CN114171035A - Anti-interference method and device - Google Patents
Anti-interference method and device Download PDFInfo
- Publication number
- CN114171035A CN114171035A CN202010952800.9A CN202010952800A CN114171035A CN 114171035 A CN114171035 A CN 114171035A CN 202010952800 A CN202010952800 A CN 202010952800A CN 114171035 A CN114171035 A CN 114171035A
- Authority
- CN
- China
- Prior art keywords
- current
- lpc filter
- coefficient
- lsp
- filter coefficient
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000001514 detection method Methods 0.000 claims description 13
- 230000000694 effects Effects 0.000 claims description 11
- 238000005516 engineering process Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 description 10
- 238000001228 spectrum Methods 0.000 description 10
- 230000005284 excitation Effects 0.000 description 9
- 230000002159 abnormal effect Effects 0.000 description 5
- 238000012935 Averaging Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 4
- 230000003247 decreasing effect Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000001208 nuclear magnetic resonance pulse sequence Methods 0.000 description 3
- 238000003786 synthesis reaction Methods 0.000 description 3
- 230000001960 triggered effect Effects 0.000 description 3
- 101100455541 Drosophila melanogaster Lsp2 gene Proteins 0.000 description 2
- 101000984710 Homo sapiens Lymphocyte-specific protein 1 Proteins 0.000 description 2
- 102100027105 Lymphocyte-specific protein 1 Human genes 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000036039 immunity Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000006641 stabilisation Effects 0.000 description 2
- 238000011105 stabilization Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 102100040006 Annexin A1 Human genes 0.000 description 1
- 101000959738 Homo sapiens Annexin A1 Proteins 0.000 description 1
- 101000929342 Lytechinus pictus Actin, cytoskeletal 1 Proteins 0.000 description 1
- 101000959200 Lytechinus pictus Actin, cytoskeletal 2 Proteins 0.000 description 1
- 101000797296 Lytechinus pictus Actin, cytoskeletal 3 Proteins 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
- G10L19/04—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
- G10L19/26—Pre-filtering or post-filtering
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention provides an anti-interference method and an anti-interference device, wherein the method comprises the following steps: receiving the coding data of the current speech frame and obtaining the current LPC filter coefficient based on the coding data; generating an all-pole filter based on the current LPC filter coefficients; if the all-pole filter is stable, determining the current LPC filter coefficient as a target LPC filter coefficient; if the all-pole filter is unstable and the adjustment frequency of the current LPC filter coefficient does not reach the preset frequency, adjusting the current LPC filter coefficient, and entering the step of generating the all-pole filter based on the current LPC filter coefficient; and if the all-pole filter is unstable and the adjustment times of the current LPC filter coefficient reach the preset times, determining the LPC filter coefficient corresponding to the nearest effective speech frame as the target LPC filter coefficient. The invention can adjust the LPC filter coefficient and ensure the stability of the filter.
Description
Technical Field
The present application relates to the field of communications technologies, and in particular, to an anti-interference method and apparatus.
Background
In modern communication systems, speech coding and decoding techniques with high coding efficiency, high speech restoration and strong interference immunity are required. Linear Predictive Coding (LPC) is widely used in various vocoder standards as a speech coding and decoding technique.
In a modern communication system, a vocoder may be installed in a transmitting device and a receiving device, the transmitting device performs an encoding operation on a current speech frame by using an encoding module of the vocoder to obtain encoded data, the encoded data is transmitted to the receiving device through a service channel, and the receiving device performs a decoding operation on the encoded data by using a decoding module of the vocoder to obtain a decoded current speech frame.
Disclosure of Invention
The applicant found in the course of research that:
in the encoding operation of the current speech frame by the transmitting device, the LPC filter coefficients are generally calculated by Durbin (levenson-Durbin algorithm), and all poles of the LPC filter coefficient synthesis filter are within the unit circle, so that the filter synthesized based on the LPC filter coefficients in the vocoder decoding module has a better filtering effect.
The LPC filter coefficients are converted into line spectral pair parameters (LSP) during the encoding process, and other parameters are generated during the encoding process, and then encoded data containing the LSP coefficients and other parameters is constructed.
Because of interference in the transmission process of the traffic channel, after encoded data reaches a receiving device through the transmission of the traffic channel, the LSP coefficients may be affected by the traffic channel to generate bit errors, which causes the LPC filter coefficients converted based on the LSP coefficients to generate bit errors, further causes a part of poles in the filter synthesized based on the LPC filter coefficients to possibly exceed a unit circle, and causes the filter to be unstable.
Under the condition that the filter is unstable, the excitation signal passing through the filter is dispersed, and finally, the recovered voice frame is abnormal harsh, that is, the voice frame cannot be normally recovered, so that the user experience is poor.
In view of this, the present application provides an anti-interference method and apparatus, which can adjust the LPC filter coefficients, thereby ensuring the stability of the filter.
In order to achieve the above object, the present invention provides the following technical features:
an interference rejection method, comprising:
receiving coding data of a current voice frame and obtaining a current LPC filter coefficient based on the coding data;
generating an all-pole filter based on the current LPC filter coefficients;
if the all-pole filter is stable, determining the current LPC filter coefficient as a target LPC filter coefficient;
if the all-pole filter is unstable and the adjustment frequency of the current LPC filter coefficient does not reach the preset frequency, adjusting the current LPC filter coefficient, and entering the step of generating the all-pole filter based on the current LPC filter coefficient;
and if the all-pole filter is unstable and the adjustment times of the current LPC filter coefficient reach preset times, determining the LPC filter coefficient corresponding to the nearest effective voice frame as a target LPC filter coefficient.
Wherein said adjusting said LPC filter coefficients comprises:
determining a reference LPC filter coefficient corresponding to a nearest effective speech frame in a plurality of speech frames after decoding operation is executed;
adjusting a first weight corresponding to the reference LPC filter coefficient and a second weight corresponding to the current LPC filter coefficient;
determining the product of the reference LPC filter coefficient and the first weight and the sum of the products of the current LPC filter coefficient and the second weight as an adjusted current LPC filter coefficient;
wherein the sum of the first weight and the second weight is 1, and the preset number of times is greater than 1.
Wherein said adjusting said LPC filter coefficients comprises:
determining a plurality of reference LPC filter coefficients corresponding to a plurality of valid speech frames among the plurality of speech frames after the decoding operation is performed;
adjusting weights of the plurality of reference LPC filter coefficients and the current LPC filter coefficient;
determining a sum of the plurality of reference LPC filter coefficients and the products of the current LPC filter coefficient and the respective weights as an adjusted current LPC filter coefficient;
wherein, the sum of the weights is 1, and the preset times is more than 1.
Wherein said adjusting said LPC filter coefficients comprises:
determining a plurality of reference LPC filter coefficients corresponding to a plurality of speech frames among the plurality of speech frames after performing the decoding operation;
determining an average of the plurality of reference LPC filter coefficients and the current LPC filter coefficient as an adjusted current LPC filter coefficient;
wherein the preset times is 1.
