WO2014075579A1 - 一种译码方法及装置 - Google Patents

一种译码方法及装置 Download PDF

Info

Publication number
WO2014075579A1
WO2014075579A1 PCT/CN2013/086670 CN2013086670W WO2014075579A1 WO 2014075579 A1 WO2014075579 A1 WO 2014075579A1 CN 2013086670 W CN2013086670 W CN 2013086670W WO 2014075579 A1 WO2014075579 A1 WO 2014075579A1
Authority
WO
WIPO (PCT)
Prior art keywords
decoding
decoding result
decoder
result
soft value
Prior art date
Application number
PCT/CN2013/086670
Other languages
English (en)
French (fr)
Inventor
吴可镝
魏岳军
唐欣
熊杰
Original Assignee
华为技术有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Publication of WO2014075579A1 publication Critical patent/WO2014075579A1/zh

Links

Classifications

    • HELECTRICITY
    • H03ELECTRONIC CIRCUITRY
    • H03MCODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/63Joint error correction and other techniques
    • H03M13/6312Error control coding in combination with data compression
    • HELECTRICITY
    • H03ELECTRONIC CIRCUITRY
    • H03MCODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/03Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
    • H03M13/05Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
    • H03M13/09Error detection only, e.g. using cyclic redundancy check [CRC] codes or single parity bit
    • HELECTRICITY
    • H03ELECTRONIC CIRCUITRY
    • H03MCODING; DECODING; CODE CONVERSION IN GENERAL
    • H03M13/00Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
    • H03M13/37Decoding methods or techniques, not specific to the particular type of coding provided for in groups H03M13/03 - H03M13/35
    • H03M13/39Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes
    • H03M13/41Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors
    • H03M13/4115Sequence estimation, i.e. using statistical methods for the reconstruction of the original codes using the Viterbi algorithm or Viterbi processors list output Viterbi decoding

