with respect to frames which are encodingly excited and LP-encoded by both of an adaptive code book and a fixed code book. Moreover, the method uses muted repetitive excitation, a composite filter for expanding and repeating a threshold adaptive band width and a repetitive pitch delay (T(m)) which is made to jitter in order to hide erased frames.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to electronic devices, and more particularly, to a method and circuit for speech encoding, transmitting, storing, and decoding / synthesizing.

[0002]

BACKGROUND OF THE INVENTION In current and foreseeable digital communications, the performance of digital voice systems using low bit rates has become increasingly important. Both dedicated channels and packetized over network (eg, voice over IP or voice over packet) transmissions benefit from compression of voice signals. Widely used linear prediction (LP:
Linear prediction digital speech compression methods model the vocal tract as a time-varying filter and a time-varying filter excitation to mimic the human voice. Linear predictive analysis And the energy Σr of the residual r (n) in the frame
By minimizing (n) ² , the LP coefficients a _i , i = 1, 2,. . . , M. Usually, the order M of the linear prediction filter is about 10 to 12. The sampling rate for forming samples s (n) is typically 8 kHz (same as public switched telephone network sampling for digital transmission). The number of samples {s (n)} in a frame is typically 80 or 160 (1
0 or 20 ms frame). The frames of the sample are variously windowed (wind
owing) operation.
The name "linear prediction" Is the linear combination of the preceding audio samples Is interpreted as an error in predicting s (n). Therefore, minimizing {r (n) ² yields {a _i } that optimizes linear prediction for the frame. The coefficient {a _i } is determined by a line spectral frequency (LSF).
nces), and may be quantized and transmitted or stored, or a line spectrum pair (LSP: li)
(ne.spectral pairs) to interpolate between subframes.

[0003] {r (n)} is the LP residual for the frame. Ideally, the LP residual is the synthesis filter 1 / A
Excitation for (z). Here, A (z) is given by equation (1).
Is the transfer function of Of course, the LP residual cannot be obtained at the decoder. Thus, the task of the encoder is to represent the LP residual so that the decoder can generate an excitation that mimics the LP residual from the encoded parameters. Physiologically, for voiced frames the excitation will be in the form of a series of pulses at the pitch frequency, and for unvoiced frames the excitation will be in the form of almost white noise.

The LP compression approach basically consists of (quantized) filter coefficients, (quantized) residuals (waveforms,
Or parameters (such as pitch) and (quantized) gain (s) are only transmitted / stored. The receiver decodes the transmitted / stored item and plays the input audio with the same perceptual characteristics. 5 and 6 show the high-level blocks of the LP system. Since the required bits for the periodic update of the quantized item are less than the direct representation of the audio signal, a reasonable LP encoder is 2
To 3 kb / s (kilobits / second).

However, due to the high error rate of wireless transmission and the large packet loss / delay for network transmission, LP decoders
One has to deal with frames where so many bits are lost that frames are ignored (erased). To maintain speech quality and intelligibility for wireless or packet-on-packet applications when frames are erased, decoders typically have a method to conceal such frame erasures. Such methods can be categorized as interpolated or iterative. The interpolation concealment method interpolates missing parameters by using both future frame parameters and past frame parameters. In general, the interpolation concealment method improves the approximation of the speech signal of the missing frame compared to the iterative concealment method using only past frame parameters.
In applications such as wireless communications, interpolation concealment (concea
l) The method must come at the expense of adding a delay to obtain future frames. For voice over packet communication, future frames are available from playout buffers. The playout buffer compensates for packet arrival jitter. The interpolation concealment method mainly increases the size of the playout buffer. An iterative concealment method that simply repeats or modifies past frame parameters is described in 729, G.C. 723.1, and GSM-E
Used in some CELP speech encoders, including FR. The iterative concealment method in these encoders does not increase the delay or the size of the playout buffer, but in the presence of erasure frames, the performance of the reconstructed speech is lower than that of the interpolation approach. Inferior. This is especially true in an environment where the ratio of erased frames is high or in a burst-like frame erased environment.