Wherein after determining the target LPC filter coefficient, further comprising:
respectively determining the LPC filter coefficient of each subframe in the current voice frame by using the target LPC filter coefficient and the LPC filter coefficient of the previous voice frame;
respectively generating all-pole filters of all the subframes based on the LPC filter coefficients of all the subframes;
obtaining a synthesized voice signal of each subframe based on the all-pole filter of each subframe;
and obtaining a decoded current speech frame according to the synthesized speech signals of the sub-frames.
Wherein, after obtaining the decoded current speech frame according to the synthesized speech signal of each sub-frame, the method further comprises:
if the current voice frame is judged to be a valid voice frame according to the voice activity detection technology and the voice features extracted from the current voice frame do not accord with the voice features, or if the current voice frame is judged to be an invalid voice frame according to the voice activity detection technology, filling preset noise data into the current voice frame and outputting the current voice frame.
Wherein said adjusting said current LPC filter coefficient comprises: adjusting a current LSP coefficient, and converting the current LSP coefficient into a current LPC filter coefficient;
the determining that the LPC filter coefficient corresponding to the nearest valid speech frame is the target LPC filter coefficient includes: determining the LSP coefficient corresponding to the nearest effective voice frame as a target LSP coefficient, and converting the target LSP coefficient into a target LPC filter coefficient.
Wherein the adjusting the current LSP coefficient includes:
determining a reference LSP coefficient corresponding to the nearest effective speech frame in a plurality of speech frames after decoding operation;
adjusting a first weight corresponding to the reference LSP coefficient and a second weight corresponding to the current LSP coefficient;
determining the product of the reference LSP coefficient and the first weight and the sum of the product of the current LSP coefficient and the second weight as the adjusted current LSP coefficient;
wherein the sum of the first weight and the second weight is 1, and the preset number of times is greater than 1.
Wherein the adjusting the current LSP coefficient includes:
determining a plurality of reference LSP coefficients corresponding to a plurality of valid speech frames in the plurality of speech frames after the decoding operation is performed;
adjusting weights of the plurality of reference LSP coefficients and the current LSP coefficient;
determining the sum of the multiple reference LSP coefficients and the product of the current LSP coefficient and the respective weight as an adjusted current LSP coefficient;
wherein, the sum of the weights is 1, and the preset times is more than 1.
Wherein the adjusting the current LSP coefficient includes:
determining a plurality of reference LSP coefficients which are in one-to-one correspondence with a plurality of voice frames in the plurality of voice frames after decoding operation is executed;
determining the average value of the plurality of reference LSP coefficients and the current LSP coefficient as an adjusted current LSP coefficient;
wherein the preset times is 1.
An apparatus comprising a memory and a processor that implements the tamper-resistant method when executing program instructions stored on the memory.
A computer-readable storage medium, on which a computer program is stored, characterized in that the program realizes the steps of the tamper-resistant method when executed by a processor. Through the technical means, the following beneficial effects can be realized:
the invention provides an anti-interference method, which can receive the coded data of the current voice frame, obtain the current LPC filter coefficient based on the coded data, and generate an all-pole filter based on the current LPC filter coefficient; and then judging the stability of the all-pole filter by utilizing a zero-pole analysis method.
And if the all-pole filter is unstable, adjusting the coefficient of the current LPC filter, generating the all-pole filter based on the coefficient of the current LPC filter, judging the stability of the all-pole filter by using a zero-pole analysis method, and circularly executing adjustment operation and judgment operation until the condition of finishing triggering adjustment jumps out of a cycle.
In case the all-pole filter is stable in the trigger-tuning end condition, it means that the current LPC filter coefficients can be restored by the tuning operation, and thus the current LPC filter coefficients can be determined as the target LPC filter coefficients.
When the number of times of adjustment reaches the preset number of times in the trigger adjustment end condition, it means that the current LPC filter coefficient cannot be restored by the adjustment operation, and therefore, the LPC filter coefficient corresponding to the nearest valid speech frame can be set as the target LPC filter coefficient. A valid speech frame is here a speech frame containing characteristics of human voice.
Through the scheme, the LPC filter coefficient can be adjusted, so that the stability of the filter is ensured, a voice frame can be normally recovered in the subsequent process, and the user experience is improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic structural diagram of an anti-interference system disclosed in an embodiment of the present application;
fig. 2 is a flowchart of a first embodiment of an anti-interference method disclosed in the present application;
fig. 3 is a flowchart of a second embodiment of an anti-interference method disclosed in the present application;
fig. 4 is a flowchart of a third embodiment of an anti-interference method disclosed in the present application;
fig. 5 is a flowchart of a fourth embodiment of an anti-interference method disclosed in the present application;
fig. 6 is a schematic structural diagram of an anti-interference apparatus according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Referring to fig. 1, the present invention provides an anti-jamming system, comprising: a transmitting device 100 and a receiving device 200. The transmitting apparatus 100 and the receiving apparatus 200 are installed with vocoders, in which an encoding module and a decoding module are integrated.
In case that the transmitting device 100 needs to transmit the current speech frame, it may perform an encoding operation on the current speech frame by using an encoding module in the vocoder to generate encoded data, and transmit the encoded data to the receiving device 200 through the traffic channel.
The receiving apparatus 200 performs a decoding operation on the encoded data using a decoding block in the vocoder, and adjusts the LPC filter coefficients in the decoding operation to obtain target LPC filter coefficients. And continuously executing decoding operation based on the target LPC filter coefficient to obtain a decoded current speech frame.
Referring to fig. 2, a first embodiment of an anti-interference method provided by the present invention is applied to a decoding module of a vocoder in a receiving device, and may include the following steps:
step S201: receiving encoded data of a current speech frame and obtaining current LPC filter coefficients based on the encoded data.
After a decoding module of the vocoder receives the coded data of the current voice frame, the decoding module executes demultiplexing operation on the coded data to obtain a demultiplexing data set; wherein the demultiplexed data sets include line spectrum pairs and line spectrum pair index parameters, among other parameters.
And extracting LSP coefficients from the line spectrum pairs by using the line spectrum pair index parameters, and performing polynomial conversion operation on the LSP coefficients to obtain the current LPC filter coefficients.
The LPC filter coefficients are typically calculated by Durbin (levenson-Durbin algorithm) and converted to LSP coefficients in the encoding operation. The inverse operation is performed in the decoding operation with the goal of converting the LSP coefficients to LPC filter coefficients.
Step S202: an all-pole filter is generated based on the current LPC filter coefficients.
Generation of LPC polynomial from current LPC filter coefficients
An all-pole filter can be represented as
Wherein the content of the first and second substances,
p is the order of LPC linear prediction, akFor the coefficients predicted by the LPC filter, z is the transform variable of the z-domain transform.
If the LSP coefficients are not disturbed in the traffic channel, the LSP coefficients are converted to the current LPC filter coefficients that are not disturbed, and the all-pole filter generated based on the current LPC filter coefficients is stable, i.e. all poles of the all-pole filter are within the unit circle.