Definitions

  • the present invention claims the priority of a Chinese patent application filed on November 15, 2012 by the Chinese Patent Office, the application number is 201210460305.1, and the invention is entitled "a decoding method and device". This is incorporated herein by reference.
  • the present invention relates to the field of communication network technologies, and in particular, to a decoding method and apparatus.
  • BACKGROUND OF THE INVENTION Voice communication is one of the most basic and important ways in current communication systems.
  • the channel coding used for current speech transmission is mostly a convolutional code.
  • the decoding of the convolutional code is usually performed using the Viterbi Algorithm (VA).
  • VA Viterbi Algorithm
  • VA Viterbi Algorithm
  • the convolutional code can also be decoded by using a List Viterbi Algorithm (LVA).
  • the LVA includes a parallel enumeration Viterbi List Viterbi Algorithm (PLVA) and a serial enumeration Viterbi Algorithm (SLVA).
  • PLVA parallel enumeration Viterbi List Viterbi Algorithm
  • SLVA serial enumeration Viterbi Algorithm
  • VA and LVA are all channel decoding for channel coding. Since the Adaptive Muti-Rate (AMR) speech coding has redundant information between the parameters of adjacent frames, the redundancy between adjacent frames of the source can be utilized, so that the decoding performance can be improved.
  • AMR Adaptive Muti-Rate
  • an embodiment of the present invention provides a decoding method and apparatus, which can improve a subjective voice quality average opinion score while reducing noise risk.
  • a decoding method including:
  • the parameters in the second decoding result are corrected, and the correction result is sent to the voice decoder.
  • a decoding apparatus including:
  • a decoder based on the error detection code auxiliary decision, configured to decode the first substream of the demodulated soft value to obtain a first decoding result
  • a speech frame repair decoder configured to correct the first decoding result to obtain a second decoding result
  • the decoding validity determining module is configured to correct the parameter in the second decoding result, and send the correction result to the voice decoder.
  • An embodiment of the present invention provides a decoding method and apparatus, which obtains a first decoding result by decoding a first substream of a demodulated soft value by using a decoder based on an error detection code auxiliary decision;
  • the first decoding result is corrected by using a speech frame repair decoder to obtain a second decoding result; the parameters in the second decoding result are corrected, and the correction result is used to recover the speech by using a speech decoder.
  • Waveform When using LVA for decoding in the prior art, the probability of missed detection is increased. Large, that is, introducing a large noise risk, and failing to make full use of the redundancy of the source.
  • the BER performance can be improved, but the improvement of BLER is very limited, subjective
  • the improvement of the voice quality MOS is also very limited.
  • the decoder and the speech frame repair decoder based on the error detection code are jointly decoded, and the decoding result is valid. Judging, which can improve the average opinion score of subjective voice quality, while reducing the risk of noise.
  • FIG. 1 is a flowchart of a decoding method according to an embodiment of the present invention
  • FIG. 2 is a block diagram of a decoding apparatus according to an embodiment of the present invention.
  • FIG. 3 is a schematic flowchart of a convolutional coding performed by a transmitting end according to another embodiment of the present invention
  • FIG. 4 is a flowchart of a decoding method according to another embodiment of the present invention
  • FIG. 5 is a flowchart of a method for performing correction by a speech frame repair decoder according to another embodiment of the present invention.
  • FIG. 6 is a block diagram of a decoding apparatus according to another embodiment of the present invention.
  • the technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. example. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts are protected by the present invention. Scope.
  • the embodiment of the present invention provides a decoding method.
  • the method can be a base station, or a user terminal, or another relay device.
  • the embodiment of the present invention is not limited. As shown in FIG. 1 , the method includes:
  • Step 101 Decode the first substream of the demodulated soft value by using a decoder based on the error detection code auxiliary decision to obtain a first decoding result.
  • the error detection code may be a Cyclic Redundancy Check (CRC) error detection code, but the error detection code in the embodiment of the present invention is not limited to the CRC error detection code, and may also be Other error detection codes.
  • the decoder based on the error detection code assisted decision includes at least an enumeration Viterbi decoding algorithm LVA decoder, or a bit inversion BF decoder.
  • the first decoding result includes: an error detection code verification result, an error detection code verification result corresponding to the optimal sequence, and a decoding bit sequence.
  • the first substream of the demodulated soft value is decoded by a decoder based on the error detection code auxiliary decision, and the decoded bit sequence in the decoding result is subjected to soft value estimation to obtain a first decoding. result.
  • Manner 1 Determine the demodulated soft value corresponding to the decoding bit sequence in the decoding result, and obtain the mean value of the demodulated soft value ranges closest to the medium distance target bit according to the demodulated soft value Determining a soft value magnitude of the target bit; or
  • Manner 2 Determine an optimal decoding bit sequence, and determine a log likelihood ratio LLR soft value according to the optimal decoding bit sequence.
  • the first substream is decoded by the decoder based on the error detection code auxiliary decision, and the second substream and the third substream of the demodulated soft value are used.
  • the channel decoder is used for decoding, and the third decoding result and the fourth decoding result are respectively obtained.
  • the channel decoder includes at least a Viterbi decoding algorithm VA and a maximum logarithmic maximum a posteriori MLP decoding algorithm.
  • VA Viterbi decoding algorithm
  • the voice bits are divided into a first substream, a second substream, and a third substream according to bit importance, wherein the first substream is the most important substream, usually added with an error detection code. Protection, while the second substream and the third substream are not subjected to error detection code protection, and the error detection code code is used to detect whether the received bit stream is correct.
  • BFI Bad Frame Index
  • Step 102 Perform the first decoding result by using a speech frame repair decoder to obtain a second decoding result.
  • Step 103 Perform correction and validity determination on the parameters in the second decoding result, and send the correction result to the voice decoder.
  • the virtual source decoder is used to recover the voice parameter from the decoding result at the physical layer, where the decoding result includes: the second decoding result, or the third decoding result and the At least one decoding result of the fourth decoding result and the second decoding result;
  • Determining the unreasonable parameter in the speech parameter by using a parameter-level error concealment processor, and modifying the unreasonable parameter; encoding the modified speech parameter to obtain a speech coding bit; and indicating the speech coding bit and the error frame BFI
  • the signal is sent to the application layer speech decoder.
  • the base station receiver performs the error detection code verification result after decoding the PLVA through the lub interface between the base station and the radio network controller (RNC). Sending to the RNC, so that the RC performs outer loop power control according to the error detection code check result.
  • RNC radio network controller
  • Embodiments of the present invention provide a decoding method, which performs joint decoding by using a decoder based on an error detection code auxiliary decision and a speech frame repair decoder, and performs validity judgment on the decoding result, thereby improving subjective speech. The average quality opinion is divided, while reducing the risk of noise.
  • the embodiment of the present invention further provides a decoding device for implementing the steps and methods in the foregoing method embodiments, where the device may be a base station, or a user terminal, or another relay device, which is not limited by the embodiment of the present invention.
  • the decoding apparatus comprises: a decoder 201 based on an error detection code assisted decision, a speech frame repair decoder 202, a decoding validity judging module 203;
  • the decoder 201 which is based on the error detection code auxiliary decision, is configured to decode the first substream of the demodulated soft value to obtain a first decoding result;
  • the error detection code may be a CRC error detection code of the Viterbi decoding algorithm.
  • the error detection code in the embodiment of the present invention is not limited to the CRC error detection code, and may be other error detection codes.
  • the decoder based on the error detection code assisted decision is an enumeration Viterbi decoding algorithm LVA decoder, or a bit inverse BF decoder.
  • the first decoding result includes: an error detection code verification result, an error detection code verification result corresponding to the optimal sequence, and a decoding bit sequence.
  • the decoder based on the error detection code assisted decision is specifically configured to: decode the first substream of the demodulated soft value, and perform soft value estimation on the decoded bit sequence in the decoding result to obtain a first decoding result; or the decoder based on the error detection code auxiliary decision, configured to decode the first substream of the demodulated soft value; the apparatus further includes: decoding soft value estimation And a module, configured to perform soft value estimation on the decoded bit sequence in the decoding result, to obtain a first decoding result.
  • the first decoding soft value estimating unit is configured to determine the demodulated soft value corresponding to the decoding bit sequence in the decoding result. Determining, according to the demodulated soft value, an average of a plurality of demodulation soft value amplitudes of the medium distance target bit, determining a soft value range of the target bit; or, a second decoding soft value estimating unit, An optimal decoded bit sequence is determined, and a log likelihood ratio LLR soft value is determined based on the optimal decoded bit sequence.
  • the speech frame repair decoder 202 is configured to correct the first decoding result to obtain a second decoding result.
  • the decoding validity judging module 203 is configured to correct the parameters in the second decoding result, and send the correction result to the speech decoder.