More specifically, according to ITU standard G. 7
29 is a 10 ms long frame (80 samples
use. Track pitch and gain parameters
To improve codebook search
10ms long frame to reduce complexity
Are two subframes of 40 samples each of 5 ms
Divided into Each subframe has an adaptive codebook
Represented by contributions and fixed (algebraic) codebook contributions
With the excitation to be performed. The adaptive codebook contribution is
Gives periodicity to the excitation. The adaptive codebook contribution is
Converted and interpolated by the time delay of the current frame pitch
Gain g to the excitation v (n) of the previous frame _pMultiplied by
is there. The algebraic codebook contribution is the actual residual and adaptive
The difference between the codebook contribution and the four-pulse vector c (n)
And gain g_cApproximate by the product of Therefore, the excitation is u
(N) = g_pv (n) + g_cc (n). Where v
(N) is from previous (decoded) frame
And g_p, G_c, And c (n) are in the current frame
It is obtained from the parameters transmitted to it.
3 and 4 show the encoding and decoding in block form.
You. Postfilters emphasize essentially any periodicity
(For example, vowels).

G. 729 handles frame erasure by reconstruction based on previously received information. That is, iterative concealment. That is, replace the missing excitation signal with one of the similar characteristics, and gradually reduce its energy by using a voicing classifier based on long-term predicted gain (calculated as part of long-term post-filter analysis). To attenuate. The long-term post-filter uses a normalized correlation method greater than 0.5 for optimal delay determination, resulting in a prediction gain of 3
Long-term predictors greater than dB (long-term pr
dictor). For the error concealment process, a 10 ms frame is declared to be periodic if one or more 5 ms subframes have a long term prediction gain greater than 3 dB. Otherwise, the frame is declared aperiodic. The erased frame inherits its class from the preceding (reconstructed) speech frame. Note that the articulation classification is continuously updated based on this reconstructed audio signal. The specific steps taken for the erased frame are as follows.

1) Iteration of synthesis filter parameters. The LP parameter of the last good frame is used.

2) Adaptive codebook gain and fixed codebook
Attenuation of clock gain. The adaptive codebook gain is
Based on the attenuated version of the bookbook gain. (M + 1)
When the frame of is deleted, g_p ^{(m + 1)}= 0.9g
_p ^(m)Use Similarly, the fixed codebook gain is
Based on an attenuated version of the fixed codebook gain. g _c
^{(m + 1)}= 0.98g_c ^(m).

3) Decay of the memory of the gain predictor. The gain predictor for a fixed codebook gain uses the energy of the previously selected algebraic codebook vector c (n). To avoid the effects of the transition, once a good frame is received, the gain predictor memory is updated with an attenuated version of the average codebook energy over the four previous frames.

4) Generation of displacement excitation. The excitation used depends on the periodicity classification. If the last reconstructed frame was classified as periodic, then the current frame is also considered periodic. In that case, only the adaptive codebook contribution is used and the fixed codebook contribution is set to zero. The pitch delay is based on the integer part of the pitch delay of the previous frame and is repeated for each subsequent frame. To avoid excessive periodicity, the pitch delay value is increased by one every next frame, up to 143. Alternatively, if the last reconstructed frame was classified as aperiodic, then the current frame is also considered aperiodic and the adaptive codebook contribution is 0.
Is set to The fixed codebook contribution is generated by randomly selecting a codebook index and a code index. By using classification, different decay rates can be used for both types of excitation (e.g., 0. 0 for periodic gain).
9, 0.98 for aperiodic gain). FIG. 2 shows a decoder with concealment parameters.

"Speech Frame Reconstruction Method for CELP Speech Encoder in Digital Cellular and Wireless Communications," by Leung et al., Proceedings Wireless Magazine (Leung et al., Voice Fra).
me Reconstruction Methods
for CELP Speech Codersin
Digital Cellular and Wir
eless Communications, Pro
c. Wireless 93 (Jully 1993)
Describes missing frame reconstruction using parametric extrapolation and interpolation for low complexity CELP encoders using four subframes per frame. However, the results of the iterative hiding method are not good.

[0013]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for concealing erased frames by frame repetition, which has improved performance for iterative concealment methods.

[0014]

SUMMARY OF THE INVENTION In accordance with the present invention, a frame is repeated with one or more of excitation signal muting, LP coefficient bandwidth expansion at a cutoff frequency, and pitch delay jittering.

