36 results about "Voice frequency" patented technology

The invention belongs to the technical field of voice interaction, and relates to a 5G communication equipment intelligent voice interaction method and system based on deep learning, and the method comprises the steps: generating a joint input signal containing voice frequency domain information and a motion state; outputting the de-noised voice segments and the corresponding environmental interference level parameters; generating a basic voiceprint ID containing a user biological feature identifier and a dynamic key fragment; calculating a cloud processing priority score according to the environmental interference level parameter and the equipment residual electric quantity value, when the score exceeds a preset threshold value, sending a voice processing request containing a basic voiceprint ID to an edge computing node, and otherwise, triggering a local semantic recognition process; and receiving a semantic recognition result from a cloud end or a local end, and generating a multi-round dialogue response instruction in combination with the context parameters in the historical interaction data of the user. According to the invention, the problem that local processing resource overload or cloud communication delay exceeding is easily caused by a fixed noise weight distribution mechanism is solved.

According to the intelligent voice broadcast system and method for the elevator, elevator car environment data are collected in real time through a multi-source heterogeneous sensor set, and passenger density distribution and movement tracks are calculated through fusion of a millimeter wave radar and a 3D visual sensor; physiological features of a handrail contact area are detected through a biological impedance sensor so as to judge the health state of a passenger, and the current signal-to-noise ratio is dynamically calculated and the voice frequency band is optimized through an environment noise analysis module. Therefore, by constructing a technical system of'multi-modal perception-edge intelligent decision-cross-modal interaction 'and adopting innovative means such as millimeter-wave radar and biological impedance fusion sensing, deep reinforcement learning of a dynamic strategy and a phased array directional sound field, the problems of broadcast stiffness, poor scene adaptability and privacy potential safety hazards are solved. According to the system and method, accurate passenger feature recognition, dynamic voice content generation and multi-channel collaborative feedback are achieved through two-dimensional innovation, and the intelligent level and service efficiency of elevator voice interaction are improved.

The invention relates to the technical field of speech analysis, in particular to a speech enhancement method and system in a dynamic noise environment. The method comprises the following steps: acquiring a voice signal within a preset time under a fixed voice frequency band, and determining a main channel and a secondary channel; framing and windowing the voice signal to obtain a voice field, obtaining a time-frequency matrix of the voice field, determining a feature vector according to a phase difference and an amplitude ratio of each time-frequency point, and obtaining a key human voice cluster based on feature vector clustering; in the key human voice cluster, direction consistency of time-frequency points is calculated based on complex numbers of the main channel and the secondary channel, each time-frequency point is compared with other time-frequency points to obtain voiceprint feature consistency, and a human voice evaluation index is formed based on the time-frequency points and the other time-frequency points; obtaining a frequency domain binary label matrix; and training a network based on the frequency domain binary label matrix, and completing voice enhancement through the network. According to the invention, the problems of many interference persons and insufficient noise interference separation capability are solved.

The invention discloses a control system of a controllable anti-recording device, and relates to the technical field of information safety protection, a multi-dimensional audio detection module adopts a quaternary cross array MEMS microphone array and can identify characteristic signals of recording equipment in a frequency band of 20 Hz to 20 kHz, an interference generation module generates a self-adaptive comb-shaped interference spectrum based on a DDS technology, dynamically avoids communication and voice frequency bands, and provides a multi-dimensional anti-recording device based on the self-adaptive comb-shaped interference spectrum. The heterogeneous computing master control module integrates a CPU and a DSP, a multi-level decision tree algorithm is built in, the composite energy management module fuses piezoelectric and electromagnetic dual-mode energy collection, and the anti-interference communication module adopts a UWB and Bluetooth 5.3 mixed link and supports dynamic frequency hopping and LDPC coding. Through combination of the multi-dimensional microphone array and an intelligent algorithm, high-precision recording equipment detection and positioning are realized, the system integration degree is high, the system can be rapidly deployed in a conference room, a confidential place and the like, and comprehensive protection is provided for information security.