On the contrary, if the LSP coefficients are interfered in the traffic channel, the LSP coefficients are converted to the current LPC filter coefficients to be interfered, and the all-pole filter generated based on the current LPC filter coefficients is unstable, i.e. part of the poles of the all-pole filter are not within the unit circle.
Step S203: and judging the stability of the all-pole filter by using a zero-pole analysis method.
And determining the pole values of all poles of the all-pole filter in a Z domain by using a zero-pole analysis method, wherein the all-pole filter is stable if all the pole values are in a unit circle, and the all-pole filter is unstable if the pole values are out of the unit circle.
And if the all-pole filter is stable, directly skipping the steps S204-S206, directly taking the current LPC filter coefficient as a target LPC filter coefficient, and continuously executing decoding operation based on the target LPC filter coefficient to obtain a decoded current speech frame.
Step S204: if the all-pole filter is unstable and does not trigger an adjustment end condition, adjusting the current LPC filter coefficient, and proceeding to step S202.
This embodiment provides three implementations of adjusting the current LPC filter coefficients:
the first implementation mode comprises the following steps: interpolation in case of a valid speech frame.
S11: and determining the reference LPC filter coefficient corresponding to the nearest adjacent effective speech frame in a plurality of speech frames after the decoding operation is executed.
It can be understood that a plurality of speech frames are already decoded before the current speech frame, a part of speech frames are noise, i.e. invalid speech frames, and a part of speech frames are voiced, i.e. valid speech frames, and the LPC filter coefficients corresponding to the nearest valid speech frame can be determined from the plurality of speech frames after the decoding operation is performed, and can be used as the reference LPC filter coefficients.
The purpose of selecting the nearest valid speech frame is that the speech frame nearest to the current speech frame will have a greater correlation or similarity to the current speech frame, so the nearest speech frame is used.
Under the condition of recovering the current voice frame, if the voice frame is a noise voice frame, no voice exists, in this case, no reference exists, and if the voice frame is a noise voice frame which is changed into a noise voice frame after being greatly interfered, the voice frame cannot be used, so that a valid voice frame is selected from a plurality of voice frames.
Therefore, with the two contents, the LPC filter coefficient corresponding to the nearest valid speech frame can be selected as the reference LPC filter coefficient from the plurality of speech frames after the decoding operation is performed.
S12: and adjusting a first weight corresponding to the reference LPC filter coefficient and a second weight corresponding to the current LPC filter coefficient.
Initially, the first weight and the second weight are each 0. And in the adjustment process, the first weight corresponding to the reference LPC filter coefficient is continuously increased, and the second weight corresponding to the current LPC filter coefficient is continuously reduced.
Because the all-pole filter synthesized with reference to the LPC filter coefficients is stable, the first weight corresponding to the reference LPC filter coefficients is continuously increased and the second weight corresponding to the current LPC filter coefficients is continuously decreased during the process of adjusting the weights.
The present embodiment provides a specific way to adjust the first weight a1 and the second weight a 2:
a1＝a1+(1/2)n；
a2＝1-a1；
wherein a sum of the first weight and the second weight is 1.
S13: and determining the reference LPC filter coefficient, the first weight and the product of the current LPC filter coefficient and the second weight as the adjusted current LPC filter coefficient.
The adjusted Current LPC filter coefficient is represented by New _ LPC, the reference LPC filter coefficient is represented by Last _ LPC, the Current LPC filter coefficient is represented by Current _ LPC, the first weight is represented by a1, and the second weight is represented by a2, so the calculation formula can be represented as:
New_LPC＝Last_LPC*a1+Current_LPC*a2。
in a first implementation, the current LPC filter coefficient may be adjusted multiple times, where the preset number of times is greater than 1.
The second implementation mode comprises the following steps: interpolation in case of multiple valid speech frames.
Similar to the first implementation, this implementation employs multiple valid speech frames.
S21: and determining a plurality of reference LPC filter coefficients corresponding to a plurality of effective speech frames in a one-to-one sequence from near to far in the plurality of speech frames after the decoding operation is performed.
S22: adjusting weights of the plurality of reference LPC filter coefficients and the current LPC filter coefficient.
Initially, the weights of the plurality of reference LPC filter coefficients and the current LPC filter coefficient are both zero.
Because the all-pole filter synthesized by the reference LPC filter coefficients is stable, the weights corresponding to the reference LPC filter coefficients are continuously increased and the weights corresponding to the current LPC filter coefficients are continuously decreased in the process of adjusting the weights.
Since the closer to the current speech frame the more representative the current speech frame, the more close to the current speech frame the reference LPC filter coefficients have a larger weight, and the more distant from the current speech frame the reference LPC filter coefficients have a smaller weight.
This embodiment provides a specific way to adjust the multiple weights, taking four weights as an example for three reference LPC filter coefficients and the current LPC filter coefficient, where initially a1 is 0.25, a2 is 0.25, a3 is 0.25, and a4 is 0.25.
a1＝a1+(1/4)n；
a2＝a2+(1/8)n；
a3＝a3+(1/16)n；
a4＝1-a1-a2-a3；
Wherein the sum of the four weights is 1.
S23: and determining the product of the reference LPC filter coefficients and the current LPC filter coefficients and the respective weights as the adjusted current LPC filter coefficients.
The adjusted Current LPC filter coefficient is represented by New _ LPC, the reference LPC filter coefficient is represented by Last _ LPC1, Last _ LPC2 and Last _ LPC3, the Current LPC filter coefficient is represented by Current _ LPC, and the weights are represented by a1, a2, a3 and a4, respectively, then the calculation formula can be expressed as:
New_LPC＝Last_LPC1*a1+Last_LPC2*a2+Last_LPC3*a3+Current_LPC*a4
in a second implementation, the current LPC filter coefficient may be adjusted multiple times, where the preset number of times is greater than 1.
The third implementation mode comprises the following steps: averaging in case of multiple speech frames.
S31: and determining a plurality of reference LPC filter coefficients corresponding to the plurality of speech frames in a one-to-one sequence from near to far according to a sliding window in the plurality of speech frames after the decoding operation is performed.
According to a preset sliding window, determining a plurality of voice frames according to the time sequence from near to far. The speech frames in this implementation may be reference LPC filter coefficients corresponding to valid speech frames or invalid speech frames.
S32: determining an average of the plurality of reference LPC filter coefficients and the current LPC filter coefficient as an adjusted current LPC filter coefficient.
The average of the plurality of reference LPC filter coefficients is determined as the adjusted current LPC filter coefficient. In this implementation, the average value is obtained, so the preset number is 1.
Step S205: and under the condition that the all-pole filter is stable in the condition of triggering the adjustment ending, determining the current LPC filter coefficient as a target LPC filter coefficient.
In the process of repeating adjustment and judgment in steps S202 to S205, if an adjustment end condition is triggered, the loop is skipped. The adjustment end condition comprises the stabilization of the all-pole filter and the adjustment times reaching the preset times.
And under the condition that the all-pole filter is stable in the trigger adjustment ending condition, the adjusted LPC filter coefficient is restored, and the all-pole filter generated based on the current LPC filter coefficient is stable, so that the current LPC filter coefficient is determined to be the target LPC filter coefficient.