  • the channel decoder when the decoder based on the error detection code assisted decision performs decoding, the channel decoder is configured to decode the second substream and the third substream of the demodulated soft value.
  • the third decoding result and the fourth decoding result are respectively obtained.
  • the channel decoder includes at least a Viterbi decoding algorithm VA and a maximum logarithmic maximum a posteriori MLP decoding algorithm.
  • the device when the device is a base station, the device further includes: a radio network controller, configured to receive the error detection code in the first decoding result sent by the decoder based on the error detection code auxiliary decision The verification result is performed by the outer loop power control according to the error detection code verification result.
  • a radio network controller configured to receive the error detection code in the first decoding result sent by the decoder based on the error detection code auxiliary decision The verification result is performed by the outer loop power control according to the error detection code verification result.
  • the apparatus provided in this embodiment may include another voice frame repair decoder for performing the functions of the voice frame repair decoder 202 and the decoding validity determining module 203 in this embodiment.
  • the first decoding result is corrected to obtain a second decoding result, the parameters in the second decoding result are corrected, and the correction result is used to recover the speech waveform by using a speech decoder.
  • An embodiment of the present invention provides a decoding apparatus, configured to decode a first substream of a demodulated soft value by using a decoder based on an error detection code to obtain a first decoding result; a repair decoder, configured to modify the first decoding result to obtain a second decoding result; and a decoding validity determining module, configured to correct a parameter in the second decoding result, and
  • the correction result uses a speech decoder to recover the speech waveform, thereby improving the subjective speech quality average opinion score while reducing the noise risk.
  • the embodiment of the present invention is described by taking the voice transmission of the uplink of the UMTS system as an example, wherein the transmitting end is as shown in FIG. 3, and the universal mobile communication specified by the 3rd Generation Partnership Project (3GPP) protocol In the uplink universal transmitter in the Universal Mobile Telecommunications System (UMTS): according to the voice bits
  • 3GPP 3rd Generation Partnership Project
  • the importance of the three substreams is A, B, and C, which respectively correspond to the first substream, the second substream, and the third substream in the embodiment of the present invention, and joins the A substream bits.
  • the error detection code check bits are protected, and the B and C substreams are not added with the error detection code check bits, and are respectively channel coded by the convolutional code encoder, and then the encoded voice bits are sent to the receiving end.
  • the receiving end After receiving the encoded voice bit, the receiving end performs decoding.
  • the embodiment of the present invention provides a decoding method, and the execution subject is described in detail for the base station. As shown in FIG. 4, the method includes: Step 401: The first substream of the adjusted soft value is decoded by a decoder based on the error detection code auxiliary decision to obtain a first decoding result.
  • the error detection code may be a CRC error detection algorithm for the enumeration Viterbi decoding algorithm.
  • the code but the error detection code in the embodiment of the present invention is not limited to the CRC error detection code, and may be other error detection codes.
  • the decoder based on the error detection code assisted decision includes at least an enumeration Viterbi decoding algorithm LVA decoder, or a bit inversion BF decoder.
  • the LVA decoder can use the parallel enumeration Viterbi decoding algorithm PLVA or the serial enumeration Viterbi decoding algorithm SLVA.
  • the embodiment of the present invention is described in detail by an LVA decoder, wherein the LVA decoder uses PLVA.
  • Step 402 Decode the second substream and the third substream of the demodulated soft value by using a channel decoder to obtain a third decoding result and a fourth decoding result respectively.
  • step 401 is performed simultaneously with step 402.
  • the first substream, the second substream, and the third substream are substreams divided according to voice bit importance from high to low;
  • the channel decoder includes at least a Viterbi decoding algorithm VA, and a maximum logarithm is maximum A posteriori MLP decoding algorithm.
  • the first substream is decoded by using a PLVA decoder to obtain a first decoding result, where the first decoding result includes: an error detection code verification result, and an error detection code verification corresponding to the optimal sequence The result and the decoded bit sequence.
  • the error detection code check result indicates the success or failure of the output bit.
  • the output bit "0" indicates “success” and the output bit " ⁇ indicates” failure.
  • LVA is generalized on the basis of VA.
  • VA only outputs an optimal sequence and The sequence performs error detection code verification, and the LVA outputs the first L optimal sequences, and performs error detection code verification on the L sequences according to the order of likelihood, wherein the most A sequence may refer to a sequence with the greatest likelihood, and may also refer to any one of the top L optimal sequences. Only when the error detection code check of the L sequence does not pass, it is considered that the error detection code of the current frame is incorrectly verified, and the decoding fails.
  • the decoded bit sequence refers to a binary hard decision bit that is output when the PLVA decoder is used.
  • the frame that has not passed the error detection code check in the first substream is further processed by the speech frame repair decoder. Further corrections. Since the speech frame repair decoder needs to know the reliability of the hard decision bits output by the channel decoder, that is, obtain the soft value of each decoding bit, where the soft value is obtained by the hard decision bit, specifically, the soft value is the PLVA translation.
  • the Log-likelihood Ratio (LLR) corresponding to the hard decision bit of the encoder output, so it is necessary to further process the binary hard decision bits outputted by the PLVA decoder, that is, the decoding result
  • the decoded bit sequence is subjected to soft value estimation to obtain a first decoding result.
  • a separate decoding soft value estimation module may be used, or a PLVA decoder may be used.
  • the decoding soft value estimation module is integrated in the essence. In the PLVA decoder.
  • the decoding soft value can be obtained in two ways: engineering approximation and maximum logarithmic maximum a posteriori probability (Max-Log-MAP, MLP) decoding method.
  • the engineering approximation method according to the change of the channel in a short time, the demodulated soft value corresponding to the decoded bit sequence in the decoding result may be determined, and the medium distance is obtained according to the demodulated soft value.
  • the engineering approximation is described by taking a 1/2 convolutional code as an example. Let ⁇ (0), ⁇ (1), ⁇ , ⁇ , - ⁇ denote the decoded bit sequence to be estimated output by the PLVA decoder, and the corresponding demodulated soft value sequence adopts PLVA.
  • the demodulated soft value before decoding by the decoder is expressed as ⁇ .
  • the corresponding output decoded soft value sequence is represented as (0), i (1), ⁇ ⁇ , ⁇ ( , - ⁇ , where c marks the L value as the translation estimated by the decoded soft value estimation module
  • the soft value of the code Specifically, the decoded soft value can be obtained according to the average of the last five demodulated soft values, and the absolute value can be approximated as: + + 1) 0 )
  • the sign of the code is equal to the sign of the input decoded bit sequence, for example,
  • the MLP decoding method is to determine the optimal decoding bit sequence and determine the log likelihood ratio LLR soft value based on the optimal decoding bit sequence.
  • the soft value amplitude corresponding to the PLVA optimal path is considered to be the soft value amplitude of the PLVA decoding output result.
  • an MLP decoder which can directly output the decoded bit output LLR soft value of the optimal sequence, that is, the decoded bit output LLR soft value of the VA, and determine the absolute value of the decoding bit of the optimal sequence as the PLVA decoding output.
  • the magnitude of the soft value determines the sign of the soft value based on the sign of the hard decision bit obtained by the PLVA decoder.
  • the speech frame repair decoder only corrects the frame when the PLVA decoding fails, that is, when the frame of the decoding bit with the error detection code check is 1 is corrected, the output decoding result when the PLVA decoding fails is The VA decoding result is the same, so the soft value of the input speech frame repair decoder can be regarded as the soft value of the PLVA decoding output.
  • the soft values utilized in the speech frame repair decoder start parameter correction are accurate.
  • Step 403 Perform the first decoding result by using a voice frame repair decoder to obtain a second decoding result.
  • the frame that fails the error detection code check in the first decoding result that performs the soft value estimation is further corrected by using a voice frame repair decoder.
  • the voice frame repair decoder may use a Soft Bit Source Decoding (SBSD) algorithm or an iterative Source-Channel Decoding (ISCD) algorithm. Iterative or iterative algorithm.
  • SBSD Soft Bit Source Decoding
  • ISCD iterative Source-Channel Decoding
  • Step 501 Acquire an input decoding soft value and a decoding bit sequence, and determine whether the decoding bit sequence can be detected by the error detecting code.
  • step 505 is performed.
  • step 502 is performed; optionally, the decoding soft value may be a decoding bit sequence, and k is a time label when receiving the decoding bit sequence, i Represents the ith bit in the decoded bit sequence.
  • Step 502 Calculate a bit error probability according to the decoded soft value.
  • bit error probability is calculated according to 1 + ex Pl L ((0)1, where W is the bit error probability.
  • Step 503 determining a parameter level transition probability according to the calculated bit error probability; according to the calculated bit error error probability, according to
  • Step 504 Calculate a posterior probability according to a parameter level transition probability.
  • the parameter-level transition probability may be a transition probability calculated in a previous frame, or may be a transition probability obtained by calculating a large number of training samples.
  • the posterior probability of the decoding is
  • Step 505 Perform a maximum A Posteriori (MAP) estimation according to the posterior probability, and obtain a second decoding result.
  • MAP maximum A Posteriori
  • Step 505 obtain a second decoding result according to the ⁇ , where the parameter V represents the second The result of the decoding.
  • step 403 only the first decoding result is further modified, and the third decoding result obtained in step 402 and the fourth decoding result need not be performed.
  • Step 404 Perform correction and validity judgment on the parameters in the decoding result.
  • the decoding result in this step includes a second decoding result, a third decoding result, and a fourth decoding result, where only the second decoding result may be further corrected, or the second decoding result and the second decoding result may be
  • the third decoding result is further modified, or the second decoding result and the fourth decoding result are further modified, or the second decoding result, the third decoding result, and the fourth decoding result are further corrected.
  • the decoding validity judgment module is used to correct the parameters in the decoding result and have a message judgment.
  • the decoding validity determining module may include a virtual source decoder (VSD) and a parameter-level error concealment (PEC);
  • the decoding result is restored to the speech parameters at the physical layer, and the upper layer speech decoder is not needed to match the existing system architecture.
  • the PEC is used to determine an unreasonable parameter in the voice parameter, and the unreasonable parameter is corrected. For example, when the posterior probability of a parameter is less than a preset threshold, the parameter is identified as an unreasonable parameter, and may also be referred to as bad. parameter. For the unreasonable parameters, error concealment correction is needed. There are many specific error concealment correction algorithms. The present invention does not limit the specific parameter level error concealment algorithm. For example, an error concealment mechanism based on BFI for full frame discarding of error frames may be employed.
  • the speech coded frame processed by the decoding validity judgment module can be regarded as a "reasonable frame", and the BFI setting can be used to indicate that the frame is a "reasonable frame";
  • Step 405 Send the correction result to the voice decoder, and the voice decoder recovers the voice waveform. And encoding the corrected speech parameters to obtain speech coded bits; transmitting the speech coded bits and the error frame indication BFI signal to the application layer speech decoder, and the speech decoder according to the received, the speech data, and the corresponding BFI indication , the voice signal can be recovered.
  • it can be AMR voice.
  • the NodeB receiver will separately translate the first one.
  • the error detection code verification result in the code result is transmitted to the RNC.
  • the RNC uses the error detection code verification result to perform outer loop power control.
  • no power control is required.
  • the MOS decoder is used separately, and the PLVA decoding and the VA joint speech frame repair decoder are used separately, and the MOS performance is improved.
  • a decoding method provided by an embodiment of the present invention decodes a speech frame by using an LVA and a speech frame repair decoder, and then performs validity judgment on the decoding result, and corrects an unreasonable parameter, and finally Sending the result to the speech decoder to complete the recovery of the speech signal can improve the subjective speech quality average opinion score while reducing the noise risk.
  • the embodiment of the present invention provides a decoding apparatus, which may be a base station, a user terminal, or another relay device, which is not limited in the embodiment of the present invention.
  • the device comprises: a decoder 701 based on the error detection code auxiliary decision, a speech frame repair decoder 702, a decoding validity judging module 703, a virtual source decoder 7031, a parameter level error concealing processor 7032, and a decoding soft a value estimation module 704, a first decoding soft value estimating unit 7041, a second decoding soft value estimating unit 7042, a channel decoder 705, a radio network controller 706;
  • the decoder 701 is configured to perform decoding on the first substream of the demodulated soft value to obtain a first decoding result.
  • the error detection code may be a CRC error detection code of the Viterbi decoding algorithm.
  • the error detection code in the embodiment of the present invention is not limited to the CRC error detection code, and may be other error detection codes.
  • the decoder based on the error detection code assisted decision is an enumeration Viterbi decoding algorithm LVA decoder, or a bit inverse BF decoder.
  • the first decoding result includes: an error detection code verification result, an error detection code verification result corresponding to the optimal sequence, and a decoding bit sequence.
  • the decoder 701 based on the error detection code auxiliary decision is specifically configured to: decode the first substream of the demodulated soft value, and perform soft value estimation on the decoded bit sequence in the decoding result, Obtaining the first decoding result; or
  • the decoder 701 is configured to decode the first substream in the demodulated soft value
  • the decoded soft value estimation module 704 is configured to decode the decoded bit in the decoding result.
  • the sequence is subjected to soft value estimation to obtain a first decoding result.
  • the decoder 701 or the decoded soft value estimation module 704 based on the error detection code auxiliary decision further includes: a first decoding soft value, when performing soft value estimation on the decoding bit sequence in the decoding result Estimation unit 7041 or second decoding soft value estimation unit 7042;
  • the first decoding soft value estimating unit 7041 is configured to determine the demodulated soft value corresponding to the decoding bit sequence in the decoding result, and obtain the closest distance to the target bit according to the demodulated soft value. a plurality of demodulation soft value amplitudes, determining a soft value magnitude of the target bit; or, a second decoding soft value estimating unit 7042, configured to determine an optimal decoding bit sequence, and according to the optimal decoding The bit sequence determines the log likelihood ratio LLR soft value. Further, when the decoder 701 based on the error detection code assisted decision decodes the first substream, the channel decoder 705 is configured to use the second substream and the third substream of the demodulated soft values.
  • the channel decoder 705 includes at least a Viterbi decoding algorithm VA and a maximum logarithmic maximum a posteriori MLP decoding algorithm.
  • the speech frame repair decoder 702 is configured to modify the first decoding result to obtain a second decoding result. Specifically, the frame that fails the error detection code check in the first decoding result is corrected.
  • the decoding validity determining module 703 is configured to correct the parameters in the second decoding result, and send the correction result to the voice decoder.
  • the decoding validity determining module 703 is specifically configured to: correct an unreasonable parameter in the second decoding result; or, the third decoding result and the fourth decoding result At least one of the decoding results and the unreasonable parameters of the second decoding result are corrected.
  • the decoding validity determining module 703 includes: a virtual source decoder 7031, a parameter level error concealing processor 7032;
  • a virtual source decoder 7031 configured to recover a speech parameter at a physical layer, where the decoding result includes: the second decoding result, or the third decoding result and the fourth At least one decoding result in the decoding result and the second decoding result;
  • a parameter-level error concealing processor 7032 configured to determine an unreasonable parameter in the speech parameter, and correct the unreasonable parameter; encode the corrected speech parameter to obtain a speech encoding bit; and encode the speech encoding bit and error
  • the frame indicates that the BFI signal is sent to the application layer speech decoder.
  • the method further includes: a radio network controller 706, configured to receive an error detection code verification result in the first decoding result sent by the LVA decoder, The outer loop power control is performed according to the error detection code verification result.
  • the apparatus provided in this embodiment may include another voice frame repair decoder for performing the voice frame repair decoder 702 and the decoding validity determining module 703 in this embodiment.
  • the function is to correct the first decoding result, obtain a second decoding result, correct the parameters in the second decoding result, and use the speech decoder to recover the speech waveform.
  • An embodiment of the present invention provides a decoding apparatus, configured to decode a first substream of a demodulated soft value by using a decoder based on an error detection code to obtain a first decoding result; a repair decoder, configured to modify the first decoding result to obtain a second decoding result; and a decoding validity determining module, configured to correct a parameter in the second decoding result, and
  • the correction result uses a speech decoder to recover the speech waveform, thereby improving the subjective speech quality average opinion score while reducing the noise risk.
  • the solution provided by the embodiment of the present invention is suitable for all systems that use the channel coding for voice transmission and provides error protection, where the channel coding may use a convolutional code, a turbo code, a turbo code, etc., in addition to the embodiment of the present invention.
  • the channel coding may use a convolutional code, a turbo code, a turbo code, etc., in addition to the embodiment of the present invention.
  • decoding not limited to wideband AMR speech coding, enhanced full rate (EFR) speech coding, and the like.
  • Computer readable media includes both computer storage media and communication media including any medium that facilitates transfer of a computer program from one location to another.
  • a storage medium may be any available media that can be accessed by a computer.
  • computer readable media may comprise RAM. ROM, EEPROM, CD-ROM or other optical disk storage, disk storage media or other magnetic storage device, or can be used for carrying or storing in the form of an instruction or data structure.
  • the desired program code and any other medium that can be accessed by the computer may suitably be a computer readable medium.
  • the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable , fiber optic cable, twisted pair, DSL or such as infrared, Wireless technologies such as wireless and microwaves are included in the fixing of the associated medium.
  • coaxial cable, fiber optic cable, twisted pair, DSL or such as infrared, Wireless technologies such as wireless and microwaves are included in the fixing of the associated medium.
  • a disk and a disc include a compact disc (CD), a laser disc, a disc, a digital versatile disc (DVD), a floppy disc, and a Blu-ray disc, wherein the disc is usually magnetically copied, and the disc is The laser is used to optically replicate the data. Combinations of the above should also be included within the scope of the computer readable media.

Landscapes

  • Physics & Mathematics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Error Detection And Correction (AREA)
  • Detection And Prevention Of Errors In Transmission (AREA)

Abstract

本发明公开一种译码方法及装置,涉及通信网络技术领域,可以提高主观语音质量平均意见分,同时降低杂音风险。本发明实施例通过将解调后的软值中第一子流采用基于检错码辅助判决的译码器进行译码,获得第一译码结果;将所述第一译码结果采用语音帧修复译码器进行修正,获得第二译码结果;将所述第二译码结果中的参数进行修正,并将修正结果发送给语音译码器。本发明实施例提供的方案适于所有利用信道编码进行语音传输并提供差错保护的系统,本发明实施例包括并不限于AMR、宽带AMR以及增强型全速率语音编码的译码。

Description

一种译码方法及装置 本申请要求于 2012 年 11 月 15 日提交中国专利局、 申请号为 201210460305.1、发明名称为"一种译码方法及装置 "的中国专利申请的优先 权, 其全部内容通过引用结合在本申请中。 技术领域 本发明涉及通信网络技术领域, 尤其涉及一种译码方法及装置。 背景技术 语音通信是当前通信系统中最基本、 最重要的方式之一。 当前语音传 输所采用的信道编码多为卷积码。 对卷积码的译码通常采用维特比译码算 法 ( Viterbi Algorithm, VA )进行译码。 然而, 单独利用 VA进行译码没有 充分利用检错码校验所提供的信息量, 恢复出的语音质量也较差。
另夕卜, 还可以利用列举维特比译码算法 (List Viterbi Algorithm, LVA ) 对卷积码进行译码。 其中 LVA包括并行列举维特比译码算法 (Parallel List Viterbi Algorithm, PLVA )、 串行列举维特比译码算法 (Serial List Viterbi Algorithm, SLVA )。 通过釆用 LVA译码后, 会输出前 L条最优译码序列, 并依照似然度从大到小的顺序对这 L条译码序列进行检错码校验, 直到某 条译码序列的检错码校验全部通过为止。 当 L条译码序列的检错码校验均 不通过时, 认为对当前帧的检错码校验错误, 译码失败。
采用 VA以及 LVA进行译码, 均为针对信道编码而进行的信道译码。 由于速率自适应 ( Adaptive Muti-Rate , AMR )语音编码本身相邻帧的参数 之间存在着冗余信息量, 因此可以利用信源相邻帧间的冗余性,从而构成可 以提升译码性能的语音帧修复译码器。
然而, 根据现有技术采用 LVA进行译码时, 会造成漏检概率的增大, 导致引入较大的杂音风险, 也未能充分利用信源的冗余性。 采用语音帧修 复译码器进行译码时, 可以提升误比特率(Bit Error Rate, BER )性能, 但 是对误块率(BLER Block Error Rate, BLER )的提升却非常有限, 对主观 语音质量平均意见分(Mean Opinion Score, MOS ) 的提升也非常有限。 发明内容 有鉴于此, 本发明的实施例提供一种译码方法及装置, 可以提高主观 语音质量平均意见分, 同时降低杂音风险。
第一方面, 提供了一种译码方法, 包括:
将解调后的软值中第一子流采用基于检错码辅助判决的译码器进行译 码, 获得第一译码结果;
将所述第一译码结果釆用语音帧修复译码器进行修正, 获得第二译码 结果;
将所述第二译码结果中的参数进行修正, 并将修正结果发送给语音译 码器。
第二方面, 提供了一种译码装置, 包括:
基于检错码辅助判决的译码器, 用于将解调后的软值中第一子流进行 译码, 获得第一译码结果;
语音帧修复译码器, 用于将所述第一译码结果进行修正, 获得第二译 码结果;
译码有效性判断模块, 用于将所述第二译码结果中的参数进行修正, 并将修正结果发送给语音译码器。
本发明实施例提供一种译码方法及装置, 通过将解调后的软值中第一 子流采用基于检错码辅助判决的译码器进行译码, 获得第一译码结果; 将 所述第一译码结果采用语音帧修复译码器进行修正, 获得第二译码结果; 将所述第二译码结果中的参数进行修正, 并将修正结果釆用语音译码器恢 复出语音波形。 与现有技术中采用 LVA进行译码时, 会造成漏检概率的增 大, 即引入较大的杂音风险, 也未能充分利用信源的冗余性, 采用语音帧 修复译码器进行译码时, 可以提升 BER性能, 但是对 BLER的提升却非常 有限, 对主观语音质量 MOS的提升也非常有限相比, 本发明实施例提供的 方案通过采用基于检错码辅助判决的译码器和语音帧修复译码器进行联合 译码, 并对译码结果进行有效性判断, 从而可以提高主观语音质量平均意 见分, 同时降低杂音风险。 附图说明 为了更清楚地说明本发明实施例或现有技术中的技术方案, 下面将对 实施例或现有技术描述中所需要使用的附图作简单地介绍, 显而易见地, 下面描述中的附图仅仅是本发明的一些实施例, 对于本领域普通技术人员 来讲, 在不付出创造性劳动性的前提下, 还可以根据这些附图获得其他的 附图。
图 1为本发明实施例提供的一种译码方法的流程图;
图 2为本发明实施例提供的一种译码装置的框图;
图 3为本发明另一实施例提供的发送端进行卷积编码的流程示意图; 图 4为本发明另一实施例提供的一种译码方法的流程图;
图 5 为本发明另一实施例提供的语音帧修复译码器执行修正的方法的 流程图;
图 6为本发明另一实施例提供的一种译码装置的框图。 具体实施方式 下面将结合本发明实施例中的附图, 对本发明实施例中的技术方案进 行清楚、 完整地描述, 显然, 所描述的实施例仅仅是本发明一部分实施例, 而不是全部的实施例。 基于本发明中的实施例, 本领域普通技术人员在没 有作出创造性劳动前提下所获得的所有其他实施例, 都属于本发明保护的 范围。
本发明实施例提供一种译码方法, 该方法的执行主体可以为基站, 或 者用户终端, 或者其他中继设备, 本发明实施例不做限制, 如图 1 所示, 该方法包括:
步骤 101,将解调后的软值中第一子流采用基于检错码辅助判决的译码 器进行译码, 获得第一译码结果;
其中 , 检错码可以为列举维特比译码算法循环冗余校验 ( Cyclic Redundancy Check, CRC )检错码, 但本发明实施例中的检错码并不限于 CRC检错码, 还可以为其他的检错码。 所述基于检错码辅助判决的译码器 至少包括列举维特比译码算法 LVA译码器, 或者比特反转 BF译码器。
第一译码结果包括: 检错码校验结果, 最优序列对应的检错码校验结 果和译码比特序列。
进一步的, 将解调后的软值中第一子流釆用基于检错码辅助判决的译 码器进行译码, 对译码结果中译码比特序列进行软值估计, 获得第一译码 结果。
任一种:
方式一: 确定译码结果中所述译码比特序列对应的所述解调后的软值, 根据所述解调后的软值获取中距离目标比特最近的若干个解调软值幅度的 均值, 确定所述目标比特的软值幅度; 或者,
方式二: 确定最优译码比特序列, 并根据所迷最优译码比特序列确定 对数似然比 LLR软值。
进一步的, 在对解调后的软值中第一子流采用基于检错码辅助判决的 译码器进行译码的同时, 将解调后的软值中第二子流、 第三子流采用信道 译码器进行译码, 分别获得第三译码结果以及第四译码结果。 所述信道译 码器至少包括维特比译码算法 VA、 最大对数最大后验 MLP译码算法。 将语音比特按照比特重要性由高到低, 分为第一子流、 第二子流和第 三子流, 其中, 第一子流为最重要的子流, 通常加上检错码码进行保护, 而第二子流和第三子流则没有进行检错码码保护, 检错码码用于检测接收 比特流是否正确。
例如, 若某个语音帧的第一子流检错码校验正确, 则认为该帧接收正 确, 错误帧指示 (Bad Frame Index, BFI ) 置为 0; 若检错码校验错误, 则 认为该帧接收错误, BFI置为 1。 BFI为单独设置的一个信号, 不包含在语 音帧中, 由物理层产生, 用于送给应用层语音译码器的, 以告知上层语音 译码器当前帧是否正确可用。
步骤 102, 将所述第一译码结果采用语音帧修复译码器进行修正, 获得 第二译码结果;
步骤 103, 将所述第二译码结果中的参数进行修正以及有效性判断, 并 将修正结果发送给语音译码器。
进一步的, 将所述第二译码结果中不合理的参数进行修正; 或者, 将所述第三译码结果与所述第四译码结果中的至少一个译码结果以及 所述第二译码结果中不合理的参数进行修正以及有效性判断。
可选的, 采用虚拟信源译码器, 将译码结果在物理层恢复出语音参数, 所述译码结果包括: 所述第二译码结果, 或者所迷第三译码结果与所述第 四译码结果中的至少一个译码结果以及所述第二译码结果;
采用参数级错误隐藏处理器确定所述语音参数中不合理参数, 将所述 不合理参数进行修正; 将修正后的语音参数进行编码获得语音编码比特; 将所述语音编码比特与错帧指示 BFI信号发送给应用层语音译码器。
需要说明的是, 当本方法的执行主体为基站时, 通过基站和无线网络 控制器 ( Radio Network Controller, RNC )之间的 lub接口, 基站接收机将 PLVA译码后的检错码校验结果发送给 RNC, 以便所述 R C根据所述检错 码校验结果进行外环功控。 本发明实施例提供一种译码方法, 通过采用基于检错码辅助判决的译 码器和语音帧修复译码器进行联合译码, 并对译码结果进行有效性判断, 从而可以提高主观语音质量平均意见分, 同时降低杂音风险。
本发明实施例进一步给出实现上述方法实施例中各步骤及方法的一种 译码装置, 该装置可以为基站, 或者用户终端, 或者其他中继设备, 本发 明实施例不做限制, 如图 2所示, 该译码装置包括: 基于检错码辅助判决 的译码器 201 , 语音帧修复译码器 202, 译码有效性判断模块 203;
基于检错码辅助判决的译码器 201 ,用于将解调后的软值中第一子流进 行译码, 获得第一译码结果;
可选的, 检错码可以为列举维特比译码算法 CRC检错码 , 但本发明实 施例中的检错码并不限于 CRC检错码, 还可以为其他的检错码。 所述基于 检错码辅助判决的译码器为列举维特比译码算法 LVA译码器, 或者比特反 转 BF译码器。
其中, 所述第一译码结果包括: 检错码校验结果, 最优序列对应的检 错码校验结果和译码比特序列。
进一步的, 所述基于检错码辅助判决的译码器具体用于: 将解调后的 软值中第一子流进行译码, 对译码结果中译码比特序列进行软值估计, 获 得第一译码结果; 或者, 所述基于检错码辅助判决的译码器, 用于将解调 后的软值中第一子流进行译码; 所述装置还包括: 译码软值估计模块, 用 于对译码结果中译码比特序列进行软值估计 , 获得第一译码结果。
进一步的, 对译码结果中译码比特序列进行软值估计时, 第一译码软 值估计单元, 用于确定译码结果中所述译码比特序列对应的所述解调后的 软值, 根据所述解调后的软值获取中距离目标比特最近的若干个解调软值 幅度的均值, 确定所述目标比特的软值幅度; 或者, 第二译码软值估计单 元, 用于确定最优译码比特序列, 并根据所述最优译码比特序列确定对数 似然比 LLR软值。 语音帧修复译码器 202, 用于将所述第一译码结果进行修正, 获得第二 译码结果;
译码有效性判断模块 203 , 用于将所述第二译码结果中的参数进行修 正, 并将修正结果发送给语音译码器。
需要说明的是, 当基于检错码辅助判决的译码器进行译码的同时, 信 道译码器, 用于将解调后的软值中第二子流、 第三子流进行译码, 分别获 得第三译码结果以及第四译码结果。 所述信道译码器至少包括维特比译码 算法 VA、 最大对数最大后验 MLP译码算法。
进一步的, 当所述装置为基站时, 所述装置还包括: 无线网络控制器, 用于接收所述基于检错码辅助判决的译码器发送的所述第一译码结果中检 错码校验结果, 根据所述检错码校验结果进行外环功控。
需要说明的是, 本实施例提供的装置中可以包括另一语音帧修复译码 器,用于执行本实施例中语音帧修复译码器 202和译码有效性判断模块 203 的功能, 即将所述第一译码结果进行修正, 获得第二译码结果, 将所述第 二译码结果中的参数进行修正, 并将修正结果釆用语音译码器恢复出语音 波形。
本发明实施例提供一种译码装置, 通过基于检错码辅助判决的译码器, 用于将解调后的软值中第一子流进行译码, 获得第一译码结果; 语音帧修 复译码器, 用于将所述第一译码结果进行修正, 获得第二译码结果; 译码 有效性判断模块, 用于将所述第二译码结果中的参数进行修正, 并将修正 结果采用语音译码器恢复出语音波形, 从而可以提高主观语音质量平均意 见分, 同时降低杂音风险。
本发明实施例以 UMTS系统上行链路的语音传输为例进行说明,其中, 发送端如图 3所示, 第三代合作伙伴计划 (The 3rd Generation Partnership Project , 3 GPP ) 协议规定的通用移动通讯系统 ( Universal Mobile Telecommunications System, UMTS ) 中上行通用发送端中: 按照语音比特 的重要性由高到低, 分为 A、 B、 C三个子流, 即分别对应本发明实施例中 的第一子流、 第二子流及第三子流, 并对 A子流比特加入检错码校验比特 进行保护, B、 C子流没有加入检错码校验比特, 分别通过卷积码编码器进 行信道编码, 然后将编码后的语音比特发送给接收端。
接收端接收到编码后的语音比特后, 进行译码, 本发明实施例提供一 种译码方法, 以执行主体为基站进行详细描述, 如图 4所示, 该方法包括: 步骤 401 ,将解调后的软值中第一子流采用基于检错码辅助判决的译码 器进行译码, 获得第一译码结果; 可选的, 检错码可以为列举维特比译码算法 CRC检错码, 但本发明实 施例中的检错码并不限于 CRC检错码, 还可以为其他的检错码。 所述基于 检错码辅助判决的译码器至少包括列举维特比译码算法 LVA译码器, 或者 比特反转 BF译码器。 LVA译码器可以釆用并行列举维特比译码算法 PLVA 或者串行列举维特比译码算法 SLVA。 本发明实施例以 LVA译码器进行详 细描述, 其中, LVA译码器采用 PLVA。
步驟 402, 将解调后的软值中第二子流、 第三子流采用信道译码器进行 译码, 分别获得第三译码结果以及第四译码结果;
需要说明的是, 步骤 401与步骤 402同时执行。
其中第一子流、 第二子流、 第三子流为按照语音比特重要性由高到低 进行划分的子流; 所述信道译码器至少包括维特比译码算法 VA、 最大对数 最大后验 MLP译码算法。
可选的, 第一子流采用 PLVA译码器进行译码, 获得第一译码结果, 其 中, 第一译码结果包括: 检错码校验结果, 最优序列对应的检错码校验结 果和译码比特序列。
具体的, 检错码校验结果表示输出的比特的成功或者失败, 例如, 输 出比特 "0" 表示 "成功 ", 输出比特 "Γ 表示 "失败,,。
LVA是在 VA的基础上推广得到的。 VA只输出一条最优序列, 并对该 序列进行检错码校验, 而 LVA会输出前 L条最优序列, 并依照似然度从大 到小的顺序对这 L条序列进行检错码校验, 其中, 本实施例中的最有序列 可以指似然度最大的序列, 也可以指前 L条最优序列中的任意一个序列。 只有 L条序列的检错码校验均不通过时, 才认为对当前帧的检错码校验错 误, 译码失败。
译码比特序列指釆用 PLVA译码器时输出的一个个二进制的硬判决比 特。
进一步的, 将解调后的软值中第一子流采用 LVA译码器进行译码后, 对第一子流中未通过检错码校验的帧, 再通过语音帧修复译码器进行进一 步的修正。 由于语音帧修复译码器需要获知信道译码器输出的硬判决比特 的可靠度, 即获得每个译码比特的软值, 这里软值通过硬判决比特获得, 具体的, 软值即 PLVA 译码器输出的硬判决比特对应的对数似然比 ( Log-likelihood Ratio, LLR ), 因此需要将 PLVA译码器时输出的一个个二 进制的硬判决比特进行进一步的处理, 即对译码结果中译码比特序列进行 软值估计, 获得第一译码结果。 需要说明的是, 进行软值估计时可以采用 单独的译码软值估计模块执行, 也可以采用 PLVA译码器, 当采用 PLVA译 码器时, 其实质为将译码软值估计模块集成在 PLVA译码器中。
译码软值的获得可以有两种方式: 工程近似法和最大对数最大后验概 率 (Max-Log-MAP, MLP )译码法。
工程近似法: 根据信道在短时间的变化不大, 因此可以确定译码结果 中所述译码比特序列对应的所述解调后的软值, 根据所述解调后的软值获 取中距离目标比特最近的若干个解调软值幅度的均值, 确定所述目标比特 的软值幅度。 例如, 以 1/2 卷积码为例来描述该工程近似法。 设 {έ(0),ό(1),· ··,Κ ,-·}表示由 PLVA译码器输出的待估计的译码比特序列 , 其对 应的解调后的软值序列即采用 PLVA译码器进行译码之前的解调软值,表示 为 ^。。),/ ^ ,/ ^。),^!^),…, ^。),^ ),…》, 其中, M标记该 L值为解调 软值, 对应的输出译码软值序列表示为 (0),i (1), ·· ·,^ ( ,-} , 其中, c标记 该 L值为译码软值估计模块估计出的译码软值。具体的, 可以根据最近的 5 个解调软值的均值获取译码软值, 则 的绝对值可以近似表示为: + + 1)0)| + + 1 )| + 2)。)|)
Figure imgf000011_0001
的正负号即等于输入的译码比特序列对应的正负号, 例如才艮据
Figure imgf000011_0002
MLP译码法: 即确定最优译码比特序列, 并根据所述最优译码比特序 列确定对数似然比 LLR软值。 可选的, 将 PLVA最优路径所对应的软值幅 值, 认为是 PLVA译码输出结果的软值幅值。 也可以采用 MLP译码器, 可 以直接输出最优序列的译码比特输出 LLR软值,即 VA的译码比特输出 LLR 软值,确定最优序列的译码比特的绝对值为 PLVA译码输出软值的幅值,根 据 PLVA译码器获得的硬判决比特的符号确定软值的符号。
当语音帧修复译码器只对 PLVA译码失败时的帧做修正时,即对译码比 特的检错码校验为 1 的帧做修正时, PLVA译码失败时的输出译码结果与 VA译码结果是相同的, 因此其输入语音帧修复译码器的软值就可以认为是 PLVA译码输出的软值。 这样, 采用 MLP译码估计法, 就可以保证语音帧 修复译码器启动参数修正时所利用的软值是精确的。
步骤 403 , 将所述第一译码结果采用语音帧修复译码器进行修正, 获得 第二译码结果;
可选的, 将进行软值估计的第一译码结果中未通过检错码校验的帧采 用语音帧修复译码器进行进一步修正。 可选的, 语音帧修复译码器可以采 用包含软比特信源译码(Soft Bit Source Decoding, SBSD )算法、 迭代信源 信道译码 ( iterative Source-Channel Decoding, ISCD ) 算法在内各种非迭代 或者迭代的算法。 如图 5所示, 本实施例以 SBSD算法为例进行详细描述: 步驟 501,获取输入的译码软值以及译码比特序列, 并判断译码比特序 列是否可以通过检错码检测, 当译码比特序列可以通过检错码检测时, 则 执行步骤 505 ,当译码比特序列不可以通过检错码检测时 ,则执行步骤 502; 可选的, 译码软值可以为 译码比特序列可以为 其中, k 表示接收到此译码比特序列时的时刻标号, i表示译码比特序列中的第 i个 比特。
步骤 502, 根据译码软值计算比特错误概率;
Pe (bk ('·)) = 1—,—— . 可选的,根据 1 + exPlL ( (0)1计算比特错误概率,其中,驱 W 表示比特错误概率。
步驟 503, 根据计算的比特错误概率, 确定参数级转移概率; 根 据 计 算 的 比 特 错 误 概 率 , 按 照
[ Pe(bk(i)) ,其他 计 算比特级转移概率; 其中, 表示取值为 V时传输的比特; 根据一个参数所包含的全部比特的转移概率, 按照
^Jb^ p^i^))计算参数级转移概率; 其, N表示该参数 的位宽, 即该参数由 N个比特表示。
步骤 504, 根据参数级转移概率, 计算后验概率;
可选的, 参数级转移概率可以为上一帧中计算的转移概率, 也可以为 通过大量的训练样本计算获得的转移概率。
p(\>^\bk,vk_l)=c-p(bk ib^)∑ (b^ sm I ,, _2) 具体的,根据