[0015]

BEST MODE FOR CARRYING OUT THE INVENTION Overview The decoder and method of the preferred embodiment for concealing frame erasure in the transmission of a signal such as CELP encoded speech comprises one or more of the following three features. (1) Muting the excitation outside the feedback loop.
This replaces adaptive codebook gain and fixed codebook gain attenuation. (2) The bandwidth of the LP synthesis filter having the threshold frequency is expanded to make the expansion coefficient different.
(3) Jitter the pitch delay to avoid the overlap of periodic repetitive frames. Features (2) and (3)
This is especially true for bursty noise leading to frame erasure. FIG. 1 shows a preferred embodiment decoder that uses all three concealment features. This is shown in FIG. This contrasts with the concealment of the decoder of the G.729 standard.

The preferred embodiment system (eg, voice over IP or voice over packet) includes the preferred embodiment concealment method in a decoder.

2. Encoder Details To describe the preferred embodiment, see Certain details of an encoding method similar to G.729 are needed. In particular, a speech encoder using LP coding with excitation contributions from both an adaptive codebook and an algebraic codebook is shown. And the concealment features of the preferred embodiment affect pitch delay, codebook gain, and LP synthesis filter. The encoding proceeds as follows.

(1) A series of digital samples s (n) are obtained by sampling the input audio signal at 8 kHz or 16 kHz (which may be pre-processed to filter out DC and low frequencies, etc.). Divide the sample stream into frames. For example, split into 80 samples or 160 samples (eg, 10 ms frame) or other convenient size. The analysis and encoding may use different sized subframes of the frame or other periods.

(2) Each frame (or subframe)
For linear prediction (LP: linear predi)
ction) analysis to find the LP (and hence LSF / LSP) coefficients and quantize the coefficients. More specifically, the LSF is a monotonically increasing frequency {f ₁ , f ₂ , f ₃ ,... Between 0 and the Nyquist frequency (4 kHz or 8 kHz for a sampling rate of 8 kHz or 16 kHz). . . f _N }.
That is, 0 <f ₁ <f ₂ . . . <F _M <f _samp / 2, where M is the order of the linear prediction filter and is usually in the range of 10 to 12. The LSF for transmission / storage is quantized by vector quantizing the difference group between the frequency group and the fourth-order moving average prediction value of the frequency group.

(3) s within the windowed range
By searching for the correlation between (n) and s (n + k),
Find the pitch delay for each subframe. Before searching
(N) may be filtered perceptually. Search is 2
It may be performed in stages. That is, it applies to the open loop search using the correlation of s (n) to find the pitch delay, and then to the excitation of the target speech x (n) in the (sub) frame and the previous (sub) frame. This is a closed-loop search that refines the pitch delay by interpolation from the maximum value of the normalized inner product <x | y> with the speech y (n) generated by the quantized LP synthesis filter of the (sub) frame. The resolution of the pitch delay may be part of the sample. This is especially true for smaller pitch delays. At this time, the adaptive codebook v (n) is the excitation of the previous (sub) frame, which has been transformed by the refined pitch delay and interpolated.

(4) Determine the adaptive codebook gain g _p as the ratio of the inner product <x | y> divided by <y | y>. here,
x (n) is the target speech in the (sub) frame, y
(N) is the (perceptually weighted) in the (sub) frame where the quantized LP synthesis filter is applied to the adaptive codebook vector v (n) from step (3)
It is voice. Thus, g _p v (n) is the adaptive codebook contribution to the excitation and g _p y (n) is the adaptive codebook contribution to the speech in the (sub) frame.

[0022] (5) (sub) for each frame, (sub) x quantization as the target speech in the frame LP synthesis filter filtered c (n) (n) -g p y (n)
Find the algebraic codebook vector c (n) by essentially maximizing the normalized correlation with That is, by removing the adaptive codebook contribution,
Get a new purpose. For more information, correlation _{<x-g p y | H} |
The square of c> Energy ^{<c | H T H | c} > as divided by the ratio is maximized, the searching possible algebraic codebook vectors c (n). Where h (n) is the impulse response of the quantized LP synthesis filter (performing perceptual filtering) and H is the diagonal h (0), h (1),. . .
With lower triangle (lower triangula)
r) Toeplitz convolution matrix. If a (sub) frame of 40 samples (5 ms) is used as the coding granularity, the vector c (n)
Has 40 positions. Forty samples are partitioned into four interleaved tracks, with one pulse in each track. Each of the three tracks has eight samples, and one track has sixteen samples.