Step S206: and under the condition that the adjusting times reach the preset times in the condition of triggering the adjustment ending, determining the LPC filter coefficient corresponding to the nearest effective speech frame as the target LPC filter coefficient.
And under the condition that the adjusting times reach the preset times in the trigger adjusting ending condition, the adjusted current LPC filter coefficient cannot be recovered, and the all-pole filter generated based on the current LPC filter coefficient is unstable.
Therefore, in order to ensure the stability of the all-pole filter, the LPC filter coefficient corresponding to the nearest valid speech frame is determined as the target LPC filter coefficient. The all-pole filter synthesized by the LPC filter coefficients corresponding to the nearest neighbor valid speech frame is stable.
Optionally, this embodiment may further include an adjustment flag, where the adjustment flag is set to "1" when the LPC filter coefficient is subjected to the adjustment operation, and otherwise the adjustment flag is set to "0".
The embodiment has the following beneficial effects:
the embodiment provides an anti-interference method, which can receive encoded data of a current voice frame, obtain a current LPC filter coefficient based on the encoded data, and generate an all-pole filter based on the current LPC filter coefficient; and then judging the stability of the all-pole filter by utilizing a zero-pole analysis method.
And if the all-pole filter is unstable, adjusting the coefficient of the current LPC filter, generating the all-pole filter based on the coefficient of the current LPC filter, judging the stability of the all-pole filter by using a zero-pole analysis method, and circularly executing adjustment operation and judgment operation until the condition of finishing triggering adjustment jumps out of a cycle.
In case the all-pole filter is stable in the trigger-tuning end condition, it means that the current LPC filter coefficients can be restored by the tuning operation, and thus the current LPC filter coefficients can be determined as the target LPC filter coefficients.
When the number of times of adjustment reaches the preset number of times in the trigger adjustment end condition, it means that the current LPC filter coefficient cannot be restored by the adjustment operation, and therefore, the LPC filter coefficient corresponding to the nearest valid speech frame can be set as the target LPC filter coefficient. A valid speech frame is here a speech frame containing characteristics of human voice.
Through the scheme, the LPC filter coefficient can be adjusted, so that the stability of the filter is ensured, a voice frame can be normally recovered in the subsequent process, and the user experience is improved.
Referring to fig. 3, the second embodiment of the anti-interference method provided by the present invention is applied to a decoding module of a vocoder in a receiving device.
It is understood that one current speech frame includes a plurality of subframes. After determining the target LPC filter coefficient corresponding to the current speech frame in the first embodiment, a decoding operation can be performed based on the target LPC filter coefficient.
In the process of performing demultiplexing operation on the encoded data in step S201, the process of obtaining a demultiplexed data set may obtain an adaptive codebook and pitch index parameters, an algebraic codebook and codebook index parameters for each subframe, in addition to the line spectrum pair and line spectrum pair index parameters.
On the basis of the first embodiment, the following steps may be further included after steps S205 and S206:
step S301: and respectively determining the LPC filter coefficient of each subframe in the current voice frame by using the target LPC filter coefficient and the LPC filter coefficient of the previous voice frame.
The standards of different vocoders may include different numbers of subframes. Taking the ACELP vocoder in the Tetra standard as an example, one current speech frame includes four subframes, and some vocoders include two subframes.
The vocoder standard is preset with the formula and weight for calculating each sub-frame, and the detailed description of the steps is given by taking the example that one current voice frame comprises four sub-frames:
q1＝0.75qn-1+0.25qn-1；
q2＝0.5qn-1+0.5qn-1；
q3＝0.25qn-1+0.75qn-1；
q4＝qn-1；
wherein, the LPC filter coefficients of four sub-frames of the current speech frame respectively adopt q1、q2、q3And q is4Is represented by qn-1The LPC filter coefficients representing the previous speech frame are shown, it should be noted that the LPC filter coefficients of the previous speech frame are also determined through steps S201 to S205 shown in fig. 2, i.e. the synthesized all-pole filter is stable.
Because the target LPC filter coefficient of the current voice frame and the LPC filter coefficient of the previous voice frame are both repaired, the LPC filter coefficient of each subframe calculated based on the target LPC filter coefficient of the current voice frame and the LPC filter coefficient of the previous voice frame is repaired, and the LPC filter coefficient does not have an error code.
Step S302: and respectively generating all-pole filters of the sub-frames based on the LPC filter coefficients of the sub-frames.
And respectively generating the all-pole filter of each subframe based on the LPC filter coefficient of each subframe, wherein the all-pole filter of each subframe is stable because the LPC filter coefficient of each subframe is repaired and no error is generated.
Step S303: an excitation signal for each sub-frame is determined based on the encoded data and the target LPC filter coefficients, respectively.
S41: and respectively extracting pitch period information of each subframe from the corresponding adaptive codebook by using the pitch index parameter of each subframe, and respectively extracting a random pulse sequence of each subframe from the corresponding algebraic codebook by using the codebook index parameter of each subframe.
S42: and respectively determining the pitch gain of each subframe by using the self-adaptive codebook of each subframe and the target LPC filter coefficient.
S43: and respectively determining the codebook gain of each subframe by using the algebraic codebook of each subframe and the target LPC filter coefficient.
S44: and determining the product of the pitch period information of each subframe and the corresponding pitch gain and the product of the random pulse sequence and the corresponding codebook gain as the excitation signal of each subframe.
Taking a subframe as an example, the excitation signal is represented by u (n), the pitch period information is represented by v (n), and the pitch gain is represented by gpRepresenting that the random pulse sequence adopts c (n), the codebook gain adopts gcIs shown to be
u(n)＝gpv(n)+gcc(n)。
The determination of the excitation signal of each subframe in step S303 is a mature technology, and the specific implementation process thereof will not be described in detail here.
Step S304: and respectively inputting the excitation signal of each subframe to the corresponding all-pole filter to obtain the synthesized voice signal of each subframe.
After determining the all-pole filter of each sub-frame and the excitation signal of each sub-frame, inputting the excitation signal of each sub-frame to the corresponding all-pole filter, i.e. the speech synthesis filter, respectively, and synthesizing the speech signal of each sub-frame so as to perform a synthesis operation on each excitation signal by using the all-pole filter.
Step S305: and according to the synthesized voice signals of each subframe, obtaining a decoded current voice frame through the amplitude and saturation control processing operation of a post filter.
And (3) carrying out recombination operation on the synthesized voice signals of each subframe according to the original front and back sequence of each subframe in the current voice frame, and obtaining the decoded current voice frame through amplitude and saturation control processing operation of a post filter.
The receiving device may directly output the current speech frame after obtaining the current speech frame.
The present embodiment can obtain the following advantageous effects:
on the basis of the beneficial effect of the embodiment, the embodiment can normally recover the voice frame, and the user experience is improved.
Referring to fig. 4, the third embodiment of the anti-interference method provided by the present invention is applied to a decoding module of a vocoder in a receiving device.
