69 results about "Abnormal voice" patented technology

The invention relates to an abnormal voice monitoring system and method based on an intelligent video. The system comprises a monitoring region field device, an intelligent monitoring and analysis device connected with the monitoring region field device and a monitoring center computer respectively connected with the monitoring region field device and the intelligent monitoring and analysis device, wherein the monitoring region field device comprises a video acquisition module, a voice receiving module triggered by the intelligent monitoring and analysis device and an anti-explosion warning module triggered by the monitoring center computer; and the intelligent monitoring and analysis device comprises a central processor which is respectively connected with the video acquisition module and the voice receiving module and has an output end connected with the monitoring center computer, a data storage module connected with the central processor, a video processing module connected with the central processor and a sound intensity detection module connected with the central processor. The abnormal voice monitoring system and the abnormal voice monitoring method based on the intelligent video have the advantages of relatively high intelligentization, relatively high instantaneity and relatively low false alarm rate.

The embodiment of the invention discloses a telephone recording data processing method. The method includes the steps of carrying out endpoint detection on input source recording data to obtain voiced data, judging whether the voiced data are invalid, abandoning the voiced data if the voiced data are invalid, otherwise, determining that the voiced data are voice data, judging whether the voice data are abnormal, if the answer is yes, abandoning the voice data, and otherwise, determining the voice data are valid voice data. According to the telephone recording data processing method, noise signals and abnormal voice signals with high energy are filtered out, effectiveness of data of a follow-up voice recognition system is guaranteed, the error rate of the follow-up voice recognition process is reduced, voice recognition efficiency is improved, and meanwhile, calculation is reduced due to the steps that the voice data are determined through endpoint detection. The embodiment of the invention further discloses a telephone recording data processing system.

The invention relates an abnormal voice detecting method based on time-domain and frequency-domain analysis. The method includes computing the background sound intensity of a monitored scene updated in real time at first, and detecting and extracting suddenly changed fragments of the sound intensity; then extracting uniform filter Mel frequency cepstrum coefficients of the suddenly changed fragments; and finally using the extracted Mel frequency cepstrum coefficients of sound of the abnormal fragments as observation sequences, inputting a trained modified hidden Markov process model, and analyzing whether the abnormal fragments are abnormal voice or not according to frequency characteristics of voice. Time sequence correlation is improved when the hidden Markov process model is added. The method is combined with time-domain extraction of suddenly changed energy frames and verification within a frequency-domain range, the abnormal voice can be effectively detected, instantaneity is good, noise resistance is high, and robustness is fine.

The invention discloses a method and system for detecting abnormal sounds. The method includes the steps that short-time energy of each frame of a collected audio signal is compared with a first short-time energy threshold value, if the short-time energy is larger than the first short-time energy threshold value, the corresponding frames are recorded to be first-grade frames, if the short-time energy is smaller than the first short-time energy threshold value, the short-time energy of each frame of the collected audio signal is compared with a second threshold value or the zero-crossing rate of each frame of the collected audio signal is compared with a zero-passing rate threshold value, the frames with the short-time energy larger than the second short-time energy threshold value or the zero-passing rates larger than the zero-passing rate threshold value are recorded to be second-grade frames, and when the number of the continuous frames which are the first-stage frames or the second-stage frames is larger than N and the current frame is the first-stage frame, it is judged that the sounds are abnormal. By means of the method, the abnormal sounds are judged by calculating the short-time energy and the zero-passing rates; as the short-time energy and the zero-passing rates belong to the time domain characteristics and are not related to frequency domain conversion and characteristic parameter calculating, the calculating complexity can be lowered. Meanwhile, as audio information collected in real time is processed, real-time processing and real-time analysis can be carried out, and abnormality can be judged in time.