The invention provides an intelligent constitution identification system based on lingual surface infrared thermal imaging and traditional Chinese medicine syndrome differentiation and an application method. According to the system, an infrared thermal image partition acquisition module, a lingual surface image normalization acquisition module and a voice feature analysis module are fused, and an intelligent database integrating traditional Chinese medicine syndrome differentiation knowledge and cases is constructed. Through a deep learning algorithm, multi-modal features such as a human body infrared thermal structure, lingual texture color, voice frequency tone and the like are extracted, an improved cross-modal attention mechanism is applied to realize data deep fusion, a dialectical analysis model is established in combination with a traditional Chinese medicine viscera meridian theory, and dynamic tracking and health risk prediction are performed on physique change by using an LSTM network. The application method covers multi-modal data synchronous acquisition, joint feature extraction, fusion dialectical analysis, dynamic health assessment and other processes, can generate a personalized diagnosis and treatment scheme including medicine and food compatibility, meridian guidance and emotion regulation, and supports doctor-patient interaction optimization.

The invention discloses a radio analog telephone communication encryption method, device and equipment and a medium, and the method comprises the steps: generating a frequency spectrum segment number and a pseudo-random sorting sequence through a preset encryption algorithm; and carrying out controllable segmentation on the total spectrum of the analog voice according to a continuous sequence from low frequency to high frequency, carrying out sequence recombination on each sub-spectrum according to a pseudo-random sequence, finally combining and outputting encrypted voice, and recovering the original voice by utilizing the same pseudo-random sequence reverse recombination at a receiving end. Therefore, the time-frequency continuity and statistical characteristics of the voice spectrum in the traditional radio analog communication are fundamentally changed. According to the method, frequency spectrums are recombined as a whole according to a pseudo-random rule, so that an eavesdropper can only obtain disordered voice signals after the frequency spectrums are rearranged even if the eavesdropper intercepts the signals, and the difficulty of passive cracking means such as frequency spectrum analysis, filtering reduction and voice recognition is remarkably increased; the problem that in the prior art, radio analog telephone communication is insufficient in encryption and easy to crack is solved.

The application provides a voice false wake-up processing method, device and equipment and a storage medium, and relates to the technical field of smart home. The method comprises the following steps: determining the response state of each smart device in a target space receiving a voice signal; determining a pre-wake-up signal of a smart device identified as a pre-wake-up state and a non-wake-up device identified as a non-wake-up state; determining that the voice signal in the target space is a valid wake-up voice based on the pre-wake-up signal and the non-wake-up device; and waking up a target smart device in the target space based on the valid wake-up voice. The application solves the defects of low voice recognition accuracy and high false wake-up voice frequency in the prior art, reduces the dependence on wake-up audio training in a specific space environment, and improves the voice recognition accuracy by determining the valid wake-up voice.

The application discloses a controllable recording prevention device control system, and relates to the technical field of information security protection.A four-element cross array MEMS microphone array is used in the multi-dimensional audio detection module, which can identify recording equipment characteristic signals in a 20Hz to 20kHz frequency band.A self-adaptive comb interference spectrum is generated by the interference generation module based on DDS technology, and the communication and voice frequency bands are dynamically avoided.The heterogeneous computing main control module integrates CPU and DSP, and is provided with a multi-level decision tree algorithm.A composite energy management module combines piezoelectric and electromagnetic dual-mode energy collection, and an anti-interference communication module uses a UWB and Bluetooth 5.3 hybrid link, supports dynamic frequency hopping and LDPC coding.The application combines a multi-dimensional microphone array with an intelligent algorithm to realize high-precision recording equipment detection and positioning, and the system has high integration degree and can be quickly deployed in conference rooms, secure locations and the like to provide all-round protection for information security.

The present invention discloses a system and method for providing one or more user-personalised voice commands to a voice-controlled electronic device with attenuated ambient noise The system (102) is configured to generate at least one user profile of one or more user profiles for each user of the one or more users to store at least one of: one or more pre-defined voice instructions and voice frequency data associated with one or more audio signals received through one or more voice- capturing units (104). The system (102) attenuates the ambient noise by comparing a plurality of frequencies against one or more predefined ambient noise frequency data and generate one or more bandpass filters. The one or more bandpass filters isolates the one or more user-personalised voice commands associated with the at least one user profile for providing one or more user-personalised voice commands to the voice-controlled electronic device (106).

To enable a patron of a venue attending a performance to feel as if he or she is ‘front and center’ even if located much farther away or angled from a stage of the venue, system and methods that enhance audio quality of audio signals especially over the voice frequency spectrum to generate a processed audio signal including an increased voice frequency spectrum are provided. The processed audio signal may be communicated to a receiving system with a headset or other transducer (e.g., loudspeakers) that allows for ambient sounds to be heard by a patron and for the patron to control volume of the processed audio signal, thereby mixing the processed audio signal with the ambient sounds. The receiving system may be rented and an operator may control usage for both performance or long-term rentals using various control features of the processed audio signals and / or receiving system.

The present application is applicable to the field of artificial intelligence technology and provides a speech segmentation method, device, server, and storage medium. The method includes: obtaining speech data and determining whether the speech data includes multiple speech frequencies; if the speech data includes multiple speech frequencies, performing speech separation on the speech data based on a preset frequency set and the speech frequencies of each part in the speech data to obtain multiple target speech data; determining target speech segmentation points of the target speech data based on speech pause information in the target speech data and target speech text corresponding to the target speech data, and performing speech segmentation on the target speech data based on the target speech segmentation points. The present application can automatically perform speech segmentation on the speech data of the target user, which helps to improve the efficiency and accuracy of speech segmentation.

The application provides an auditory bionic voiceprint recognition method and device, electronic equipment and a storage medium. The auditory bionic voiceprint recognition method comprises the following steps: obtaining a to-be-recognized voice signal, and performing cutting processing on the to-be-recognized voice signal to obtain a plurality of to-be-recognized voice segments; inputting each to-be-recognized voice segment into a cochlea bionic filter to obtain a voice frequency feature; inputting the voice frequency feature into a pre-trained voiceprint recognition model to obtain a voiceprint recognition result; and the voiceprint recognition model is obtained based on a cochlea nucleus bionic network. The cochlea bionic filter and the voiceprint recognition model provided by the application have higher adaptability to noise and complex environments, thereby improving the reliability and efficiency of voiceprint recognition.

Microphone signals of a primary headphone are processed and either a first transparency mode of operation is activated or a second transparency mode of operation. In another aspect, a processor enters different configurations in response to estimated ambient acoustic noise being lower or higher than a threshold, wherein in a first configuration a transparency audio signal is adapted via target voice and wearer voice processing (TVWVP) of a microphone signal to boost detected speech frequencies in the transparency audio signal, and in a second configuration the TVWVP is controlled to, as the estimated ambient acoustic noise increases, reduce boosting of, or not boost at all, the detected speech frequencies in the transparency audio signal. Other aspects are also described and claimed.

The application discloses a mechanical adjustable sound cavity microphone, which comprises a shell, a microphone assembly arranged on the inner side of the sound cavity hole at the bottom of the shell, a microphone, a sound cavity, a microphone position adjusting assembly, a supporting rod, a limiting buckle, a supporting cup, a rack, a driven gear, an adjusting gear, an adjusting wheel and a rack rail bottom plate. The position of the microphone in the sound cavity is adjusted through mechanical gear cooperation, the sound cavity can be adjusted and optimized according to the voice frequency characteristics of the user, the sound cavity frequency response characteristics of specific users are optimized, the user adjusts the microphone sound cavity structure while reading a piece of sample text according to the prompt, and the structure is solidified when the best frequency response characteristics are obtained, so that the recognition and interaction effect of the voice is greatly improved. The sound cavity can also be automatically adjusted through the adjustment of the driving device, the adjustment of the driving device is controlled through software and a central control system, the sound cavity can be intelligently and automatically adjusted, and the voice recognition of specific users can be adapted.

The invention belongs to the field of building intercom, and particularly relates to an audio compensation method in an intercom system, which comprises the following steps of: S1, filling two empty mute frames; s2, setting a statistical flag bit, wherein the statistical flag bit is used for entering S4 to record the starting time when the voice frequency just establishes a call; s3, judging whether the statistical mark is set or not, if the statistical mark is set, entering S4, and if the statistical mark is not set, entering S5; s4, recording an initial millisecond number S, wherein the initial millisecond number S is used for judging whether statistics is cut off or not in S7; s5, the length L of the pcm audio data is obtained and used for S12, the actual playing amount G is calculated; and S6, recording an ending millisecond number E. According to the method, a statistical model is established and used for discovering and positioning audio trembling, and after a problem is positioned, sound is compensated through a mute data feeding method so as to eliminate the problem of sound trembling.

The application provides a signal processing method for enhancing human voice, an electronic device and a storage medium, and relates to the technical field of signal processing. In the scene of playing audio and video by the electronic device, the human voice gain parameter can be determined according to the environmental noise and the background voice frequency domain signal and the human voice frequency domain signal corresponding to the multi-channel audio stream, then the human voice frequency domain signal is gain adjusted according to the human voice gain parameter and the normalized spectral flux, and the enhanced human voice frequency domain signal is obtained. Since the gain of the human voice signal can be dynamically adjusted for the input multi-channel audio signal under different environmental noises, the effect of enhancing the human voice is achieved, and the phenomenon of large and small hearing is avoided, so the application scheme can achieve the effect that the video voice can be heard clearly without disturbing others when playing the video in a quiet environment, or the user can hear the video voice clearly when playing the video in a noisy environment, and the user experience can be improved.

The invention relates to a voice separation method and device, a model training method and device, equipment, a medium and a vehicle. The method comprises the following steps: acquiring a multi-channel voice signal of a current frame and a voiceprint frequency domain feature vector of each person in a vehicle, performing frequency domain transformation on the multi-channel voice signal of the current frame to obtain a multi-channel voice frequency domain signal, and aligning the voiceprint frequency domain feature vector of each person in the vehicle with the multi-channel voice frequency domain signal on a time frame to obtain a multi-channel voice frequency domain signal. The target voiceprint frequency domain feature vector of each person is obtained, the target voiceprint frequency domain feature vector of each person and the multi-channel voice frequency domain signal are input into a voiceprint auxiliary separation model, and an audio mask corresponding to each person in the vehicle is obtained through the voiceprint auxiliary separation model; the audio mask corresponding to each person in the vehicle is utilized to perform plural masking processing on the multi-channel voice frequency domain signal to obtain the voice signal corresponding to each person in the vehicle, and the audio mask corresponding to each person in the vehicle is utilized to perform voice separation, so that the separation performance is improved.

To enable a patron of a venue attending a performance to feel as if he or she is 'front and center' even if located much farther away or angled from a stage of the venue, system and methods that enhance audio quality of audio signals especially over the voice frequency spectrum to generate a processed audio signal including an increased voice frequency spectrum are provided. The processed audio signal may be communicated to a receiving system (e.g., wireless headset, cochlear implants, mobile electronic devices, etc.) with a headset or other transducer (e.g., loudspeakers) that allows for ambient sounds to be heard by a patron and for the patron to control volume of the processed audio signal, thereby mixing the processed audio signal with the ambient sounds. The receiving system may be rented and an operator may control usage for both performance or long-term rentals using various control features.

The invention provides a sound conversion model training method and device, and a sound conversion method and device, and relates to the technical field of data processing, and the method comprises the steps: obtaining sound production person audio samples of a plurality of different sound production persons; wherein each sounder audio sample corresponds to one fundamental frequency label; training a preset sound conversion model according to the automatic speech recognition sample features and the sound producer sample feature vectors corresponding to the sound producer audio samples to obtain a first sound conversion model; training the first sound conversion model based on the forged automatic speech recognition features corresponding to each speaker audio forged sample of the target speaker and the speaker sample feature vector to obtain a second sound conversion model; wherein the sounder audio forged sample is obtained according to a real sounder audio sample of the target sounder; and the second sound conversion model is used for outputting fundamental frequency information corresponding to the voice frequency of the sounder based on the automatic voice recognition features corresponding to the voice frequency of the sounder and the feature vector of the sounder.

Microphone signals of a primary headphone are processed and either a first transparency mode of operation is activated or a second transparency mode of operation. In another aspect, a processor enters different configurations in response to estimated ambient acoustic noise being lower or higher than a threshold, wherein in a first configuration a transparency audio signal is adapted via target voice and wearer voice processing (TVWVP) of a microphone signal to boost detected speech frequencies in the transparency audio signal, and in a second configuration the TVWVP is controlled to, as the estimated ambient acoustic noise increases, reduce boosting of, or not boost at all, the detected speech frequencies in the transparency audio signal. Other aspects are also described and claimed.

The application belongs to the technical field of acoustic antennas, and particularly relates to a dual-band high-directivity topological acoustic receiving antenna. The left side surface, the upper side surface, the lower side surface and the rear end surface are mutually spliced to form a single-port acoustic waveguide, so that the right side of the single-port acoustic waveguide is an open port, the single-port acoustic waveguide is a parallelogram structure, and a topological phononic crystal and a sound-absorbing sponge are arranged in the single-port acoustic waveguide. Through the structure, high-directivity reception of two different voice frequency band acoustic waves can be realized, and an anti-interference high-security acoustic communication function is realized. Meanwhile, the acoustic receiving antenna has the characteristics of dual working frequency bands, small size, light weight and convenience in carrying, has high-directivity receiving effect on two independent wide frequency band acoustic signals, provides a feasible solution for voice frequency band acoustic signal directivity anti-interference and security transmission, and can be applied to the field of artificial intelligence robots.

The present invention discloses a vehicle-mounted voice processing method and system. The present invention first filters the frequency band of the sound signal below the lower limit of the human voice frequency, so that it is not necessary to perform spectrum analysis on the frequency band of the sound signal between the lower limit of the noise signal frequency and the lower limit of the voice frequency; then performs spectrum analysis on the frequency band of the sound signal below the upper limit of the noise signal frequency, so that it is not necessary to perform spectrum analysis on the frequency band of the sound signal between the upper limit of the noise signal frequency and the upper limit of the voice frequency, thereby greatly reducing the frequency band of the sound signal that needs spectrum analysis, and the workload and cost of vehicle-mounted voice noise reduction are small.

The invention discloses an intelligent child development detection bracelet based on multi-modal mouth shape analysis. The bracelet comprises a wearable bracelet main body, the multi-mode sensor module comprises a visual sensor used for capturing mouth shape dynamic images and an audio sensor used for collecting voice; the data processing unit is used for preprocessing sensor data and executing multi-mode mouth shape analysis; the intelligent development evaluation module is used for generating a development score and a risk index according to the analysis result; the user interface module is used for displaying information and providing feedback; according to the invention, a high-resolution visual sensor and a multi-microphone array are integrated on the bracelet, so that multi-modal data collaborative acquisition of dynamic images and synchronous voices of mouth shapes of children is realized, and fusion analysis is carried out by adopting a model based on a convolutional neural network and a recurrent neural network. The time sequence relevance among the lip motion trail, the mouth opening amplitude and the voice frequency spectrum feature can be extracted, and therefore the problem of one-sidedness of single-mode monitoring is solved.

The invention discloses a control method and control equipment for preventing mistaken door opening of a motor train unit, relates to the technical field of high-speed rail traffic safety detection and is used for preventing a driver from opening a train door mistakenly. In the method, the control equipment firstly performs rapid and low-power-consumption preliminary screening by utilizing rhythm characteristics of broadcast prompt tones to determine a pre-selected opening side, so that simultaneous physical detection on two sides is avoided. Then, aiming at the preselected opening side, through emitting a modulation infrared pulse and analyzing a mean value and a variance of a reflection signal, the existence of a static entity platform is physically confirmed from two dimensions of whether an object exists and whether the object is static, and under the condition that the static entity platform exists on the preselected opening side, the preselected opening side is opened; and the control equipment outputs an unlocking pulse signal to the door opening button protection device on the pre-selected opening side, so that a door opening mistaken accident caused by voice frequency misrecognition or driver fatigue is prevented, and the safety of passengers getting on and off the bus is improved.

The invention discloses an electroencephalogram language decoding method and application thereof, and the method comprises the following steps: opening recording software, playing and recording the pronunciation of a character needing to be recorded, and obtaining a recording file, the pronunciation being the sound made by artificial intelligence; 2, performing Fourier transform on the sound recording file to obtain an amplitude-frequency diagram, selecting frequencies corresponding to a plurality of peak values in a frequency range of 1-1100Hz according to the amplitude-frequency diagram so as to complete audio decoding and obtain N decoding frequencies, writing the N decoding frequencies obtained in the step 2 into a sound production program from small to large, and playing for each decoding frequency for 1-20s, and opening recording software for recording, finally executing a sound production program, and storing the recording software to obtain an electroencephalogram voice frequency file. The playing of the electroencephalogram speech file obtained by the electroencephalogram speech decoding method has a regulation and control effect on sleep, and the electroencephalogram speech decoding method is not limited by languages and can decode multiple languages.

The application discloses a kind of wireless data transmission formula sea wave height measuring device and measuring method, mainly solve the problem of high cost of existing sea wave height measuring equipment.The device includes float component and measuring component, two are fixed as a whole, float component includes liquid level sensor circuit, electromagnetic valve, high-pressure carbon dioxide cabin and air bag air bag, liquid level sensor contacts seawater after electromagnetic valve is opened, carbon dioxide gas rushes into air bag, so that it floats on sea surface;Measuring component includes acceleration sensor, data processor, voice conversion chip, ultrashort wave transmitter, power amplifier circuit, antenna and power supply.Measuring component keeps the state of stable motion under float component, and calculates sea wave height by collecting sea wave acceleration data, and converts it into voice frequency modulation signal and emits to the air above sea surface, for pilot to listen to the voice broadcast of sea wave height.The device is small in size, light in weight, low in measurement cost, and can be used to ensure the safe take-off and landing of aircraft on water surface.

To enable a patron of a venue attending a performance to feel as if he or she is ‘front and center’ even if located much farther away or angled from a stage of the venue, system and methods that enhance audio quality of audio signals especially over the voice frequency spectrum to generate a processed audio signal including an increased voice frequency spectrum are provided. The processed audio signal may be communicated to a receiving system with a headset or other transducer (e.g., loudspeakers) that allows for ambient sounds to be heard by a patron and for the patron to control volume of the processed audio signal, thereby mixing the processed audio signal with the ambient sounds. The receiving system may be rented and an operator may control usage for both performance or long-term rentals using various control features of the processed audio signals and / or receiving system.

The application relates to the field of air traffic control and speech recognition technology, and particularly relates to an air traffic control instruction end-to-end speech recognition method under high noise conditions, which comprises the following steps: pre-processing to-be-recognized speech to extract original speech frequency features; adopting an adaptive attention noise reduction module to perform noise reduction intensity control, frequency band gain control and pitch tracking processing on the original speech frequency features to obtain noise reduction speech; pre-processing the noise reduction speech to extract noise reduction speech frequency features; adopting a speech recognition module to perform shared encoder coding, connection time sequence classification decoder decoding and attention-based encoder-decoder decoding on the noise reduction speech frequency features to obtain output speech recognition transcription text; and the application can improve the speech recognition accuracy of air traffic control instructions under high noise conditions.

In some embodiments, a method includes: dividing an audio input into speech segments and non-speech segments using at least one processor; estimating a time-varying noise spectrum of the non-speech segment for each frame in each non-speech segment using at least one processor; estimating a speech spectrum of the speech segment for each frame in each speech segment using at least one processor; identifying one or more non-speech frequency components in the speech spectrum for each frame in each speech segment; comparing the one or more non-speech frequency components with one or more corresponding frequency components in a plurality of estimated noise spectra; and selecting an estimated noise spectrum from the plurality of estimated noise spectra based on the comparison results.

In one aspect, a device includes a processor and storage with instructions executable to identify a first frequency range of sound associated with an environment and to determine whether the first frequency range matches a second frequency range at least to within a threshold. The second frequency range is associated with a voice of a digital assistant. Responsive to the first frequency range matching the second frequency rage at least to within the threshold, the instructions are executable to change a setting for the digital assistant so that the voice of the digital assistant outputs audio in a third frequency range that does not match the first frequency range at least to within the threshold.