On the basis of adjusting the flag bit to "1", on the basis of the third embodiment of the interference avoidance method shown in fig. 3, after step S305, the following steps may be further included:
step S401: and judging whether the current voice frame is a valid voice frame or not by utilizing a voice activity detection technology. If so, the process proceeds to step S402, otherwise, the process proceeds to step S405.
Using voice activity detection technology (VAD technology) to determine whether the current voice frame is a valid voice frame, that is, whether there is voice in the current voice frame is determined, if there is voice, step S402 is performed, if there is no voice, it indicates that the current voice frame is noise, and step S405 is directly performed.
Step S402: and if the current voice frame is an effective voice frame representing the voice, extracting voice characteristics from the current voice frame.
For the accuracy of detection, another scheme is used again to detect whether the current speech frame is really voiced. Speech features, such as formant features, are extracted from the current speech frame.
Step S403: and judging whether the voice features accord with the human voice features, if so, entering a step S404, and otherwise, entering a step S405.
Judging whether the voice characteristics accord with the voice characteristics, if so, indicating that the current voice frame is a valid voice frame containing voice, marking the identifier of the valid voice frame for the current voice frame, otherwise, marking the identifier of the invalid voice frame for the current voice frame.
Step S404: and if the voice feature is determined to be the human voice feature, outputting the current voice frame.
If the current voice frame is an effective voice frame containing human voice, the current voice frame is directly output, namely the receiving equipment plays the current voice frame so that the user can listen to the voice.
Step S405: if the current voice frame is an invalid voice frame representing noise or a voice characteristic non-human voice characteristic, filling preset noise data into the current voice frame to output the current voice frame, or performing a mute operation on the current voice frame.
If the current speech frame is a noise speech frame or does not contain human voice, the current speech frame is indicated to contain invalid environmental noise or sharp noise, and in order to improve user experience, preset noise data is filled in the current speech frame, wherein the preset noise data is adjusted noise of a comfortable human body, such as non-sharp white noise. Then, the current speech frame is output.
Alternatively, the muting operation is performed directly on the current speech frame, and one current speech frame is usually a short time, for example, 30ms, so the muting operation can eliminate the noise without affecting the user experience.
The present embodiment can obtain the following advantageous effects:
the embodiment fundamentally identifies whether the abnormal sharp sound exists in the voice, eliminates the abnormal sharp sound fundamentally and restores the voice frame, and avoids the problems of missing detection, false detection, voice frame information loss and the like of the abnormal sharp sound in the prior art.
The LPC filter adjusting process in the first embodiment and the two detection processes in the third embodiment are effectively combined, the LPC filter can be adjusted to identify and eliminate abnormal sharp sounds and restore speech frames from the source, the quality of the current speech frame can be detected from the result of the two detection processes, and if the current speech frame is noise, the current speech frame is muted or is filled with preset noise, so that smooth transition is realized, and user experience is improved.
According to the embodiment, hardware settings of the receiving equipment and the sending equipment do not need to be changed, and under the condition that no cost is increased, the anti-interference capability of the voice decoding process can be improved, so that the vocoder is more suitable for a complex service channel environment, and the customer experience is improved.
The embodiment has strong universality, can be suitable for the voice channel coding and decoding standards based on the LPC linear prediction technology, and can be applied to the voice coding and decoding standards of 2G, 3G and 4G.
Referring to fig. 5, the fourth embodiment of the anti-interference method provided in the present invention is applied to a decoding module of a vocoder in a receiving device, and may include the following steps:
step S501: receiving the coding data of the current speech frame and obtaining the current LSP coefficient based on the coding data.
After a decoding module of the vocoder receives the coded data of the current voice frame, the decoding module executes demultiplexing operation on the coded data to obtain a demultiplexing data set; wherein the demultiplexed data sets include line spectrum pairs and line spectrum pair index parameters, among other parameters.
And extracting the current LSP coefficient from the line spectrum pair by using the line spectrum pair index parameter.
Step S502: converting the current LSP coefficients to current LPC filter coefficients.
The LPC filter coefficients are typically calculated by Durbin (levenson-Durbin algorithm) and converted to LSP coefficients in the encoding operation. The inverse operation is performed in the decoding operation with the goal of converting the LSP coefficients to LPC filter coefficients.
Step S503: an all-pole filter is generated based on the current LPC filter coefficients.
Current LPC filter coefficient adoption
Representing generation of LPC polynomial from current LPC filter coefficients
An all-pole filter can be represented as
Wherein the content of the first and second substances,
p is the order of LPC linear prediction, akFor the coefficients predicted by the LPC filter, z is the transform variable of the z-domain transform.
If the LSP coefficients are not disturbed in the traffic channel, the LSP coefficients are converted to the current LPC filter coefficients that are not disturbed, and the all-pole filter generated based on the current LPC filter coefficients is stable, i.e. all poles of the all-pole filter are within the unit circle.
On the contrary, if the LSP coefficients are interfered in the traffic channel, the LSP coefficients are converted to the current LPC filter coefficients to be interfered, and the all-pole filter generated based on the current LPC filter coefficients is unstable, i.e. part of the poles of the all-pole filter are not within the unit circle.
Step S504: and judging the stability of the all-pole filter by using a zero-pole analysis method.
And determining the pole values of all poles of the all-pole filter in a Z domain by using a zero-pole analysis method, wherein the all-pole filter is stable if all the pole values are in a unit circle, and the all-pole filter is unstable if the pole values are out of the unit circle.
And if the all-pole filter is stable, directly skipping the steps S204-S206, directly taking the current LPC filter coefficient as a target LPC filter coefficient, and continuously executing decoding operation based on the target LPC filter coefficient to obtain a decoded current speech frame.
Step S505: if the all-pole filter is unstable and an adjustment end condition is not triggered, adjusting the current LSP coefficient, and proceeding to step S202.
This embodiment provides three implementation manners for adjusting the current LSP coefficient:
the first implementation mode comprises the following steps: interpolation in case of a valid speech frame.
S11: and determining the reference LSP coefficient corresponding to the nearest effective speech frame in the plurality of speech frames after the decoding operation is performed.
It can be understood that a plurality of speech frames are already decoded before the current speech frame, a part of the speech frames are noise, i.e. invalid speech frames, and a part of the speech frames are voice, i.e. valid speech frames, and an LSP coefficient corresponding to the nearest valid speech frame can be determined from the plurality of speech frames after the decoding operation is performed, and used as a reference LSP coefficient.
The purpose of selecting the nearest valid speech frame is that the speech frame nearest to the current speech frame will have a greater correlation or similarity to the current speech frame, so the nearest speech frame is used.
Under the condition of recovering the current voice frame, if the voice frame is a noise voice frame, no voice exists, in this case, no reference exists, and if the voice frame is a noise voice frame which is changed into a noise voice frame after being greatly interfered, the voice frame cannot be used, so that a valid voice frame is selected from a plurality of voice frames.
Therefore, with the two contents, the LSP coefficient corresponding to the most adjacent valid speech frame can be selected as the reference LSP coefficient from the plurality of speech frames after the decoding operation is performed.
S12: and adjusting a first weight corresponding to the reference LSP coefficient and a second weight corresponding to the current LSP coefficient.