[0023] _{(6) | x-g p} y-g c z | by the minimizing, determine the algebraic codebook gain g _c. Here, as before, x (n) is the target speech in the (sub) frame, g _p is the adaptive codebook gain, and y (n) applies to v (n). A quantized LP synthesis filter, where z (n) is the signal in the frame generated by applying the quantized LP synthesis filter to the algebraic codebook vector c (n).

(7) Quantize gains g _p and g _c for insertion as part of the codeword. Algebraic codebook gains may be factored and predicted. The gains may be quantized together in a vector quantization codebook. next,
The excitation for the (sub) frame is u (n) = g _p v
It is quantized by (n) + g _c c (n) and the excitation memory is updated for use in the next (sub) frame.

It should be noted that all of the quantized items typically have different values, and the moving average of the values of the previous frame is used as a predictor. That is, only the difference between the actual value and the predicted value is encoded.

The final codewords for encoding the (sub) frames are the quantized LSF coefficients, the adaptive codebook pitch delay, the algebraic codebook vector, and the quantized adaptive and algebraic codebooks. Gain,
.

3. Details of the Decoder FIG. 1 shows the decoder and the decoding method of the preferred embodiment. This decoder and decoding method essentially reverses the coding steps of the above coding method, and provides the feature of iterative concealment for erasure frame reconstruction as described in the next section. Is what you do. FIG. 4 shows a decoder without concealment features, proceeding as follows for the m-th (sub) frame. (1) The quantized LP coefficient a _j ^(m) is decoded. Since the coefficients can be in the form of a differential LSP, a moving average of the decoded coefficients of the previous frame may be used. The LP coefficient may be interpolated every 20 samples (subframes) in the LSP domain. (2) Quantized pitch delay T of adaptive codebook
^(m) decoding the, decoded before the pitch delay is applied to the (sub) frame of the excitation u ^(m-1) (n) by (time conversion and interpolation) to the vector v ^(m) (N) is formed. This is the feedback loop of FIG. (3) Decode the algebraic codebook vector c ^(m) (n). (4) Decode the gains g _p ^(m) and g _c ^(m) of the quantized adaptive and algebraic codebooks. (5) Use the items from steps (2)-(4) to calculate the excitation for the m th (sub) frame as u ^(m) (n) =
It is formed as g _p ^(m) v ^(m) (n) + g _c ^(m) c ^(m) (n). (6) A voice is synthesized by applying from the LP synthesis filter in step (1) to the excitation in step (5). (7) Apply any post-filtering and other shaping operations.

4. FIG. 1 shows the preferred embodiment concealment features of the preferred embodiment decoder,
This is in contrast to FIG. Specifically, it is assumed that the m-th frame has been decoded, but the (m + 1) -th frame has been deleted, and the (m + 2) -th frame,. . . , The (m + j) th frame. . . The same applies to Next, the concealment feature of the preferred embodiment comprises the (m + j) th frame in one or more of the following decoder steps.

(1) (quantized) filter coefficient a _k
^{(m + j)} is the (quantized) coefficient a _{k of the} previous good frame
^The LP synthesis filter is assumed to be a bandwidth-expanded version of ^(m). Is determined. j = 1, 2,. . . For successively erased frames, And the bandwidth expansion factor γ ⁽ⁿ⁾ is in the range [0.8, 1..
0]. FIG. 1 illustrates this bandwidth extension applied to a synthesis filter. The decoder updates the bandwidth extension by the following formula for each frame. Here, C _B is the bursty frame erasure counter which counts the number of frames subsequent erased, LSFB
W _MiN is the minimum LSF bandwidth of the last good frame. The ith LSF bandwidth (LSFBW _i ) is | f _{i + 1} −f
_i |. The smaller the LSF bandwidth, the corresponding peak (formant) in the LPC spectrum
Becomes sharp. That is, LSFBW _min is the minimum LSF
Since it is BW _i , the bandwidth expansion factor may be reduced only when at least one pair of LSF frequencies is close (sharp formant). Note that as γ ^{(n )} decreases, the poles of the synthesis filter Move radially toward the origin, thereby expanding the formant peak.

Therefore, when the m-th frame is a good frame and the (m + 1) -th frame is deleted, C _B =
At 1, the updated magnification factor is γ ^{(m + 1)} = min (1.0
5γ ^{(m )} , 1.0). (Γ ^{(m + 1)} = 1.05γ ^(m)
In the case of ≦ 1), γ ^(m) had to be about 0.953 or less. This means that at least one of the preceding four frames has a reduced γ ^(n), which means two or more consecutive erased frames. )
However, there are (m + 2) th frames or more erased frames, and the LSFBW _{min of} the mth frame is 100
Below Hz, the coefficient γ ^{(m + j)} gradually decreases to the limit of 0.8. This allows any sharp formant (LSFBW _min <100 Hz) in the m-th frame to be shifted to the (m +
2) Prevents affecting the combined quality of the concealment reconstruction for the frame and subsequent successive erased frames.
That is, the synthesis filter is used to conceal the erased (m + j) -th frame. It is. Here, the filter coefficients a _k ^(m) are obtained from the last good frame.

Also, for good frames following a burst of frame erasures, γ ^{(m + j)} is still applied to the decoded filter coefficients and γ ^{(m + j + 1)} = min (1. 05
γ ^{(m + j)} , 1.0) gradually increases to 1.0 in order to smoothly recover from frame erasure.

(2) Replace the quantized pitch delay T ^{(m + 1)} of the adaptive codebook for concealing the erased (m + 1) frame with T ^(m) from the previous good m-th frame. Determine equally. However, for two or more consecutive erased frames, T ^{(m + j)} (j = 2,3...) Is determined by adding a random 3% jitter to T ^(m) . Thereby, G. As in 729, T ^{(m + j + 1)} is exactly T ^{(m + j)} +1
Thus, it is possible to avoid reconstructing an excessively periodic concealment signal without accumulating the estimation error that may occur in the case of (1).
This concealment pitch delay is determined by the excitation u of the previous (sub) frame.
^(m) Apply to (n) to form an adaptive codebook vector v ^{(m + j)} (n). In short, a random number in the range of [-0.03T ^(m) , 0.03T ^(m) ] is applied to T ^(m) , and rounded to the nearest 1/3 or an integer according to the range, thereby successively. T for erased frame
^{(m + j)} . FIG. 1 shows the jitter and the feedback loop shows the use of the excitation of the previous frame.

(3) Algebraic codebook vector c
^{(m + j)} (n) is defined as a random vector of the type c ^(m) (n). That is, G. In the case of 729-type coding, it is assumed that four vectors out of 40 zero components are ± 1 pulses.

(4) The quantized adaptive codebook gain g _p ^{(m + j)} and algebraic codebook gain g _c ^{(m + j)} are simply determined to be equal to g _p ^(m) and g _c ^(m). . Where g _p ^{(m + j)}
The upper limit _{max (1.2-0.1 (C B -1)} , 0.
8). Again, C _B is the count of the number of successive erased frames, that is, the burst. This upper limit prevents unexpected surges of excitation signal energy. By using such an unattenuated gain, the excitation energy is maintained. However, as explained in step (5), the excitation is weakened (muted) before the synthesis by applying the coefficient g _E ^{(m + j} ).

(5) Deleted (m + 1) th (sub)
Excitation for the frame from steps (2)-(4)
Use item u^{(m + 1)}(N) = g_p ^{(m + 1)}v
^{(m + 1)}(N) + g_c ^{(m + 1)}c^{(m + 1)}(N).
Next, as shown in FIG.
Excitation muting coefficient g on the outside_E ^{(m + 1)}Apply This
This eliminates, but eliminates, the excessive attenuation of the excitation.
Frame that follows the frame containing the beginning of the vowel
Surges of voice energy that can occur are still avoided
You. Excitation muting section for each subframe (5 ms)
Several g_E ⁽ⁿ⁾Is updated to be within the range [0.0,1.0].
You. The update is performed every frame (10 ms) as follows.
New muting counter C_MDepends on C_BIf> 1, C_M= 4 otherwise g_p ^{(m + 1)}<1.0 at C_M> 0
If C_MOtherwise reduce C by 1_MChange
Not where C_BAgain in succession of erased frames
A burst-like counter for counting the number, g_p ^{(m + 1)}Is
Algebraic codebook gain from step (4).
Then g_E(N) The update is as follows. g_E ^{(n + 1)}= 0.95499 g_E ⁽ⁿ⁾ C_M ^{(n + 1)}If> 0 g_E ^{(n + 1)}= Min (1.09648 g_E ⁽ⁿ⁾, 1.0) Otherwise, the excitation for the synthesis filter is g_E ^{(m + 1)}u
^{(m + 1)}(N). Similarly, the (m + 1) -th successive disappearance
For the last frame, the corresponding g_p ^{(m + j)}v
^{(m + j)}(N) + g_c ^{(m + j)}c^{(m + j)}Using (n), g_E
^{(m + j)}Muting with.

(6) A speech is synthesized by applying the excitation of step (5) from the LP synthesis filter of step (1).

(7) Apply any post-filtering and other shaping operations.

5. Alternative Preferred Embodiment The alternative preferred embodiment performs only one or two of the three concealment features of the preferred embodiment. In fact, it is possible to omit the bandwidth expansion of the LP coefficients for erasure frames and for good frames after erasure frame bursts. This only changes the synthesis filter and does not affect excitation muting or pitch delay jittering.

Another alternative preferred embodiment omits pitch delay jittering, but uses excitation muting and L
With the bandwidth expansion of the P coefficient, the G.P. 729, an increment can be used.

In addition, an alternative preferred embodiment omits excitation muting, along with pitch delay jittering and bandwidth expansion of the synthesis filter coefficients, as well as G.264. 729 configuration is used.

Finally, the preferred embodiment uses only one of the three features (excitation muting, pitch delay jittering, and bandwidth expansion of the synthesis filter coefficients), while G. 729.

6. System Preferred Embodiment FIGS. 5 and 6 show, in functional block form, a preferred embodiment system that uses the preferred embodiment encoding and decoding. This also applies to speech and other signals that can be effectively CELP encoded. Encoding and decoding may be performed by a digital signal processor (DSP), or a general-purpose programmable processor, or a dedicated circuit or system on a chip,
For example, the control can be performed by both the DSP and the RISC processor on the same chip that controls the RISC processor. The codebook is stored in the memory of both the encoder and the decoder. Internal or external ROM, Flash EEP
A ROM or a program stored in a DSP or a programmable processor can perform signal processing. Analog-to-digital and digital-to-analog converters provide real-world coupling, and modulators and demodulators (and antennas for the air interface) provide for transmission waveforms. The encoded voice can be packetized and transmitted via a network such as the Internet.

7. Modifications The preferred embodiment can be modified in various ways while maintaining one or more of the characteristics of erasure frame concealment due to bandwidth expansion of the synthesis filter coefficients, pitch delay jittering, and excitation muting. For example, the size of the periods (frames and sub-frames) and the sampling rate can vary. The bandwidth expansion factor is C _B > 0 or C _B >
2 can be applied. Multipliers 0.95 and 1.
05 and the limits 0.8 and 1.0 can be varied. The threshold of 100 Hz can be changed. Pitch delay jitter can be increased or decreased as a percentage of pitch delay, and can be applied to the first erased frame. The size of the jitter can be changed according to the number of successive erased frames or the erase density. The excitation muting can be changed non-linearly depending on the number of consecutive erased frames or the erase density.
The multipliers 0.95499 and 1.09648 can be changed.

This application claims priority from US Provisional Patent Application No. 60 / 167,197, filed November 23, 1999.

With respect to the above description, the following items are further disclosed. (1) A method for decoding digital speech, comprising: (a) summing an adaptive codebook contribution and a fixed codebook contribution to form an excitation for an erased period of encoded digital speech. Step, wherein the adaptive codebook contribution is determined from the excitation, pitch, and first gain of a temporally previous period of the encoded digital speech, and wherein the fixed codebook contribution is temporally Decoding digital speech, comprising: forming an excitation, derived from a second gain of the previous period; (b) muting the excitation; and (c) filtering the muted excitation. Method.

(2) The method for decoding digital audio according to (1), wherein (a) the filtering includes synthesis by a synthesis filter coefficient obtained from a filter coefficient of the period preceding the time. A method for decoding digital audio.

(3) A method for decoding digital speech, wherein (a) determining a filter coefficient from a bandwidth-expanded version of the filter coefficient of a temporally previous period of the encoded digital speech. Thereby forming a synthesis filter for the erased period of the encoded digital audio; and (b) filtering the excitation for the erased period with the synthesis filter for the erased period. And a method for decoding digital voice.

(4) Decoding of the digital voice described in item 3
(A) with respect to the erased period
The filter coefficient a of the synthesis filter ₁, A_Two,. . .
a_MIs before the synthesis filter for the temporally previous period
The filter coefficient b₁, B_Two,. . . b_MFor a₁= F
b₁, A_Two= F^Twob_Two,. . . , A_M= F^Mb_MHas a relationship
Where f is the bandwidth expansion factor,
Method.

(5) A method for decoding digital speech, wherein (a) the sum of the adaptive codebook contribution and the fixed codebook contribution forms an excitation for the erased period of the encoded digital speech. Step, wherein the adaptive codebook contribution is determined from the excitation, pitch, and first gain of a temporally previous period of the encoded digital speech, and wherein the fixed codebook contribution is temporally A method for decoding digital speech, comprising: forming an excitation, derived from a second gain of a previous said period; and (b) filtering the muted excitation.

(6) The digital audio decoding method according to (5), wherein (a) the filtering is performed by muting and subsequent synthesis obtained from a synthesis filter coefficient of the temporally preceding period. The digital speech decoding method, comprising: synthesizing with a filter coefficient.

(7) The method for decoding digital speech according to item 6, wherein (a) a bandwidth-expanded version of the synthesis filter coefficient of the encoded digital speech in a temporally preceding period. And determining a synthesis filter coefficient for the period from the following.

(8) a decoder for a CELP coded signal, comprising: (a) a fixed codebook vector decoder; (b) a fixed codebook gain decoder; and (c) an adaptive codebook gain. A decoder; (d) an adaptive codebook pitch delay decoder; (e) an excitation generator coupled to each of the decoders; (f) a synthesis filter; and (g) an output of the excitation generator and A muting gain coupled between the input of the synthesis filter, and (h) when the received frame is erased, each of the decoders generates an alternate output, and the excitation generator comprises: A decoder for generating an alternative excitation, wherein the synthesis filter generates alternative filter coefficients, and wherein the muting gain mutes the alternative excitation.

(9) The decoder according to item 8, wherein
(A) The fixed codebook decoder and the adaptive codebook decoder both generate the alternative outputs by repeating the output for the previous frame.

(10) A decoder for LP-coded frames code-excited by both the adaptive codebook and the fixed codebook. For concealment of the erased frame, use a muted repetitive excitation, a threshold adaptive bandwidth extension iterative synthesis filter, and a jittered repetition pitch delay.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a decoder according to a preferred embodiment.

FIG. 2 is a diagram illustrating a known decoder concealment method.

FIG. 3 is a block diagram of a known encoder.

FIG. 4 is a block diagram of a known decoder.

FIG. 5 is a diagram showing an LP system.

FIG. 6 is a diagram showing an LP system.

[Explanation of symbols]

g _p ^(m) adaptive codebook gain g _c ^(m) fixed codebook gain T ^(m) pitch delay

Claims (2)

[Claims] 1. A method for decoding digital speech, comprising: (a) providing excitation for an erased period of encoded digital speech by a sum of an adaptive codebook contribution and a fixed codebook contribution. Forming the adaptive codebook contribution from the excitation, pitch, and first gain of the encoded digital speech in a temporally earlier period, wherein the fixed codebook contribution is time Digital audio comprising: an excitation formation step, derived from a second gain of the previous period, (b) muting the excitation, and (c) filtering the muted excitation. Decoding method. 2. A decoder for CELP encoded signals, comprising: (a) a fixed codebook vector decoder; (b) a fixed codebook gain decoder; and (c) an adaptive codebook gain decoding. (D) an adaptive codebook pitch delay decoder; (e) an excitation generator coupled to each of the decoders; (f) a synthesis filter; and (g) an output of the excitation generator and the synthesis. A muting gain coupled between the input of the filter and: (h) each of the decoders generates an alternative output when the received frame is canceled, and the excitation generator generates an alternative output. Wherein the synthesis filter generates alternative filter coefficients and the muting gain mutes the alternative excitation.