Figure imgf000012_0001
计算 后验
概 率 ; 其 中 , c 表 示 归 一 化 因 子 ,
Figure imgf000013_0001
需要说明的是, 当步骤 501 中译码比特序列可以通过检错码检测时, 本次译码的后验概率为
(b(v) ^ y ) = \ 如果 (0 = bk ( ), = {0, 1, · · ·, N 1}
[o, 其他 可选的, 将本次译码的后验概率进行存储。
步驟 505 ,根据后验概率,执行最大后验( Maximum A Posteriori , MAP ) 估计, 获得第二译码结果; 可选的, 根据 ^ ^^获得第二译码结果, 其中, 参数 V 表示第二译码结果。
根据步骤 501-步驟 505的方法可以完成为通过检错码校验的帧的修正。 需要说明的是, 步骤 403中仅对第一译码结果进行进一步修正, 而步骤 402 获得的第三译码结果以及第四译码结果则不需要执行此步骤。
步骤 404 , 将译码结果中的参数进行修正以及有效性判断;
本步骤中的译码结果包括第二译码结果, 第三译码结果以及第四译码 结果, 其中, 可以仅对第二译码结果进行进一步修正, 也可以对第二译码 结果和第三译码结果进行进一步修正, 或者对第二译码结果和第四译码结 果进行进一步修正, 或者对第二译码结果、 第三译码结果和第四译码结果 进行进一步修正。
可选的, 采用译码有效性判断模块对译码结果中的参数进行修正以及 有消息判断。具体的,译码有效性判断模块可以包括虚拟信源译码器( Virtual source decoder, VSD ) 以及参数级错误隐藏 ( Parameter-level Error Concealment, PEC );
采用 VSD, 将译码结果在物理层恢复出语音参数, 不需要调用上层的 语音译码器, 从而与现有系统架构相匹配。 采用 PEC, 确定所述语音参数中不合理参数, 将所述不合理参数进行 修正; 例如, 某参数的后验概率小于预设门限时, 则标识该参数为不合理 参数, 也可以称为坏参数。 针对不合理参数, 需要做错误隐藏修正, 具体 的错误隐藏修正算法有很多种, 本发明对具体的参数级错误隐藏算法不进 行限制。 例如, 可以采用基于 BFI对错帧进行整帧丢弃的错误隐藏机制。 经过译码有效性判断模块处理后的语音编码帧,可以被看做是 "合理的帧", 此时可将 BFI设置指示该帧为 "合理的帧";
步驟 405 ,将修正结果发送给语音译码器,语音译码器恢复出语音波形。 将修正后的语音参数进行编码获得语音编码比特; 将所述语音编码比 特与错帧指示 BFI信号发送给应用层语音译码器, 语音译码器根据收到, 语音数据、以及对应的 BFI指示,即可恢复出语音信号。这里,可以为 AMR 语音。
需要说明的是,在 UMTS的上行链路中,由于接收机在基站 NodeB里, 通过 NodeB和无线网络控制器(Radio Network Controller, RNC )之间的 Iub 接口, NodeB 接收机单独将将第一译码结果中检错码校验结果传给 RNC。 RNC利用检错码校验结果进行外环功控。 在 UMTS的下行链路中, 即接收机在用户终端时, 则不需要功率控制。
采用本发明实施例提供的增强译码方法,相对单独采用 VA译码器、单 独利用 PLVA译码以及 VA联合语音帧修复译码器, 在 MOS性能上均有增 益。
本发明实施例提供的一种译码方法, 通过采用 LVA与语音帧修复译码 器联合对语音帧进行译码, 之后对译码结果进行有效性判断, 并将不合理 的参数进行修正, 最终将结果送入语音译码器, 完成语音信号的恢复, 可 以提高主观语音质量平均意见分, 同时降低杂音风险。
本发明实施例提供一种译码装置, 该装置可以为基站, 也可以为用户 终端, 还可以为其他中继设备, 本发明实施例不做限制。 如图 6 所示, 该 装置包括: 基于检错码辅助判决的译码器 701 , 语音帧修复译码器 702, 译 码有效性判断模块 703,虚拟信源译码器 7031,参数级错误隐藏处理器 7032, 译码软值估计模块 704, 第一译码软值估计单元 7041 , 第二译码软值估计 单元 7042, 信道译码器 705, 无线网络控制器 706;
基于检错码辅助判决的译码器 701,用于将解调后的软值中第一子流进 行译码, 获得第一译码结果;
可选的, 检错码可以为列举维特比译码算法 CRC检错码 , 但本发明实 施例中的检错码并不限于 CRC检错码, 还可以为其他的检错码。 所述基于 检错码辅助判决的译码器为列举维特比译码算法 LVA译码器, 或者比特反 转 BF译码器。
其中, 第一译码结果包括: 检错码校验结果, 最优序列对应的检错码 校验结果和译码比特序列。
进一步的, 所述基于检错码辅助判决的译码器 701 具体用于: 将解调 后的软值中第一子流进行译码 , 对译码结果中译码比特序列进行软值估计 , 获得第一译码结果; 或者,
所述基于检错码辅助判决的译码器 701 ,用于将解调后的软值中第一子 流进行译码;译码软值估计模块 704,用于对译码结果中译码比特序列进行 软值估计, 获得第一译码结果。
进一步的, 对译码结果中译码比特序列进行软值估计时, 所述基于检 错码辅助判决的译码器 701或者所述译码软值估计模块 704还包括: 第一 译码软值估计单元 7041或者第二译码软值估计单元 7042;
第一译码软值估计单元 7041, 用于确定译码结果中所述译码比特序列 对应的所述解调后的软值, 根据所述解调后的软值获取中距离目标比特最 近的若干个解调软值幅度的均值, 确定所述目标比特的软值幅度; 或者, 第二译码软值估计单元 7042, 用于确定最优译码比特序列, 并根据所 述最优译码比特序列确定对数似然比 LLR软值。 进一步的, 在基于检错码辅助判决的译码器 701 对第一子流进行译码 时,信道译码器 705,用于将解调后的软值中第二子流、第三子流进行译码, 分别获得第三译码结果以及第四译码结果, 其中第一子流、 第二子流、 第 三子流为按照语音比特重要性由高到低进行划分的子流; 所述信道译码器 705至少包括维特比译码算法 VA、 最大对数最大后验 MLP译码算法。
语音帧修复译码器 702, 用于将所述第一译码结果进行修正, 获得第二 译码结果; 具体的, 将第一译码结果中未通过检错码校验的帧进行修正。
译码有效性判断模块 703 , 用于将所述第二译码结果中的参数进行修 正, 并将修正结果发送给语音译码器。
进一步的, 所述译码有效性判断模块 703 具体用于: 将所述第二译码 结果中不合理的参数进行修正; 或者, 将所述第三译码结果与所述第四译 码结果中的至少一个译码结果以及所述第二译码结果中不合理的参数进行 修正。
进一步的, 所述译码有效性判断模块 703包括: 虚拟信源译码器 7031 , 参数级错误隐藏处理器 7032;
虚拟信源译码器 7031 , 用于将译码结果在物理层恢复出语音参数, 所 述译码结果包括: 所述第二译码结果, 或者所述第三译码结果与所述第四 译码结果中的至少一个译码结果以及所述第二译码结果;
参数级错误隐藏处理器 7032, 用于确定所述语音参数中不合理参数, 将所述不合理参数进行修正; 将修正后的语音参数进行编码获得语音编码 比特; 将所述语音编码比特与错帧指示 BFI信号发送给应用层语音译码器。
需要说明的是, 当本实施例中的装置为基站时, 还包括无线网络控制 器 706, 用于接收所述 LVA译码器发送的所述第一译码结果中检错码校验 结果, 根据所述检错码校验结果进行外环功控。
需要说明的是, 本实施例提供的装置中可以包括另一语音帧修复译码 器,用于执行本实施例中语音帧修复译码器 702和译码有效性判断模块 703 的功能, 即将所述第一译码结果进行修正, 获得第二译码结果, 将所述第 二译码结果中的参数进行修正, 并将修正结果釆用语音译码器恢复出语音 波形。
本发明实施例提供一种译码装置, 通过基于检错码辅助判决的译码器, 用于将解调后的软值中第一子流进行译码, 获得第一译码结果; 语音帧修 复译码器, 用于将所述第一译码结果进行修正, 获得第二译码结果; 译码 有效性判断模块, 用于将所述第二译码结果中的参数进行修正, 并将修正 结果采用语音译码器恢复出语音波形, 从而可以提高主观语音质量平均意 见分, 同时降低杂音风险。
需要说明的是, 本发明实施例提供的方案, 适合于所有利用信道编码 进行语音传输并提供差错保护的系统, 其中, 信道编码可以采用卷积码、 涡轮码 turbo Code等,另外本发明实施例包括并不限于宽带 AMR语音编码 的译码、 增强型全速率 (Enhanced Full Rate, EFR )语音编码的译码等。
通过以上的实施方式的描述, 所属领域的技术人员可以清楚地了解到 本发明可以用硬件实现, 或固件实现, 或它们的组合方式来实现。 当使用 软件实现时, 可以将上述功能存储在计算机可读介质中或作为计算机可读 介质上的一个或多个指令或代码进行传输。 计算机可读介质包括计算机存 储介质和通信介质, 其中通信介质包括便于从一个地方向另一个地方传送 计算机程序的任何介质。 存储介质可以是计算机能够存取的任何可用介质。 以此为例但不限于: 计算机可读介质可以包括 RAM. ROM, EEPROM、 CD-ROM或其他光盘存储、 磁盘存储介质或者其他磁存储设备、 或者能够 用于携带或存储具有指令或数据结构形式的期望的程序代码并能够由计算 机存取的任何其他介质。 此外。 任何连接可以适当的成为计算机可读介质。 例如, 如果软件是使用同轴电缆、 光纤光缆、 双绞线、 数字用户线 (DSL ) 或者诸如红外线、 无线电和微波之类的无线技术从网站、 服务器或者其他 远程源传输的, 那么同轴电缆、 光纤光缆、 双绞线、 DSL或者诸如红外线、 无线和微波之类的无线技术包括在所属介质的定影中。 如本发明所使用的, 盘(Disk ) 和碟(disc ) 包括压缩光碟 ( CD ), 激光碟、 光碟、 数字通用光 碟(DVD )、 软盘和蓝光光碟, 其中盘通常磁性的复制数据, 而碟则用激光 来光学的复制数据。 上面的组合也应当包括在计算机可读介质的保护范围 之内。
总之, 以上所述仅为本发明技术方案的较佳实施例而已, 并非用于限 定本发明的保护范围。 凡在本发明的精神和原则之内, 所作的任何修改、 等同替换、 改进等, 均应包含在本发明的保护范围之内。

Claims

权利要求
1、 一种译码方法, 其特征在于, 包括:
将解调后的软值中第一子流釆用基于检错码辅助判决的译码器进行译 码, 获得第一译码结果;
将所述第一译码结果采用语音帧修复译码器进行修正, 获得第二译码 结果;
将所述第二译码结果中的参数进行修正以及有效性判断, 并将修正结 果发送给语音译码器。
2、根据权利要求 1所述的方法,其特征在于, 所述第一译码结果包括: 检错码校验结果, 最优序列对应的检错码校验结果和译码比特序列。
3、 根据权利要求 2所述的方法, 其特征在于, 所述将解调后的软值中 第一子流釆用基于检错码辅助判决的译码器进行译码, 获得第一译码结果 包括:
将解调后的软值中第一子流采用基于检错码辅助判决的译码器进行译 码, 对译码结果中译码比特序列进行软值估计, 获得第一译码结果。
4、 根据权利要求 3所述的方法, 其特征在于, 所述对译码结果中译码 比特序列进行软值估计包括:
确定译码结果中所述译码比特序列对应的所迷解调后的软值, 根据所 述解调后的软值获取中距离目标比特最近的若干个解调软值幅度的均值, 确定所述目标比特的软值幅度; 或者,
确定最优译码比特序列, 并根据所述最优译码比特序列确定对数似然 比 LLR软值。
5、 根据权利要求 1所述的方法, 其特征在于, 所述将解调后的软值中 第一子流采用基于检错码辅助判决的译码器进行译码的同时, 还包括: 将解调后的软值中第二子流、 第三子流采用信道译码器进行译码, 分 别获得第三译码结果以及第四译码结果;
其中第一子流、 第二子流、 第三子流为按照语音比特重要性由高到低 进行划分的子流; 所述信道译码器至少包括维特比译码算法 VA、 最大对数 最大后验 MLP译码算法。
6、 根据权利要求 1或 5所述的方法, 其特征在于, 所述将所述第二译 码结果中的参数进行修正以及有效性判断包括:
将所述第二译码结果中不合理的参数进行修正; 或者,
将所述第三译码结果与所述第四译码结果中的至少一个译码结果以及 所述第二译码结果中不合理的参数进行修正以及有效性判断。
7、 根据权利要求 6所述的方法, 其特征在于, 所述将不合理的参数进 行修正以及有效性判断包括:
釆用虚拟信源译码器, 将译码结果在物理层恢复出语音参数, 所述译 码结果包括: 所述第二译码结果, 或者所述第三译码结果与所述第四译码 结果中的至少一个译码结果以及所述第二译码结果;
采用参数级错误隐藏处理器确定所述语音参数中不合理参数, 将所述 不合理参数进行修正; 将修正后的语音参数进行编码获得语音编码比特; 将所述语音编码比特与错帧指示 BFI信号发送给应用层语音译码器。
8、 根据权利要求 1或 2所述的方法, 其特征在于, 在所述将解调后的 软值中第一子流釆用基于检错码辅助判决的译码器进行译码, 获得第一译 码结果之后, 还包括:
将所述第一译码结果中检错码校验结果发送给无线网络控制器 RNC , 以便所述 RNC根据所述检错码校验结果进行外环功控。
9、 根据权利要求 1或 3所述的方法, 其特征在于,
所述基于检错码辅助判决的译码器至少包括列举维特比译码算法 LVA 译码器, 或者比特反转 BF译码器。
10、 一种译码装置, 其特征在于, 包括: 基于检错码辅助判决的译码器, 用于将解调后的软值中第一子流进行 译码, 获得第一译码结果;
语音帧修复译码器, 用于将所述第一译码结果进行修正, 获得第二译 码结果;
译码有效性判断模块, 用于将所述第二译码结果中的参数进行修正, 并将修正结果发送给语音译码器。
11、 根据权利要求 10所述的装置, 其特征在于, 所述第一译码结果包 括: 检错码校验结果, 最优序列对应的检错码校验结果和译码比特序列。
12、 根据权利要求 10所述的装置, 其特征在于,
所述基于检错码辅助判决的译码器具体用于: 将解调后的软值中第一 子流进行译码, 对译码结果中译码比特序列进行软值估计, 获得第一译码 结果; 或者,
所述基于检错码辅助判决的译码器, 用于将解调后的软值中第一子流 进行译码; 所述装置还包括: 译码软值估计模块, 用于对译码结果中译码 比特序列进行软值估计, 获得第一译码结果。
13、 根据权利要求 12所述的装置, 其特征在于, 所述基于检错码辅助 判决的译码器或者所述译码软值估计模块包括:
第一译码软值估计单元 , 用于确定译码结果中所述译码比特序列对应 的所述解调后的软值, 根据所述解调后的软值获取中距离目标比特最近的 若干个解调软值幅度的均值, 确定所述目标比特的软值幅度; 或者,
第二译码软值估计单元, 用于确定最优译码比特序列, 并根据所述最 优译码比特序列确定对数似然比 LLR软值。
14、 根据权利要求 10所述的装置, 其特征在于, 所述装置还包括: 信道译码器, 用于将解调后的软值中第二子流、 第三子流进行译码, 分别获得第三译码结果以及第四译码结果;
其中第一子流、 第二子流、 第三子流为按照语音比特重要性由高到低 进行划分的子流; 所述信道译码器至少包括维特比译码算法 VA、 最大对数 最大后验 MLP译码算法。
15、 根据权利要求 10或 14所述的装置, 其特征在于, 所述译码有效 性判断模块具体用于:
将所述第二译码结果中不合理的参数进行修正; 或者,
将所述第三译码结果与所述第四译码结果中的至少一个译码结果以及 所述第二译码结果中不合理的参数进行修正。
16、 根据权利要求 15所述的装置, 其特征在于, 所述译码有效性判断 模块包括:
虚拟信源译码器, 用于将译码结果在物理层恢复出语音参数, 所述译 码结果包括: 所述第二译码结果, 或者所述第三译码结果与所述第四译码 结果中的至少一个译码结果以及所述第二译码结果;
参数级错误隐藏处理器, 用于确定所述语音参数中不合理参数, 将所 述不合理参数进行修正; 将修正后的语音参数进行编码获得语音编码比特; 将所述语音编码比特与错帧指示 BFI信号发送给应用层语音译码器。
17、 根据权利要求 10-16中任一项所述的装置, 其特征在于, 所述装置为基站, 或者为用户终端, 或者为中继设备。
18、根据权利要求 17所述的装置, 其特征在于, 当所述装置为基站时, 所述装置还包括:
无线网络控制器, 用于接收所述基于检错码辅助判决的译码器发送的 所述第一译码结果中检错码校验结果, 根据所述检错码校验结果进行外环 功控。
19、 根据权利要求 10或 12所述的方法, 其特征在于,
所述基于检错码辅助判决的译码器至少包括列举维特比译码算法 LVA 译码器, 或者比特反转 BF译码器。
PCT/CN2013/086670 2012-11-15 2013-11-07 一种译码方法及装置 WO2014075579A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201210460305.1 2012-11-15
CN201210460305.1A CN103812511B (zh) 2012-11-15 2012-11-15 一种译码方法及装置

Publications (1)

Publication Number Publication Date
WO2014075579A1 true WO2014075579A1 (zh) 2014-05-22

Family

ID=50708791

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2013/086670 WO2014075579A1 (zh) 2012-11-15 2013-11-07 一种译码方法及装置

Country Status (2)

Country Link
CN (1) CN103812511B (zh)
WO (1) WO2014075579A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111435838B (zh) * 2019-01-14 2022-06-14 华为技术有限公司 译码方法、装置及设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1722276A (zh) * 2004-04-02 2006-01-18 日立环球储存科技荷兰有限公司 用于使用多个交错删除指针来检测和校正错误的技术
CN101309086A (zh) * 2008-06-27 2008-11-19 东南大学 里德-所罗门码级联反馈系统卷积码的系统的译码方法
EP2141816A1 (en) * 2008-07-03 2010-01-06 Victor Company Of Japan, Limited Encoding method, encoding apparatus, decoding method, and decoding apparatus using block code
CN101848002A (zh) * 2010-06-18 2010-09-29 上海交通大学 Rs级联网格调制码的迭代译码装置及其译码方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1722276A (zh) * 2004-04-02 2006-01-18 日立环球储存科技荷兰有限公司 用于使用多个交错删除指针来检测和校正错误的技术
CN101309086A (zh) * 2008-06-27 2008-11-19 东南大学 里德-所罗门码级联反馈系统卷积码的系统的译码方法
EP2141816A1 (en) * 2008-07-03 2010-01-06 Victor Company Of Japan, Limited Encoding method, encoding apparatus, decoding method, and decoding apparatus using block code
CN101848002A (zh) * 2010-06-18 2010-09-29 上海交通大学 Rs级联网格调制码的迭代译码装置及其译码方法

Also Published As

Publication number Publication date
CN103812511B (zh) 2017-04-26
CN103812511A (zh) 2014-05-21

Similar Documents

Publication Publication Date Title
US8327234B2 (en) Code block reordering prior to forward error correction decoding based on predicted code block reliability
US20050003782A1 (en) Methods and apparatus for channel quality indicator determination
CN109964426B (zh) 用于解析接收信号的信号接收器和方法、通信系统
WO2002045326A1 (fr) Decodeur et procede de decodage
JPH10303759A (ja) 誤り訂正復号装置および誤り訂正復号方式
JP2007525095A (ja) 無線通信中の送信出力の制御において用いられる方法、装置、及びシステム
BRPI0619223A2 (pt) aparelho e método para decodificar uma mensagem recebida com informação a priori
WO2007087512A2 (en) Apparatus and methods for jointly decoding messages based on apriori knowledge of modified codeword transmission
EP2210360B1 (en) Apparatus and method for decoding in mobile communication system
US20170141876A1 (en) Signal processing method and device
CA2694933C (en) Code block reordering prior to forward error correction decoding based on predicted code block reliability
CN104079376A (zh) 使用前向纠错的数据通信方法和设备
EP1506634B1 (en) Blind transport format detection for transmission link
WO2014075579A1 (zh) 一种译码方法及装置
WO2010096949A1 (zh) 在利用网络编码重传的系统中的解码方法和接收设备
US7861137B2 (en) System for identifying localized burst errors
US9037942B2 (en) Modified joint source channel decoder
WO2013159364A1 (zh) 空口语音帧修复译码方法、信源边信息获取方法及设备
CN102104448B (zh) 检测语音帧的编码速率的方法和设备
JP2008054235A (ja) 変調方式判定装置、受信装置、変調方式判定方法及び変調方式判定プログラム
US7616636B2 (en) Method and device for receiving data blocks
WO2002052729A2 (en) Decoder, system and method for decoding turbo block codes
JP3979266B2 (ja) ブラインドレート検出装置、復号装置、通信装置、ブラインドレート検出方法および復号方法
JP5186324B2 (ja) 通信装置及び通信方法
JP2000269937A (ja) デジタルデータ復号器

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 13855620

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 13855620

Country of ref document: EP

Kind code of ref document: A1