Initially, the first weight and the second weight are each 0. And in the adjustment process, the first weight corresponding to the reference LSP coefficient is continuously increased, and the second weight corresponding to the current LSP coefficient is continuously reduced.
Because the all-pole filter synthesized by the reference LSP coefficients is stable, the first weight corresponding to the reference LSP coefficients is continuously increased and the second weight corresponding to the current LSP coefficients is continuously decreased in the process of adjusting the weights.
The present embodiment provides a specific way to adjust the first weight a1 and the second weight a 2:
a1＝a1+(1/2)n；
a2＝1-a1；
wherein a sum of the first weight and the second weight is 1.
S13: and determining the reference LSP coefficient, the first weight and the product of the current LSP coefficient and the second weight as the adjusted current LSP coefficient.
The adjusted Current LSP coefficient is represented by New _ LSP, the reference LSP coefficient is represented by Last _ LSP, the Current LSP coefficient is represented by Current _ LSP, the first weight is represented by a1, and the second weight is represented by a2, so the calculation formula can be represented as:
New_LSP＝Last_LSP*a1+Current_LSP*a2。
in a first implementation, the current LSP coefficient may be adjusted multiple times, where the preset number of times is greater than 1.
The second implementation mode comprises the following steps: interpolation in case of multiple valid speech frames.
Similar to the first implementation, this implementation employs multiple valid speech frames.
S21: and determining a plurality of reference LSP coefficients corresponding to a plurality of effective speech frames in a one-to-one sequence from near to far in the plurality of speech frames after the decoding operation is performed.
S22: adjusting weights of the plurality of reference LSP coefficients and the current LSP coefficient.
Initially, the weights of the plurality of reference LSP coefficients and the current LSP coefficient are all zero.
Because the all-pole filter synthesized by the multiple reference LSP coefficients is stable, the weights corresponding to the multiple reference LSP coefficients are continuously increased and the weights corresponding to the current LSP coefficients are continuously decreased in the process of adjusting the weights.
Since the closer to the current speech frame the more representative the current speech frame, the more close to the reference LSP coefficient of the current speech frame the greater its weight, and the farther away from the current speech frame the less its weight.
In this embodiment, a specific way to adjust the multiple weights is given, taking four weights as an example of three reference LSP coefficients and the current LSP coefficient, where initially a1 is 0.25, a2 is 0.25, a3 is 0.25, and a4 is 0.25.
a1＝a1+(1/4)n；
a2＝a2+(1/8)n；
a3＝a3+(1/16)n；
a4＝1-a1-a2-a3；
Wherein the sum of the four weights is 1.
S23: and determining the product sum of the multiple reference LSP coefficients and the current LSP coefficient and the respective weight as the adjusted current LSP coefficient.
The adjusted Current LSP coefficient is represented by New _ LSP, the reference LSP coefficient is represented by Last _ LSP1, Last _ LSP2, and Last _ LSP3, the Current LSP coefficient is represented by Current _ LSP, and the weights are represented by a1, a2, a3, and a4, respectively, so the calculation formula can be represented as:
in a second implementation, the New _ LSP1 a1+ Last _ LSP2 a2+ Last _ LSP3 a3+ Current _ LSP a4 may adjust the Current LSP coefficient multiple times, where the preset number of times is greater than 1.
The third implementation mode comprises the following steps: averaging in case of multiple speech frames.
S31: and determining a plurality of reference LSP coefficients corresponding to the plurality of speech frames in a one-to-one sequence from near to far in a sliding window in the plurality of speech frames after the decoding operation is performed.
According to a preset sliding window, determining a plurality of voice frames according to the time sequence from near to far. The voice frame in this implementation may be a reference LSP coefficient corresponding to a valid voice frame or an invalid voice frame.
S32: and determining the average value of the plurality of reference LSP coefficients and the current LSP coefficient as the adjusted current LSP coefficient.
And determining the average value of the plurality of reference LSP coefficients as the adjusted current LSP coefficient. In this implementation, the average value is obtained, so the preset number is 1.
Step S506: and under the condition that the all-pole filter is stable in the condition of triggering the adjustment ending, determining the current LSP coefficient as a target LSP coefficient, and converting the target LSP coefficient into a target LPC filter coefficient.
In the process of repeating adjustment and judgment in steps S502 to S506, if an adjustment end condition is triggered, the loop is skipped. The adjustment end condition comprises the stabilization of the all-pole filter and the adjustment times reaching the preset times.
And under the condition that the all-pole filter is stable in the trigger adjustment ending condition, the LSP coefficient after adjustment is recovered, the current LSP coefficient is converted into the current LPC filter coefficient based on the current LSP coefficient, and the all-pole filter generated by the LPC filter coefficient is stable, so that the current LPC filter coefficient is determined to be the target LPC filter coefficient.
Step S507: and under the condition that the adjustment times reach the preset times in the condition of triggering the adjustment ending, determining the LSP coefficient corresponding to the nearest effective voice frame as a target LSP coefficient, and converting the target LSP coefficient into a target LPC filter coefficient.
And under the condition that the adjustment times reach the preset times in the trigger adjustment ending condition, indicating that the LSP coefficient after adjustment cannot be recovered.
Therefore, in order to ensure the stability of the all-pole filter, the LSP coefficient corresponding to the nearest valid speech frame is determined as the current LSP coefficient, and the current LSP coefficient is converted into the current LPC filter coefficient and used as the target LPC filter coefficient.
Optionally, this embodiment may further include an adjustment flag, where the adjustment flag is set to "1" when the LPC filter coefficient is subjected to the adjustment operation, and otherwise the adjustment flag is set to "0".
After obtaining the target LPC filter coefficient in the fourth embodiment, the operation may be continued according to the schemes in the second embodiment and the third embodiment, which are not described herein again.
An apparatus is provided that includes a memory and a processor that implements an interference rejection method when executing program instructions stored on the memory.
A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the interference immunity method.
Referring to fig. 6, the present invention provides an anti-jamming device, comprising:
a receiving unit 61, configured to receive encoded data of a current speech frame and obtain a current LPC filter coefficient based on the encoded data;
a generating unit 62 for generating an all-pole filter based on the current LPC filter coefficients;
a first determining unit 63, configured to determine, if the all-pole filter is stable, that the current LPC filter coefficient is a target LPC filter coefficient;
an adjusting unit 64, configured to adjust the current LPC filter coefficient if the all-pole filter is unstable and the number of times of adjustment of the current LPC filter coefficient does not reach a preset number of times, and enter the step of generating the all-pole filter based on the current LPC filter coefficient;
a second determining unit 65, configured to determine, if the all-pole filter is unstable and the number of adjustment times of the current LPC filter coefficient reaches a preset number, that the LPC filter coefficient corresponding to the nearest valid speech frame is the target LPC filter coefficient.
In a first aspect: the LPC filter coefficients are directly adjusted.
