45 results about "PESQ" patented technology

When the PESQ and similar algorithms are used to measure speech quality, a particular voice call is set up to transmit only test voice signals over a communications network. This enables the test voice signals to be easily identified and provides a means of determining the amount of degradation that occurs as a result of transmission of the test voice signals. However, one problem is that in packet switched networks the transmission characteristics change with time. Thus the estimated MOS score obtained cannot be assumed to give an accurate speech quality measure for a voice call later made between the same two points. By adding test voice information to an ongoing voice call and transmitting that test voice information integrally with the ongoing voice call these problems are addressed. The test voice information is sent in packets during silent periods in the voice call such that the voice call is not affected by the test voice information. An identifier is used to identify or label the packets comprising test voice information. In the case that real-time transport protocol is used, the identifier is preferably a payload type value.

A system and method for testing radio coverage of a land mobile radio system. The system configuration generally employs a fixed unit, a mobile unit and the two-way radio system (not part of the invention) subject to testing. The test system and method are compatible with any land-mobile radio system (conventional, trunked, using proprietary or standards-based protocols, FDMA or TDMA) operating in any frequency band. One of three methods may be selected to perform testing at any one time: continuous testing, grid testing or receive-only testing. In the continuous and grid modes, both the talk-out and talk-in paths are measured. Test results are recorded in a computer file after being converted to a digital format, and are compared to an original “reference file” via the Perceptual Evaluation of Speech Quality (PESQ) algorithm as defined by ITU-T standard P.862. The PESQ output is then converted to a Delivered Audio Quality (DAQ) score by application of an empirical set of weighting factors to certain of the PESQ parameters.

The invention discloses an objective evaluation method for VoIP speech, which comprises the following steps of: acquiring original speech at a transmitting end of VoIP; acquiring distortion speech introducing damage at a receiving end; differentiating the original speech and the distortion speech through a pre-processing module to acquire statement information, calculating coder / decoder damage of aligned statements by using a coder / decoder damage assessment module A, calculating the quality damage of the stretched or compressed statements by using a coder / decoder damage assessment module B, and evaluating the influence of end-to-end delay and internal delay on the speech by using a delay damage assessment module; and finally acquiring an objective evaluation score of speech quality by using a fitting module. The method evaluates the speech quality by using a network coder / decoder, an adaptive jitter buffer and the like, is more accurate compared with the method only using perceptual evaluation of speech quality (PESQ), considers the delay damage, and also can evaluate the session quality. Meanwhile, compared with the correlation degree with a subjective score acquired by an E model, the method is more suitable for objectively evaluating the speech with network damage.

A system and method for testing radio coverage of a land mobile radio system. The system configuration generally employs a fixed unit, a mobile unit and the two-way radio system (not part of the invention) subject to testing. The test system and method are compatible with any land-mobile radio system (conventional, trunked, using proprietary or standards-based protocols, FDMA or TDMA) operating in any frequency band. One of three methods may be selected to perform testing at any one time: continuous testing, grid testing or receive-only testing. In the continuous and grid modes, both the talk-out and talk-in paths are measured. Test results are recorded in a computer file after being converted to a digital format, and are compared to an original “reference file” via the Perceptual Evaluation of Speech Quality (PESQ) algorithm as defined by ITU-T standard P.862. The PESQ output is then converted to a Delivered Audio Quality (DAQ) score by application of an empirical set of weighting factors to certain of the PESQ parameters.

This invention provides a method for testing phonetic quality in which, a calling terminal in a tested network segment of IP network utilizes the IP address parameters of the calling and the called end to deliver reference phonetic documents to the called terminal, the called end utilizes the IP address parameters of the calling and called ends to receive the step down phonetic document formed at the local end transferred by the tested network section finally, the reference and step down phonetic documents are used to evaluate the phonetic quality to get the load phonetic quality of the tested network section. This invention provides the test device including: a user interface module, an outboard interface module, a delivery module a receive module, a file forward module and PESQ computing module.

The invention provides a test method for objectively evaluating the voice quality of a pickup system. The test method mainly comprises the following steps of: testing sound field frequency response through a testing environment or direct calculation according to frequency response test signals; calculating an isostatic compensation function; carrying out isostatic compensation on standard test voice signals; playing the standard test voice signals treated with isostatic compensation, collecting the played voice signals through a pickup device of the pickup system to be tested so as to generate a degraded voice file; comparing the standard test voice signals with degraded voice signals to calculate PESQ value; calculating MOS-LQO value by utilizing a projected curve; and carrying out sample averaging to obtain the test result. The test method overcomes the limitation that the existing test method for objectively evaluating the voice quality can only carry out end-to-end measurement, and the test method can be used for testing the voice quality of the on-site monitoring pickup system.

The invention relates to a speech quality evaluation system of mobile communication network. The evaluation system mainly comprises a tested speech network and a data transmission network, and is characterized by consisting of dialing test terminals, a dialing test control server, a central dialing test terminal, a database server, an analysis management server, a Web server and a GIS server, wherein each server exchanges data through a relevant interface table in the database server. The evaluation system analyzes audio signals sent among the dialing test terminals through embedded program so as to identify the problems such as single pass, intermittence, silence, strange noise, and the like; moreover, when required by the central dialing test terminal, receiving audio frequency is transmitted back to obtain objective evaluation index PESQ value of speech quality during communication through professional software; the dialing test terminals obtain communication link parameters at the same time and report the parameters along with analysis conclusion; and the central dialing test terminal carries out relevant analysis estimation of speech quality grade and network parameters through comparison.

The present invention discloses a mobile communication speech quality remote inspection system. The present invention comprises an inspection central platform (200) and a remote inspector (100) with an antenna (4), and the remote inspector (100) is connected with the inspection central platform (200) through an inspected system; a voice quality inspecting component (700) is arranged in the inspection central platform (200), and a voice quality inspecting component of the voice MOS assessment software based on the PESQ algorithm is arranged in the voice quality inspecting component (700). The present invention also provides a method for performing the mobile communication speech quality remote inspection by utilizing the above inspection system, the method can perform inspection and assessment to the speech quality objectively and effectively, therefore to judge the running state of the signal source equipment or the communication network, and to sensing the feeling of the user during the mobile communication.

The invention discloses a deep-neural-network-based objective evaluation method for speech quality. The method comprises: step one, constructing and training a deep neural network by using data generated based on three kinds of speech features of a noisy speech as the input of the deep neural network and a PESQ score of an actual target speech as an output target of the deep neural network; and step two, inputting a q-frame speech feature vector, as input data, of a to-be-evaluated speech into the trained deep neural network and outputting a quality evaluation score of the to-be-evaluated speech. With the deep-neural-network-based objective evaluation method disclosed by the invention, the target speech quality can be evaluated objectively under the condition of only having a target speechsignal; and the objective evaluation of the target speech is associated with evaluation made by the actual PESQ algorithm highly. The deep-neural-network-based objective evaluation method is suitablefor the variable-speed speech; a phenomenon that the speech quality can not be evaluated objectively because of speech speed changing is avoided; and the evaluation result is accurate. The speech quality can be evaluated directly without a pure reference signal.

The invention discloses an objective assessment method based on voice enhancement algorithm subjective assessment. The method is a new objective combination method proposed by utilizing multi-linear regression analysis based on five voice enhancement objective assessment methods as follows: PESQ, LLR, CEP, fwSNRseg and fwSNRsegVar. An objective score for assessing the quality of a voice signal, an objective score for assessing the quality of background noise and an objective score for assessing overall quality are obtained by the combinations of different weighting coefficients and five objective methods. The subjective and objective correlation coefficients of the method are superior to a plurality of other traditional objective algorithms, the standard deviation of error is lower, and the assessment of performance of the voice enhancement algorithm is easier, therefore, the invention has great practical value.

The invention discloses an end-to-end voice enhancing method based on RefineNet. Firstly, a time-frequency analysis network is constructed to carry out coding analysis on voice signals, then the RefineNet is used for learning the characteristic mapping from noisy voice to pure voice, and finally, enhanced voice signals are generated through decoding. On the basis, an improved method for fusing evaluation indexes with a training loss function and a multi-objective fusion learning strategy for simultaneously taking an STOI and an SDR as optimization targets are provided. In tests under different noise environments and different signal-to-noise ratios, the indexes of the method provided by the invention are obviously superior to those of representative traditional methods and non-end-to-end and end-to-end deep learning methods in the aspects of the STOI, the PESQ and the SDR, the definition and the intelligibility of voice can be better improved, and a better voice enhancement effect is achieved.

The invention provides a method for detecting whether frames drop and counting the number of the dropped frames according to number continuity by using the characteristic that an adaptation layer (AL2) of an H.324M protocol stack can carry frame numbers, and gives a method for processing video interpolated frames, audio interpolated frames and wrong frames. Alignment of a sending frame sequence and a receiving frame sequence is replaced with non-accurate alignment of a PEVQ (Perceptual Evaluation Of Video Quality) algorithm and a PESQ (Perceptual Evaluation Of Speech Quality) algorithm, the evaluation is carried out by using aligned receiving and sending speech and video frame sequences, therefore, processing speed and accuracy of the original algorithm can be greatly improved. By using the sending rate of the H.324M protocol, relevant parameters are accurately calculated and an accurate non-aligned AVI (Audio Video Interleaved) file is synthesized by sending audio and video data withfixed size by using the characteristics that the sending time can be dynamically regulated and the sending rate of the H.324M protocol can be set in advance, and audio and video synchronization evaluation and video telephone integral quality evaluation are accurately carried out by using an AVET (Advanced Video Quality Evaluation) algorithm.

The invention provides a voice sample collection method based on a network dubbing game; the method comprises the following steps: 101, a voice sample collecting node: playing a video, dubbing with video captions, allowing the user to change the original self voice on a client end, and enabling a backstage server to machine-score the original audio frequency according to PESQ objective voice quality evaluation and speech recognition; 102, voice sampling subjective evaluation node: extracting the voice sample and displaying same on a game home page, and allowing the user to audition and evaluate the displaying voice sample and select a conformity or an unconformity button; allowing the backstage to artificially audition the voice sample with more unconformity buttons selected by the user, and removing the voice sample from the sample database if the selected sample cannot indeed comply with the original text after artificial audition. An artificial field recording voice sample scheme is high in recording cost and long time duration, and an artificial telephone recording scheme voice sample contains channel noises; the voice sample collection method can solve said problems.

The invention discloses a quasi-clean voice construction method for voice quality objective evaluation. An improved minimum value control recursion average algorithm and a multi-spectrum subtraction are adopted to obtain quasi-clean voice of distorsion voice. The method mainly comprises the steps of (1) distinguishing a voice segment and a non-voice segment of the distorsion voice; (2) respectively evaluating noise power spectrums of the voice segment and the non-voice segment according to the division of the voice segment and the non-voice segment; (3) calculating the quasi-clean voice power spectrum of the distorsion voice according to noise spectrum evaluation values of the non-voice segment and the voice segment. The quasi-clean voice construction method for voice quality objective evaluation has the advantages that the quasi-clean voice and the distorsion voice serve as input voice of a PESQ algorithm, and an objective evaluation value of the distorsion voice is obtained.

A speech enhancement method based on a hybrid masking learning target comprises the steps of performing traditional feature extraction of speech signals, dividing the acquired speech signals into a training set and a test set, and respectively extracting traditional features of the speech signals of the training set and the test set, respectively extracting amplitude spectrum features of STFT domains of the speech signals of the training set and the test set, constructing a deep stacking residual network, constructing a learning target, training the deep stacking residual network by using theextracted traditional features of the training set, the extracted amplitude spectrum features of the STFT domain and the learning target, and inputting the extracted traditional features of the test set and the amplitude spectrum features of the STFT domain into the trained deep stacking residual network to obtain a predicted learning target, performing ISTFT on the predicted learning target to obtain an enhanced speech signal, and calculating a PESQ value of the speech signal. Noise information is not reserved in the speech-dominated time frequency unit, so that the calculation amount is reduced, and neural network learning is easy to train to improve the intelligibility and quality of speech.

The invention belongs to the hearing aid voice quality evaluation technology field and especially relates to a hearing aid voice quality W-PESQ objective evaluation method based on gain compensation. The method comprises steps that firstly, an end point detection method is utilized to carry out alignment processing of a recorded hearing aid output voice signal and an original pure voice signal; signal interception processing on the recorded hearing aid output voice signal and the original pure voice signal after alignment processing is carried out; gain compensation of the pure voice signal without flowing through a hearing aid is carried out; a gain compensation filter is designed, gain adjustment on the original pure voice signal is carried out according to a frequency domain gain curve of the present hearing aid so as to realize fitting of a frequency domain response curve of the gain compensation filter and a gain curve of the hearing aid; lastly, objective evaluation on hearing aid voice quality is carried out through employing a W-PESQ algorithm. The method is advantaged in that a hearing disorder population is considered, the gain compensation filter is designed, through subjective verification, the method has excellent subjective and objective agreement, and great practical values are realized.

An invention is provided for testing a quality of communication data received from a SUT. The invention includes storing reference test data comprising a plurality of data segments, and receiving degraded test data also comprising a plurality of data segments, from the SUT. The data segments are located in the degraded test data, and data segments in the degraded test data are matched to related data segments in the reference test data. Further, the data segments in the degraded test data are compared to corresponding data segments in the reference test data using a fixed point operation. In one aspect, the degraded test data can be normalized prior to locating the data segments, for example, utilizing a fixed point Fourier transform. Also, a receive filter can be applied to the normalized test data utilizing a fixed point operation. Further, a perceptual evaluation signal quality (PESQ) result can be generated based on the comparison of the data segments in the degraded test data to corresponding data segments in the reference test data. The PESQ result optionally can be stored to a quality of service (QoS) data file.

The invention discloses a color-ring back tone (CRBT) quality detection method, belonging to CRBT quality detection technical field. The CRBT tone quality detection method comprises the steps that a CRBT to be tested is dialled through a subscriber terminal and recorded as an object ring tone file; the object ring tone file is compared with a source ring tone file on a CRBT platform by a PESQ (Perceptual Evaluation of Speech Quality) algorithm, a MOS (Man Option Score) value is output and a CRBT effect class is obtained. The CRBT quality detection method is objective and effective and reaches more than 95% of correlation between the judgement of ring tone class and the subjective judgement.

Disclosed are a method and a system for training a neural network. According to one embodiment, the method includes receiving a noisy signal, generating a denoised output signal, determining a signaldistortion ratio (SDR) loss function based on the denoised output signal, determining a perceptual evaluation of speech quality (PESQ) loss function based on the denoised output signal, and optimizingan overall loss function based on the PESQ loss function and the SDR loss function.

The invention relates to a multichannel speech enhancement method based on a time-frequency domain binary mask for receiving speech signals by an array. A binary mask is calculated by utilizing a network model to output speech existence probability estimation, and classification of a signal time-frequency domain and corresponding beamforming parameter estimation are realized through the binary mask, so that a better speech enhancement effect is obtained. The method comprises the following steps: firstly, performing time-frequency domain speech existence probability estimation on an array receiving signal by utilizing a network model, and then calculating a threshold value by utilizing an estimation result and the receiving signal, thereby calculating binary mask estimation and beam formingrelated parameter estimation so as to realize multi-channel speech enhancement. Compared with an existing array received signal speech enhancement algorithm, the method has a higher output signal-to-noise ratio and a higher subjective speech quality assessment PESQ score.

The invention provides a noise estimation method based on recursive least tracking for sound pickup devices. A recursive average technology is adopted for a smoothing factor in the invention, and during noise cross-power spectrum estimation, the smoothing factor is correspondingly adjusted according to the actual signal-to-noise ratio of each frequency point. Simulation results show that a low noise mean square estimation error and a better perceptual evaluation of speech quality (PESQ) score are obtained when the noise estimation method is applied to a speech enhancement system.

The invention provides a test method for objectively evaluating the voice quality of a pickup system. The test method mainly comprises the following steps of: testing sound field frequency response through a testing environment or direct calculation according to frequency response test signals; calculating an isostatic compensation function; carrying out isostatic compensation on standard test voice signals; playing the standard test voice signals treated with isostatic compensation, collecting the played voice signals through a pickup device of the pickup system to be tested so as to generate a degraded voice file; comparing the standard test voice signals with degraded voice signals to calculate PESQ value; calculating MOS-LQO value by utilizing a projected curve; and carrying out sample averaging to obtain the test result. The test method overcomes the limitation that the existing test method for objectively evaluating the voice quality can only carry out end-to-end measurement, and the test method can be used for testing the voice quality of the on-site monitoring pickup system.

The invention provides a method for testing the speech quality of an audio conference. The method comprises the steps that source audio corresponding to each member participating in the audio conference is acquired; each member constructs synthetic audio which is similar to target audio based on the source audio through an audio system, that is, each member mixes the source audio of the other members after deducting the source audio thereof so as to form synthetic audio, and the synthetic audio is the synthetic audio corresponding to each member; and each member carries out PESQ (Perceptual Evaluation of Speech Quality) calculation on the synthetic audio corresponding to each member and target audio data sent by a server of the audio conference by using an ITU-T P.862 proposal, and records a numerical value of the speech quality so as to complete testing for the speech quality. The method is simple and accurate in testing.

A shipborne communication monitoring system disclosed by the present invention comprises a voice signal transmitting terminal, the voice signal transmitting terminal is connected with a marine communication base station through a wireless signal, the voice signal transmitting terminal is also electrically connected with a recording module, the recording module is electrically connected with a recording file transmission module, and the recording module is used for recording voice information of the voice signal transmitting terminal. The recording file transmission module is used for sending recorded voice information to the recording file receiving end through the marine communication base station, the server and the wired network, and the recording file receiving end and the voice signal receiving end are both in telecommunication connection with the voice tester. According to the invention, the voice of the voice signal transmitting end can be recorded and sent to the recording file receiving end, the voice quality can be detected more accurately by comparing real-time voice with voice recording, detection can be carried out through a PESQ system and manual reinspection, the detection modes are diversified, and the detection result is more accurate; and a delay calculation module for calculating communication delay is also arranged, so that the delay of a communication signal can be monitored.

The application provides a voice quality detection system and method, and a PESQ control terminal; the voice quality detection system comprises the voice quality perception evaluation PESQ control terminal, a sound card, a first connecting line, a second connecting line, a first terminal and a second terminal; the PESQ control terminal is connected with the sound card; the sound card is connectedwith the first terminal through the first connecting line; the sound card is connected with the second terminal through the second connecting line; a normal sound card can serve as an audio frequencyemission and reception device in the PESQ system, thus reducing the voice quality detection costs.

The invention discloses an objective evaluation method for VoIP speech, which comprises the following steps of: acquiring original speech at a transmitting end of VoIP; acquiring distortion speech introducing damage at a receiving end; differentiating the original speech and the distortion speech through a pre-processing module to acquire statement information, calculating coder / decoder damage ofaligned statements by using a coder / decoder damage assessment module A, calculating the quality damage of the stretched or compressed statements by using a coder / decoder damage assessment module B, and evaluating the influence of end-to-end delay and internal delay on the speech by using a delay damage assessment module; and finally acquiring an objective evaluation score of speech quality by using a fitting module. The method evaluates the speech quality by using a network coder / decoder, an adaptive jitter buffer and the like, is more accurate compared with the method only using perceptual evaluation of speech quality (PESQ), considers the delay damage, and also can evaluate the session quality. Meanwhile, compared with the correlation degree with a subjective score acquired by an E model, the method is more suitable for objectively evaluating the speech with network damage.

The invention discloses a speech enhancement quality evaluation method and device, a terminal and a storage medium, and the method comprises the steps: using a clean speech signal as an original signal, superposing different types of noises before speech enhancement to generate a noise speech signal, and generating a speech enhancement signal through speech enhancement algorithm processing; and finally, importing the clean original speech signal and the speech enhancement signal into the PESQ to obtain a speech enhancement quality evaluation score. The method can randomly simulate noise speechsignals of different scenes, and realizes more flexible and convenient speech enhancement quality evaluation in large-scale scene testing.

The present application provides a voice quality detection system, method, and PESQ control terminal, wherein the voice quality detection system includes a voice quality sensory evaluation PESQ control terminal, a sound card, a first connection line, a second connection line, a first terminal and a second connection line. terminal, the PESQ control terminal is connected to the sound card, the sound card is connected to the first terminal through the first connection line, the sound card is connected to the first terminal through the second connection line, and the ordinary sound card is used as the audio sending and receiving device in the PESQ system to realize the reduction of voice The purpose of quality inspection cost.

The invention discloses a deep-neural-network-based objective evaluation method for speech quality. The method comprises: step one, constructing and training a deep neural network by using data generated based on three kinds of speech features of a noisy speech as the input of the deep neural network and a PESQ score of an actual target speech as an output target of the deep neural network; and step two, inputting a q-frame speech feature vector, as input data, of a to-be-evaluated speech into the trained deep neural network and outputting a quality evaluation score of the to-be-evaluated speech. With the deep-neural-network-based objective evaluation method disclosed by the invention, the target speech quality can be evaluated objectively under the condition of only having a target speechsignal; and the objective evaluation of the target speech is associated with evaluation made by the actual PESQ algorithm highly. The deep-neural-network-based objective evaluation method is suitablefor the variable-speed speech; a phenomenon that the speech quality can not be evaluated objectively because of speech speed changing is avoided; and the evaluation result is accurate. The speech quality can be evaluated directly without a pure reference signal.