The invention discloses an abnormal emotion automatic detection and extraction method and an abnormal emotion automatic detection and extraction system on the basis of the short-time analysis. The method comprises the following steps of: extracting an emotion characteristic sequence from a voice signal to be detected; calculating the likelihood of the emotion characteristic sequence and an abnormal emotion model in a preset emotion model and calculating the likelihood of the emotion characteristic sequence and a non-abnormal emotion model in the preset emotion model; according to the likelihood of the emotion characteristic sequence and the abnormal emotion model and the likelihood of the emotion characteristic sequence and the non-abnormal emotion model, calculating the likelihood ratio; and judging whether the likelihood ratio is greater than a set threshold value, determining the voice signal to be detected is abnormal emotion voice if yes, or determining the voice signal to be detected is a non-abnormal voice signal. Due to the utilization of the abnormal emotion automatic detection and extraction method and the abnormal emotion automatic detection and extraction system, the automatic high-efficiency judgment on the abnormal emotion in the voice signal can be implemented and the automatic processing efficiency of mass customer service data is improved.

A converter transformer operating state on-line audio analyzing and monitoring system belongs to the field of monitoring devices. The system comprises a sound collection unit, a sound transmission unit, a sound analysis unit, a historical sound library, and an analysis result comparison and analysis unit, which are sequentially connected. The sound collecting unit collects actual operating sounds of a converter transformer and converts the sounds into electrical signals. The sound transmission unit transmits the electrical signals converted from the collected actual operating sounds of the converter transformer to the sound analysis unit. The sound analysis unit conducts audio decomposition of the electrical signals converted from the collected actual operating sounds of the converter transformer and forms typical amplitude and frequency curves of the converter transformer on different operating conditions. The historical sound library stores the typical amplitude and frequency curves of the converter transformer on different operating conditions for comparison and comparative analysis. The analysis result comparison and analysis unit is used for analysis and comparison of newly collected audios of the converter transformer, analysis and comparison of historical sounds, and automatic alarming when the collected operating sounds of the converter transformer is abnormal.

The invention provides a voice evaluation method and device. The voice evaluation method comprises the steps that voice data to be tested are received; voice recognition is conducted on the voice data to be tested to obtain voice recognition data; the voice recognition data are analyzed to defect whether the voice recognition data are abnormal voice data or not, and an evaluation result of the voice data to be tested is generated according to a detection result. The voice evaluation method can detect and recognize abnormal voice, so that abnormal voice is deleted from the voice to be tested, the abnormal voice can not join evaluation of normal voice, influence of the abnormal voice on the evaluation result is greatly reduced, the accuracy of the evaluation result is improved, the evaluation requirement of a user is met, and user experience is improved.

The embodiment of the invention provides an abnormal voice classifying method and device. The abnormal voice classifying method comprises the steps that framing processing is conducted on an abnormal voice signal to obtain N frames of signals, wherein N is an integer greater than 1; filtering processing is conducted on each signal in the N frames of signals according to a filter model to obtain energy feature information of the abnormal voice signal, wherein the filter model comprises M sub-filters, frequency intervals formed by the starting frequencies and stopping frequencies of the M sub-filters are continuous and are not overlapped mutually, different values of output energies of statistic Mel-spectra of multiple abnormal voice signal samples different in type on the M sub-filters are smaller energy threshold values, and M is an integer greater than 1. According to the energy feature information of the abnormal voice signal, an abnormal voice signal classifying result is determined. By adopting the abnormal voice classifying method and the abnormal voice classifying device, the abnormal voice classifying accuracy can be improved.

The embodiment of the invention discloses a warning method of intelligent wearing equipment and the intelligent wearing equipment. The warning method of the intelligent wearing equipment comprises thefollowing steps: S1, an awakening voice acquired by the intelligent wearable equipment in real time is obtained; S2, the awakening voice is matched with a trained abnormal voice model; whether the matching value thereof is larger than a preset matching threshold value or not is judged; if the matching value is larger than the preset matching threshold value, the recording function is started to record a warning audio of a preset duration; the warning audio is sent to a preset audio receiving end. Compared with a traditional communication method for the intelligent wearable equipment, the real-time collected awakening voice is matched with the trained abnormal voice model. When it is found that abnormal voice exists in the environment where the intelligent wearable equipment is located, the warning audio of the preset duration is recorded and automatically sent to the preset audio receiving end. The whole process is safe and hidden. The intelligent wearing equipment is not required tobe operated by a user by itself. Therefore, the technical problem that the safety of the existing intelligent wearing equipment is low and emergency situations cannot be handled can be solved.

The invention provides a data processing method of the abnormal audio which is used to process the abnormal audio data at the input terminal of the wireless digital. The method comprises that builds the buffer area, judges if the received audio data is abnormal after receiving the audio data, if the data is not abnormal, puts successively the received audio data to the buffer area according to the predetermined buffer form, otherwise, carries out the replace processing of the audio data according to the buffer form and saves the processed audio data into the buffer area. The invention also provides a processing device of the abnormal audio data which is used to process the abnormal audio data at the input terminal of the wireless digital. The device comprises a buffer area, a judge and control unit. Therefore, the method and the device of the invention has great versatility, can also restore the abnormal audio data based on the product of providing the wireless access and also has traceability and large applicable scope, and simple and feasible application.

The invention discloses a voiceprint recognition method based on dynamic time warping and voice activity detection. The voiceprint recognition method comprises the following steps: firstly, recordingvoice via a voice activity detection method, and implementing cutting processing on the voice so as to filter off blank voice and abnormal voice; then, calculating a Mel-frequency cepstral coefficientof an obtained sample, and extracting characteristics of target voice; then, recording template voice, and calculating a difference degree via a dynamic time warping algorithm; and finally, recordingto-be-recognized voice via a voice recording method, extracting voice characteristics via a target voice characteristic extraction method, and comparing the target voice with the existing template voice one by one, so that a goal of voiceprint recognition is achieved. The voiceprint recognition method provided by the invention, in comparison with an existing voiceprint recognition method, can avoid the participation of excessive voice samples in the establishment of a voiceprint recognition model, and meanwhile, the voiceprint recognition method is friendly to low-performance processors and portable equipment; and the voiceprint recognition method can accelerate recognition speed, reduce resource occupation and achieve voiceprint offline recognition.

The invention discloses a substation equipment sound anomaly identification method and system. The method comprises the steps that: the sound information of target equipment of a transformer station is obtained, and the sound information is pre-processed; voiceprint features are extracted from the preprocessed sound information, and voiceprint features at least comprise FBank features, decibels, fundamental frequencies, short-time energy, short-time zero-crossing rates and correlation coefficients; the extracted voiceprint features are input into a trained voiceprint recognition model, and a recognition result is output; and the running state of the target equipment is judged based on the recognition result. According to the method and system of the invention, directional acquisition of data is carried out through a multi-element microphone array, so that acquired sound is relatively pure; sound signals acquired by a single microphone is the superposition of multi-directional sound signals; and therefore, the accuracy of sound data acquisition can be improved.

The invention relates to an intelligent decision making technology, in particular to a classification model, and provides a sound anomaly detection method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining to-be-detected audio data; performing segmentation processing on the audio data to be detected to obtain a plurality of audio sequences; performing feature extraction processing on the plurality of audio sequences to obtain a plurality of audio feature data; classifying the plurality of pieces of audio feature data, and determining category identifiers corresponding to the plurality of pieces of audio feature data; determining target audio feature data in the multiple pieces of audio feature data according to the category identifiers corresponding to the multiple pieces of audio feature data; and outputting abnormal alarm information according to the target audio feature data. The invention also relates to a block chain technology, and the obtained abnormal alarm information can be stored in a block chain.

The invention relates to a voice recognition method, device, system, and equipment and a computer readable storage medium. The method comprises the following steps: acquiring awakening voiceprint characteristics in an awakening audio when a terminal is awakened; acquiring voice recognition feedback data of a dialogue audio according to the awakening voiceprint characteristics and dialogue voiceprint characteristics in the dialogue audio after the terminal is awakened; and sending the voice recognition feedback data to the terminal to enable the terminal to display the voice recognition feedback data. With the method, audio false recognition of a non-terminal awakening user can be avoided, the false recognition rate of abnormal voices such as noise and the like of the non-terminal awakeninguser is effectively reduced, so that the voice recognition accuracy is effectively improved.

The invention discloses a voice anomaly detection method and device, a storage medium and electronic equipment. The voice anomaly detection method comprises the steps of receiving a voice data streamobtained when a voice task is executed; storing the voice data stream as a plurality of continuous voice segments according to the sequence of the receiving time; reading each voice segment accordingto the storage sequence of the voice segments, and detecting whether the read voice segments are abnormal or not; and when the number of the continuously abnormal voice segments is greater than a first preset threshold, judging that the voice task is abnormal. According to the voice anomaly detection method provided by the embodiment of the invention, the voice anomaly of the electronic equipmentcan be rapidly detected through the number of the voice segments with continuous anomalies, and the timeliness of voice anomaly detection of the electronic equipment is improved.

The invention provides a voice processing method and a device thereof, wherein the method comprises the steps of segmenting a mixed value to N voice segments through end point detection, wherein N isa natural number which is higher than or equal with 2; performing Bayes information criterion (BIC) detection on two random adjacent voice segments in the N voice segments, abandoning abnormal voice segments in BIC detection, and obtaining effective voice segments of a target object. The voice processing method and the device thereof can settle a problem of incapability of quickly and effectivelyseparating the specific target voice from the fixed voice in speaking of a specific target, thereby realizing an effect of quickly separating the specific target voice from the mixed voice.

The invention provides an ultrathin loudspeaker which comprises a flower file, a voice coil arranged in the flower file, and a lead board arranged at the back surface of the flower file, wherein the periphery of the flower file comprises at least two through holes which are internally provided with a fixed pad respectively, and leads respectively pass through the fixed pads to be connected with the voice coil and the lead board. In the ultrathin loudspeaker, the fixed pad causes the leads not to collide with the through holes (ventilation window) of the flower file or a centering support chipto generate abnormal voice, and not to generate short circuit due to collision and abrasion, thus improving the fatigue resistance strength of the leads, increasing the power acceptance capability ofthe loudspeaker, prolonging the service life of the loudspeaker, and improving the safety factor; in addition, the service life of a whole LCD TV installed with the loudspeaker is also greatly prolonged.

The embodiment of the invention discloses a voice processing method and related equipment, and the method is applied to the technical field of data processing, and comprises the steps: obtaining a voice signal of terminal equipment if the terminal equipment is in a preset environment when the terminal equipment is in a call state; and performing restoration processing on the voice signal, and outputting a restored voice signal. By adopting the voice processing mode, the call quality of the terminal equipment in a weak signal environment or a sound abnormal environment can be improved.

The invention discloses a method for monitoring quality of service. The method comprises the following steps: obtaining a service feature and a signaling feature of acurrent voice service; performingmatching search in a service feature library of an abnormal voice service based on the service feature of the current voice service; when the service feature of the current voice service matches the service feature of the abnormal voice service in the service feature library, based on the signaling feature of the current voice service, performing matching search in a signaling feature library of the abnormal voice service; when the signaling feature of the current voice service matches the signaling feature of the abnormal voice service in the signaling feature library, using a defect of quality of service corresponding to the signaling feature of the abnormal voice service matching the signaling feature of the current voice service as the defect of quality of service of the current voiceservice. Meanwhile, the invention further discloses an apparatus for monitoring quality of service

The present invention relates to a horizontal stage video camera for automatically tracking voice source, whose voice detection circuit is composed of a voice acquisition unit, a dual-channel preamp circuit, a voice signal conversion circuit, a dual-channel integral circuit, an A / D converting circuit, a triggering circuit and a PTZ controller. The virtual distance between voice signals after expansion of the strength difference is several times or furthermore larger than the physics distance when they are acquired by the voice acquisition unit, the invention takes a simple method to eliminate the disturbance of reflection voice to the main voice source, greatly facilitates the processing of subsequent data and improves the sensitivity of the complete appliance; the invention has low cost, smaller volume, suitability fro the industrial production and specific adaptation to abnormal voice source or places automatically responding to the specific voice.

A dividing module divides a voice signal into voice frames. A likelihood value generation module compares each of the voice frames with a first voice model and a second voice model to generate first likelihood values and second likelihood values. A decision module decides a windows size according to the first likelihood values and the second likelihood values. An accumulation module accumulates the first likelihood values and the second likelihood values inside the window size to generate a first sum and a second sum. A determination module determines whether the voice signal is abnormal according to the first sum and the second sum. While the voice has a big change in the environment, the decision module can dynamically adapt the windows size for decreasing the false rate of the detection and speeding up the determining of the abnormal voice.

The invention discloses a voice log service method and system in man-machine conversation. The method comprises the steps: generating an answer statement according to a received user conversation statement, and generating a voice log corresponding to the conversation in the round; Judging whether the voice log of the current round of conversation is abnormal or not according to preset abnormal logcharacteristics; And when it is judged that the voice log of the current round of conversation is abnormal, at least marking the voice log of the current round of conversation as an abnormal voice log and storing the abnormal voice log. In the embodiment of the invention, Whether a voice log generated in the man-machine conversation service providing process is an abnormal log or not is judged through preset abnormal log characteristics; and only when it is determined that the voice logs of the conversation in the current round are abnormal logs, the voice logs are marked as abnormal voice logs and stored, so that the voice logs needing to be stored are reduced, on one hand, the storage expenditure is saved, and on the other hand, a user can conveniently and rapidly inquire the abnormal logs and locate the abnormity in the man-machine conversation process.

The invention discloses an abnormal sound detection model training method and device and a computer storage medium, and the method comprises the following steps: intercepting a sound segment with a preset duration into N sub-segments, carrying out the sampling and filtering of each sub-segment by employing H band-pass filters with different frequency bands, and forming N*H*W three-dimensional feature tensors; inputting the plurality of three-dimensional feature tensors into a three-dimensional convolutional neural network for training, wherein the plurality of three-dimensional feature tensorscorrespond to a plurality of sound segments with preset durations; the plurality of sound segments with the preset duration comprise a positive sample with abnormal sound and a negative sample without abnormal sound; and calculating loss by adopting a loss function for simultaneously evaluating the positive sample and the negative sample, and updating parameters of an abnormal sound detection model. The problems of inaccurate identification and low operation efficiency in the existing sound anomaly detection are solved.

The invention relates to a voice recognition technology, discloses an air conditioning control system based on abnormal voice recognition and a control method of the air conditioning control system. The problem that in the application of the voice recognition technology in the prior art to household appliances, only a voice command is used for controlling the household appliances, and an active regulation means for the household appliances is absent is solved. The system comprises a voice recognition unit, a control unit, an execution unit, a warning unit and an abnormal voice corpus. The voice recognition unit is used for recognizing the voice in a room, and if the voice is recognized as abnormal voice, the control unit is informed. The control unit is used for controlling the warning unit to perform warning prompt according to a control strategy after receiving the notification of the voice recognition unit, and simultaneously sending a corresponding control command to the executionunit to control the execution unit to execute a corresponding action. The system is suitable for home use.

The invention discloses a voice anomaly detection method and device, and the method comprises steps: obtaining voice information to be detected and determining corresponding voice information features, and then inputting the voice information features into a pre-trained recognition model; obtaining a recognition result for the to-be-detected voice information so as to perform voice anomaly detection on the to-be-detected voice information; if it is determined that the to-be-detected voice information is abnormal voice information and the voice information re-acquired from the voice source corresponding to the to-be-detected voice information is still abnormal voice information under the condition that the standby service equipment is started, determining that a network voice attack behavior exists, and through a preset exception processing strategy, processing the voice source. Thus, anomaly detection is carried out on the intelligent voice response window session, attack behaviors of lawbreakers can be recognized, and the preset anomaly processing strategy can be adopted to process the voice source, so that the attack behaviors of the lawbreakers are prevented, and the system stability is maintained.

The invention discloses a voice anomaly detection method based on sound source characteristics. The method comprises the following steps that voice data is collected in real time through a sensor; theobtained voice segment data is preprocessed; iterative self-adaptive inverse filtering is conducted on the voice data of a voice segment to obtain glottis wave signals; characteristic parameters areextracted from the glottis wave signals, and amplitude quotient and glottal closure time ratio data are normalized; extracted characteristic data is input into an ideal SVM model for classification; classification labels are obtained and used for judging conditions of speakers, and condition labels of the speakers are output and transmitted to an execution module for feedback. The method has the advantages that as for stressful voice under mental stress, an identification method for extracting acoustic characteristic parameters lacking physical significance on the basis of a traditional linearvoice generation module is got rid of, and a sound source estimation module is established; a voice generation inverse filtering technology is used for analyzing and extracting characteristic parameters based on human vocal cord vibration to carry out abnormal voice detection.

The embodiment of the invention discloses an intelligent wearable equipment reminding method and intelligent wearable equipment. The intelligent wearable equipment reminding method comprises the following steps: S1, acquiring an awakening voice acquired by the intelligent wearable equipment in real time; and S2, matching the awakening voice with a trained abnormal voice model, judging whether a matching value is higher than a preset matching threshold value, and if yes, dialing a telephone number of a preset contact person. Compared with the traditional intelligent wearable equipment communication method, the intelligent wearable equipment reminding method disclosed by the invention can dial without a fussy unlocking step or an accurate dialing instruction, when the awakening voice acquired by the intelligent wearable equipment is matched with the trained abnormal voice model and the matching value meets the requirements, the telephone number can be automatically dialed, the whole process is rapid and covert, a user can be beneficially, rapidly and covertly dial the telephone number when in an emergency situation, safety of the intelligent wearable equipment is improved, and the technical problems that the current intelligent wearable equipment is relatively low in safety and the emergency situation can not be handled are solved.

The invention belongs to the field of sound signal processing, discloses an abnormal sound detection system and method based on time-frequency domain characteristics, and solves the problem of low abnormal sound detection and recognition rate in the prior art. The method comprises the steps of a, extracting an effective sound segment in sound data of collected equipment, wherein the effective sound segment at least comprises a stable rotation sound segment of the collected equipment; b, determining an abnormal audio frequency band range, obtaining a characteristic component containing a abnormal audio frequency band through frequency band decomposition, performing windowing framing calculation on the characteristic component to calculate time domain characteristics, and identifying a soundabnormal frame through a threshold value; and c, based on the recognition result of the sound abnormal frame, calculating frequency domain characteristics, inputting the frequency domain characteristics into a pre-trained machine learning model for classification, scoring according to a classification result, and determining an equipment abnormal sound detection result.

The invention belongs to the technical field of industrial sound anomaly detection, and particularly relates to an industrial sound anomaly detection system based on phase coding and equipment information. The method can complement the fineness of an industrial sound signal by means of phase information, obtains high-precision acoustic features, can model the high-precision acoustic features, focuses on improving the sensitivity of the system to abnormal information, senses subtle differences between different industrial devices, avoids acoustic interference generated between the devices, and improves the reliability of the system. And the modeling result is subjected to high-efficiency and high-precision anomaly score judgment, and the industrial sound anomaly detection performance is excellent.