A first implementation of the first aspect:
the adjusting unit 64 includes:
a first determining coefficient unit, configured to determine, among a plurality of speech frames subjected to a decoding operation, a reference LPC filter coefficient corresponding to a nearest valid speech frame;
a first adjusting weight unit for adjusting a first weight corresponding to the reference LPC filter coefficient and a second weight corresponding to the current LPC filter coefficient;
a first product-sum unit for determining a sum of a product of the reference LPC filter coefficient and a first weight and a product of the current LPC filter coefficient and a second weight as an adjusted current LPC filter coefficient;
wherein the sum of the first weight and the second weight is 1, and the preset number of times is greater than 1.
A second implementation of the first aspect:
the adjusting unit 64 includes:
a second determination coefficient unit configured to determine a plurality of reference LPC filter coefficients corresponding to a plurality of valid speech frames, among the plurality of speech frames on which the decoding operation has been performed;
a second adjusting weight unit for adjusting weights of the plurality of reference LPC filter coefficients and the current LPC filter coefficient;
a second product-sum unit for determining a sum of the plurality of reference LPC filter coefficients and products of the current LPC filter coefficient and the respective weights as an adjusted current LPC filter coefficient;
wherein, the sum of the weights is 1, and the preset times is more than 1.
A third implementation of the first aspect:
the adjusting unit 64 includes:
a third determination coefficient unit configured to determine, among the plurality of speech frames subjected to the decoding operation, a plurality of reference LPC filter coefficients corresponding to the plurality of speech frames;
an averaging unit configured to determine an average of the plurality of reference LPC filter coefficients and the current LPC filter coefficient as an adjusted current LPC filter coefficient;
wherein the preset times is 1.
In a second aspect: and adjusting the LSP parameters and indirectly adjusting the LPC filter coefficients.
The adjusting unit 64 includes:
an LSP adjusting unit, which is used for adjusting the current LSP coefficient and converting the current LSP coefficient into the current LPC filter coefficient;
the determining that the LPC filter coefficient corresponding to the nearest valid speech frame is the target LPC filter coefficient includes: determining the LSP coefficient corresponding to the nearest effective voice frame as a target LSP coefficient, and converting the target LSP coefficient into a target LPC filter coefficient.
A first implementation of the second aspect:
the adjusting LSP unit includes:
a first LSP determining unit, configured to determine, in the multiple speech frames after performing the decoding operation, a reference LSP coefficient corresponding to a nearest valid speech frame;
a first LSP adjustment weight unit, configured to adjust a first weight corresponding to the reference LSP coefficient and a second weight corresponding to the current LSP coefficient;
a first LSP product-sum unit configured to determine a sum of a product of the reference LSP coefficient and a first weight and a product of the current LSP coefficient and a second weight as an adjusted current LSP coefficient;
wherein the sum of the first weight and the second weight is 1, and the preset number of times is greater than 1.
Second implementation of the second aspect:
the adjusting LSP unit includes:
a second LSP-determining coefficient unit, configured to determine, in the plurality of speech frames after performing the decoding operation, a plurality of reference LSP coefficients corresponding to the plurality of valid speech frames;
a second LSP weighting adjustment unit for adjusting the weighting of the plurality of reference LSP coefficients and the current LSP coefficient;
a second LSP product-sum unit for determining a sum of the plurality of reference LSP coefficients and products of the current LSP coefficients and respective weights as an adjusted current LSP coefficient;
wherein, the sum of the weights is 1, and the preset times is more than 1.
A third implementation of the second aspect:
the adjusting LSP unit includes:
a third LSP determining unit, configured to determine, in the multiple speech frames after performing the decoding operation, multiple reference LSP coefficients that correspond to the multiple speech frames one to one;
an LSP averaging unit, configured to determine an average value of the multiple reference LSP coefficients and the current LSP coefficient as an adjusted current LSP coefficient;
wherein the preset times is 1.
The device further comprises:
a subframe determining unit 66, configured to determine, by using the target LPC filter coefficient and the LPC filter coefficient of the previous speech frame, an LPC filter coefficient of each subframe in the current speech frame respectively;
a sub-frame generation filter unit 67 for generating all-pole filters for respective sub-frames based on LPC filter coefficients for the respective sub-frames;
a synthesized sub-frame voice signal unit 68, configured to obtain a synthesized voice signal of each sub-frame based on the all-pole filter of each sub-frame;
a decoding unit 69, configured to obtain a decoded current speech frame according to the synthesized speech signal of each subframe.
An output unit 610, configured to, if the current speech frame is determined to be a valid speech frame according to a speech activity detection technique and the speech features extracted from the current speech frame do not conform to the speech features, or if the current speech frame is determined to be an invalid speech frame according to the speech activity detection technique, fill preset noise data in the current speech frame and output the current speech frame.
Through the technical means, the following beneficial effects can be realized:
the invention provides an anti-interference method, which can receive the coded data of the current voice frame, obtain the current LPC filter coefficient based on the coded data, and generate an all-pole filter based on the current LPC filter coefficient; and then judging the stability of the all-pole filter by utilizing a zero-pole analysis method.
And if the all-pole filter is unstable, adjusting the coefficient of the current LPC filter, generating the all-pole filter based on the coefficient of the current LPC filter, judging the stability of the all-pole filter by using a zero-pole analysis method, and circularly executing adjustment operation and judgment operation until the condition of finishing triggering adjustment jumps out of a cycle.
In case the all-pole filter is stable in the trigger-tuning end condition, it means that the current LPC filter coefficients can be restored by the tuning operation, and thus the current LPC filter coefficients can be determined as the target LPC filter coefficients.
When the number of times of adjustment reaches the preset number of times in the trigger adjustment end condition, it means that the current LPC filter coefficient cannot be restored by the adjustment operation, and therefore, the LPC filter coefficient corresponding to the nearest valid speech frame can be set as the target LPC filter coefficient. A valid speech frame is here a speech frame containing characteristics of human voice.
Through the scheme, the LPC filter coefficient can be adjusted, so that the stability of the filter is ensured, a voice frame can be normally recovered in the subsequent process, and the user experience is improved.
The functions described in the method of the present embodiment, if implemented in the form of software functional units and sold or used as independent products, may be stored in a storage medium readable by a computing device. Based on such understanding, part of the contribution to the prior art of the embodiments of the present application or part of the technical solution may be embodied in the form of a software product stored in a storage medium and including several instructions for causing a computing device (which may be a personal computer, a server, a mobile computing device or a network device) to execute all or part of the steps of the method described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts among the embodiments are referred to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (12)
1. An interference rejection method, comprising:
receiving coding data of a current voice frame and obtaining a current LPC filter coefficient based on the coding data;
generating an all-pole filter based on the current LPC filter coefficients;
if the all-pole filter is stable, determining the current LPC filter coefficient as a target LPC filter coefficient;
if the all-pole filter is unstable and the adjustment frequency of the current LPC filter coefficient does not reach the preset frequency, adjusting the current LPC filter coefficient, and entering the step of generating the all-pole filter based on the current LPC filter coefficient;
and if the all-pole filter is unstable and the adjustment times of the current LPC filter coefficient reach preset times, determining the LPC filter coefficient corresponding to the nearest effective voice frame as a target LPC filter coefficient.
2. The method of claim 1, wherein the adjusting the LPC filter coefficients comprises:
determining a reference LPC filter coefficient corresponding to a nearest effective speech frame in a plurality of speech frames after decoding operation is executed;
adjusting a first weight corresponding to the reference LPC filter coefficient and a second weight corresponding to the current LPC filter coefficient;
determining the product of the reference LPC filter coefficient and the first weight and the sum of the products of the current LPC filter coefficient and the second weight as an adjusted current LPC filter coefficient;
wherein the sum of the first weight and the second weight is 1, and the preset number of times is greater than 1.
3. The method of claim 1, wherein the adjusting the LPC filter coefficients comprises:
determining a plurality of reference LPC filter coefficients corresponding to a plurality of valid speech frames among the plurality of speech frames after the decoding operation is performed;
adjusting weights of the plurality of reference LPC filter coefficients and the current LPC filter coefficient;
determining a sum of the plurality of reference LPC filter coefficients and the products of the current LPC filter coefficient and the respective weights as an adjusted current LPC filter coefficient;
wherein, the sum of the weights is 1, and the preset times is more than 1.
4. The method of claim 1, wherein the adjusting the LPC filter coefficients comprises:
determining a plurality of reference LPC filter coefficients corresponding to a plurality of speech frames among the plurality of speech frames after performing the decoding operation;
determining an average of the plurality of reference LPC filter coefficients and the current LPC filter coefficient as an adjusted current LPC filter coefficient;
wherein the preset times is 1.
5. The method of any of claims 2-4, further comprising, after determining the target LPC filter coefficients:
respectively determining the LPC filter coefficient of each subframe in the current voice frame by using the target LPC filter coefficient and the LPC filter coefficient of the previous voice frame;
respectively generating all-pole filters of all the subframes based on the LPC filter coefficients of all the subframes;
obtaining a synthesized voice signal of each subframe based on the all-pole filter of each subframe;
and obtaining a decoded current speech frame according to the synthesized speech signals of the sub-frames.
6. The method of claim 5, wherein after obtaining the decoded current speech frame from the synthesized speech signals of the sub-frames, the method further comprises:
if the current voice frame is judged to be a valid voice frame according to the voice activity detection technology and the voice features extracted from the current voice frame do not accord with the voice features, or if the current voice frame is judged to be an invalid voice frame according to the voice activity detection technology, filling preset noise data into the current voice frame and outputting the current voice frame.
7. The method of claim 1,
said adjusting said current LPC filter coefficient comprises: adjusting a current LSP coefficient, and converting the current LSP coefficient into a current LPC filter coefficient;
the determining that the LPC filter coefficient corresponding to the nearest valid speech frame is the target LPC filter coefficient includes: determining the LSP coefficient corresponding to the nearest effective voice frame as a target LSP coefficient, and converting the target LSP coefficient into a target LPC filter coefficient.
8. The method of claim 7, wherein said adjusting the current LSP coefficient comprises:
determining a reference LSP coefficient corresponding to the nearest effective speech frame in a plurality of speech frames after decoding operation;
adjusting a first weight corresponding to the reference LSP coefficient and a second weight corresponding to the current LSP coefficient;
determining the product of the reference LSP coefficient and the first weight and the sum of the product of the current LSP coefficient and the second weight as the adjusted current LSP coefficient;
wherein the sum of the first weight and the second weight is 1, and the preset number of times is greater than 1.
9. The method of claim 7, wherein said adjusting the current LSP coefficient comprises:
determining a plurality of reference LSP coefficients corresponding to a plurality of valid speech frames in the plurality of speech frames after the decoding operation is performed;
adjusting weights of the plurality of reference LSP coefficients and the current LSP coefficient;
determining the sum of the multiple reference LSP coefficients and the product of the current LSP coefficient and the respective weight as an adjusted current LSP coefficient;
wherein, the sum of the weights is 1, and the preset times is more than 1.
10. The method of claim 7, wherein said adjusting the current LSP coefficient comprises:
determining a plurality of reference LSP coefficients which are in one-to-one correspondence with a plurality of voice frames in the plurality of voice frames after decoding operation is executed;
determining the average value of the plurality of reference LSP coefficients and the current LSP coefficient as an adjusted current LSP coefficient;
wherein the preset times is 1.
11. An apparatus comprising a memory and a processor that when executing program instructions stored on the memory implements the method of any of claims 1-10.
12. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010952800.9A CN114171035A (en) | 2020-09-11 | 2020-09-11 | Anti-interference method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010952800.9A CN114171035A (en) | 2020-09-11 | 2020-09-11 | Anti-interference method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114171035A true CN114171035A (en) | 2022-03-11 |
Family
ID=80475369
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010952800.9A Pending CN114171035A (en) | 2020-09-11 | 2020-09-11 | Anti-interference method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114171035A (en) |
-
2020
- 2020-09-11 CN CN202010952800.9A patent/CN114171035A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2535893B1 (en) | Device and method for lost frame concealment | |
| US9224399B2 (en) | Apparatus and method for concealing frame erasure and voice decoding apparatus and method using the same | |
| KR101868926B1 (en) | Noise signal processing and generation method, encoder/decoder and encoding/decoding system | |
| WO2001059766A1 (en) | Background noise reduction in sinusoidal based speech coding systems | |
| WO2008148321A1 (en) | An encoding or decoding apparatus and method for background noise, and a communication device using the same | |
| RU2707727C1 (en) | Audio signal processing device, audio signal processing method and audio signal processing program | |
| KR20070007851A (en) | Hierarchy encoding apparatus and hierarchy encoding method | |
| KR20160124877A (en) | Voice frequency code stream decoding method and device | |
| JP2004138756A (en) | Voice coding device, voice decoding device, and voice signal transmitting method and program | |
| CN114550732A (en) | Coding and decoding method and related device for high-frequency audio signal | |
| US20130085751A1 (en) | Voice communication system encoding and decoding voice and non-voice information | |
| JP3558031B2 (en) | Speech decoding device | |
| CN106683681B (en) | Method and device for processing lost frame | |
| CN1244090C (en) | Speech coding with background noise reproduction | |
| WO2024021747A1 (en) | Sound coding method, sound decoding method, and related apparatuses and system | |
| JP6626123B2 (en) | Audio encoder and method for encoding audio signals | |
| US8595000B2 (en) | Method and apparatus to search fixed codebook and method and apparatus to encode/decode a speech signal using the method and apparatus to search fixed codebook | |
| CN114171035A (en) | Anti-interference method and device | |
| JP4236675B2 (en) | Speech code conversion method and apparatus | |
| WO2000063878A1 (en) | Speech coder, speech processor, and speech processing method | |
| JP6859379B2 (en) | Equipment and methods for comfortable noise generation mode selection | |
| JP2001343984A (en) | Sound/silence discriminating device and device and method for voice decoding | |
| JP2003029790A (en) | Voice encoder and voice decoder | |
| JP4764956B1 (en) | Speech coding apparatus and speech coding method | |
| WO2005031708A1 (en) | Speech coding method applying noise reduction by modifying the codebook gain |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |