30results about How to "Good voice effect" patented technology

The invention provides a double-network digital mobile communication terminal and a double-network communication method thereof. The communication terminal comprises a private network communication unit, which is used for private network signal communication; a public network communication unit, which is used for public network signal communication; and a central control processing module, which is respectively coupled with the private network communication unit and the public network communication unit to control the private network communication unit and the public network communication unit.

The invention discloses a method and a device for handling receiving voice, which belongs to the technical field of communication. The method comprises the steps of: acquiring the use condition information of a voice output device; respectively regulating the signal intensity of the receiving voice received from each frequency band according to the use condition information of the voice output device, and outputting the regulated receiving voice through the voice output device, wherein the frequency bands are obtained by dividing the voice bandwidth of the voice output device. The device comprises an acquisition module and a handling module. In the invention, by acquiring and in accordance with the use condition information of the voice output device, the signal intensities of the receiving voice received from each frequency band are regulated respectively, the signal intensity of voice in a frequency band with higher signal intensity attenuation is intensified greatly, the signal intensity of voice in a frequency band with lower signal intensity attenuation is intensified slightly, and thus, the problems that the receiving voice is sharp and the volume is low caused by high signal intensity attenuation of the voice in a low frequency band in the process of using a mobile phone can be solved, and the integral voice effect of the receiving voice can be improved.

In frequency signals obtained by converting input audio signals from time-domain signals to frequency-domain signals, a level control value setting unit 5 establishes a level control value for reducing the levels of spectrums at a noise-components level. A level control value smoothing unit 6 carries out a smoothing process of smoothing the level control value established by the level control value setting unit 5 temporally. A spectral adjustment unit 8 multiplies the level control value after the smoothing process by the frequency signals, performing a level control.

In frequency signals obtained by converting input audio signals from time-domain signals to frequency-domain signals, a level control value setting unit 5 establishes a level control value for reducing the levels of spectrums at a noise-components level. A level control value smoothing unit 6 carries out a smoothing process of smoothing the level control value established by the level control value setting unit 5 temporally. A spectral adjustment unit 8 multiplies the level control value after the smoothing process by the frequency signals, performing a level control.

The embodiment of the invention relates to the technical field of image processing and particularly relates to a video conference management method and device. The method is applied to electronic equipment in communication connection with multiple cameras, each camera is applied to a conference desk arranged in a conference hall, the conference desk is provided with microphones, and the microphones are in communication connection with the electronic equipment. The method comprises the steps of acquiring a real-time image shot by each camera, splicing the real-time images so as to obtain a real-time integral image of the conference hall, judging whether at least one conventioneer in multiple conventioneers is in contact with the microphones so as to collect voice according to the real-timeintegral image, if yes, controlling the microphones to be switched on, receiving real-time voice information collected by the microphones, and synchronously sending the real-time integral image and the real-time voice information. According to the method, the intelligent level of a video conference system can be increased, and the voice effect and interaction performance of the video conference system can be improved.

The invention relates to a wireless communication anti-interference system, which belongs to the technical field of voice signal processing devices in the field of communications. The wireless communication anti-interference system comprises an input interface module and an output interface module which are connected through signal transmission, and is characterized in that: a speech channel switching module is arranged between the input interface module and the output interface module in a connection mode; a noise prediction module is arranged at an output end of the input interface module connected with the speech channel switching module in a fit mode; the noise prediction module is in control connection with the speech channel switching module; and a voice enhancement module which is in transmission connection with the speech channel switching module is arranged on the speech channel switching module in a fit mode. The wireless communication anti-interference system has novel design and reasonable structure, and can enable all speech channels to achieve the optimum voice effect by using one voice enhancement module only through the combined use of the noise prediction module and the speech channel switching module; and moreover, compared with the prior art that each speech channel needs to be provided with a voice enhancement module, the wireless communication anti-interference system greatly simplifies the structure, and significantly reduces the cost.

The invention discloses a camera with a built-in speech microphone. The camera comprises a base, a hose connection bar, the built-in speech microphone, a camera main body, a rotary joint, a snapshot button and an anti-jamming device, wherein the base is connected with one end of the hose connection bar, and the other end of the hose connection bar is connected with the camera main body through the rotary joint; the anti-jamming device is arranged in the camera main body; the built-in speech microphone is arranged at the lower end of the camera main body; and the snapshot button is arranged at the upper end of the camera main body. The camera main body can rotate by 360 degrees, the anti-jamming function is strong, a favorable speech sound effect is achieved, and one-click shooting can be realized.

The invention discloses an automatic processing method for city window weather dubbing of a TV weather forecast program. The method comprises the following steps: translation: translating city weather codes to obtain city names and weather and / or temperature entries; matching: matching the translated entries with the corresponding storage entries in a voice library; synthesizing: synthesizing the matched storage voice entries into audio data according to the sequence of the city name and weather and / or temperature; and time setting: by inserting mute date before and behind the entry audio data to generate dubbing data, enabling the time length of the dubbing data to be equal to a set time length. The invention also discloses a system adopting the method. The automatic dubbing synthesized voice is good in effect and can replace human voice, thereby improving program quality, avoiding external influence, freeing hosts, preventing mistake making and improving work efficiency by times.

The invention discloses a method and device for realizing stereo call on an ESCO link, a medium and a server, and belongs to the technical field of voice processing. The method for realizing the stereo call on the ESCO link comprises the following steps: establishing Bluetooth communication between a microphone and intelligent terminal equipment through the ESCO link; receiving human voice data and accompaniment data from the intelligent terminal equipment through the microphone; and carrying out sound mixing processing on the human voice data and the accompaniment data through the microphone, and sending a sound mixing processing result to another intelligent terminal equipment. According to the application, the call effect and quality are improved, and the user experience is improved.

The embodiment of the invention provides a thin client communication method and device, and relates to the communication field. A thin client can be in direction communication with user equipment while coding and decoding voice media stream during communication, thereby promoting user experience and avoiding problems of long voice time delay, poor voice quality and the like in the prior art. The method comprises the following steps: sending a first call request message to the user equipment through protocol IP telephone gateways which are interconnected through a network by the thin client; receiving a first response message that corresponds to the first call request message and is sent by the user equipment through the IP telephone gateways by the thin client; performing communication with the user equipment by the thin client according to the first response message, wherein the thin client codes first speech information from a user into the first voice media stream during communication, then sends to the user equipment, or decodes the second voice media stream from the user equipment into second speech information during communication, and then outputs to the user.

An embodiment of the invention relates to the technical field of communication, and discloses a method and a device for tracking a speaker by using a microphone, and computing equipment. The method comprises the steps of: recognizing the speaker through sonar imaging; acquiring mouth coordinates of the speaker and a distance from the microphone to the mouth coordinates; and controlling the microphone to directly face the mouth coordinates of the speaker in a first preset range in real time by a mechanical arm and dynamically track the speaker. According to the method and the device of the invention, the accurate position of the speaker can be obtained from a sonar, the development of a microphone front-end algorithm to be more centralized in the development of a certain amplitude angle through the dynamic adjustment of an angle between the microphone and the speaker, can save the development time, and achieves a better voice effect.

The invention provides a voice call-out method, system and device and a storage medium, and the method comprises the steps that a call module carries out the voice call-out of a customer, receives a voice reply of the customer after successful call-out, calls a voice processing engine to generate a machine reply according to the voice reply, and transmits the machine reply to the customer. According to the invention, interaction between a voice outbound machine and a client is realized, interaction behaviors are monitored, outbound standardization and universality which are not based on a specific scene are improved, repeated development is avoided, and machine resources are fully utilized.

The invention discloses a method and a device for handling receiving voice, which belongs to the technical field of communication. The method comprises the steps of: acquiring the use condition information of a voice output device; respectively regulating the signal intensity of the receiving voice received from each frequency band according to the use condition information of the voice output device, and outputting the regulated receiving voice through the voice output device, wherein the frequency bands are obtained by dividing the voice bandwidth of the voice output device. The device comprises an acquisition module and a handling module. In the invention, by acquiring and in accordance with the use condition information of the voice output device, the signal intensities of the receiving voice received from each frequency band are regulated respectively, the signal intensity of voice in a frequency band with higher signal intensity attenuation is intensified greatly, the signal intensity of voice in a frequency band with lower signal intensity attenuation is intensified slightly, and thus, the problems that the receiving voice is sharp and the volume is low caused by high signal intensity attenuation of the voice in a low frequency band in the process of using a mobile phone can be solved, and the integral voice effect of the receiving voice can be improved.

Embodiments of the present invention provide a thin terminal communication method and apparatus, relating to the communications field. The thin terminal is capable of directly communicating with a user equipment, and encoding and decoding a phonetic media stream in a communication process, which improves user experience, and avoids the problems of a great phonetic delay and the poor phonetic quality in the prior art. The method comprises: sending, by the thin terminal, a first call request message via a protocol IP telephone gateway interconnected between networks to the user equipment; receiving, by the thin terminal, a first response message corresponding to the first call request message and from the user equipment via the IP telephone gateway; communicating, by the thin terminal, with the user equipment according to the first response message, sending, by the thin terminal, a first phonetic media stream to the user equipment after first utterance information from the user in the communication process is encoded into a first phonetic media stream, or outputting, by the thin terminal, to the user after a second phonetic media stream from the user equipment in the communication process is decoded into second utterance information.

The invention relates to a method and a system for dynamically packing speech frames; the method comprises the following steps: both sides transmitting speech frames packet the speech frames in terms of single frame when no blockage occurs in the transmission network, and dynamically adjust the number of packing frames with set step-length, and packet the speech frames with the adjusted number when the blockage occurs in the transmission network; and the system comprises a base station controller and a mobile exchange center. The invention adopts single-frame packing when no blockage occurs in the network, thus guaranteeing no time delay of speech; and the invention adopts multi-frame packing when the blockage occurs, and then adopts single-frame packing again after the blockage is relieved, thus overcoming the inherent defects of single single-frame packing or multi-frame packing, and being capable of maximally optimizing the speech effect on the transmission network.

The embodiment of the present invention relates to the field of communication technology, and discloses a method, device, and computing device for tracking a speaker using a microphone. The method includes: identifying the speaker through sonar imaging; The distance of the mouth coordinates; through the real-time control of the mechanical arm, the microphone is directly aligned with the mouth coordinates of the speaker within the first preset range, and the speaker is dynamically tracked. The embodiment of the present invention can obtain from sonar To the exact position of the speaker, by dynamically adjusting the angle between the microphone and the speaker, the front-end algorithm development of the microphone is more focused on the development of a certain range of angles, which can save development time and achieve better voice effects.

The present invention provides a method, system, device and storage medium for a voice outbound call. The method includes: a calling module makes a voice outbound call to a customer, receives a voice reply from the customer after a successful call, and invokes a voice processing engine to generate an outbound voice call based on the voice reply. The machine replies and transmits the machine reply to the customer. The invention realizes the interaction between the voice outbound machine and the customer, and monitors the interaction behavior, and at the same time improves the standardization and versatility of the outbound call that can not be based on a specific scene, is beneficial to avoid repeated development, and make full use of machine resources.

The invention relates to the technical field of hearing-aid speech enhancement, and particularly relates to a speech enhancement method and device based on dual-channel neural network time-frequency masking and hearing-aid equipment. The method comprises the steps: receiving a speech signal through employing two microphones under noise and reverberation conditions; performing preliminary speech enhancement on each path of received microphone signals through the trained single-channel neural network, and removing the noise in the same direction as the target speech; different from traditional positioning which blindly depends on signal energy, enabling the self-adaptive ratio mask to provide a view of a target signal, accurately identifying a speech dominant time frequency unit on each microphone channel, using the time frequency units for sound source positioning, and ensuring that high positioning precision is obtained under noise and reverberation conditions; and inputting the calculated weight into a WPD beam former to remove noisy speech in different directions from the target speech and suppress room reverberation so as to obtain enhanced speech with good speech quality and high intelligibility.

The present invention disclosed a voice enhancement system based on low -cost wearable sensors.The system includes wearable devices and intelligent devices, which include wearable devices include micro -control unit, audio processing unit, filter, in -ear headset and piezoelectric ceramics.Under the control, the sound signals and cervical vibration signals in the ear canal are collected when the user speaks, and it is transmitted to the smart terminal.Time -frequency diagram or time sequence, input to the deep learning model of pre -training, and obtain the target quality voice signal. The resolution of the target quality voice signal and the user's hearing sense is better than the signal of the wearable device collection.The present invention can convert low -cost wearable sensor signals into high -quality signals in the case of protecting user privacy and suitable for daily use.

The application discloses a data processing method, device, electronic equipment and storage medium, and relates to the technical fields of voice interaction technology, automatic driving, smart home and deep learning. The specific implementation scheme is: obtain 16-bit model data to be processed; obtain a 4-bit load instruction; load four 4-bit model data from the 16-bit model data into a 64-bit register according to the 4-bit load instruction . Therefore, the present application can load four 4-bit model data from 16-bit model data into 64-bit registers by adding a 4-bit load instruction, thereby improving the processing efficiency of low-bit model data and effectively reducing network communication. The sequential access throughput caused by the cost cannot be improved and broken through. Furthermore, by adding custom instructions, the chip can store a larger neural network without increasing the chip area to achieve better voice effects.

The embodiment of the present invention provides a microphone matrix adjustment method, and the method comprises the steps: determining the shape of a socket matrix board according to a preset position of a target product; setting at least one microphone unit on a socket matrix board according to a voice algorithm to obtain a microphone matrix; detecting the microphone matrix, and adjusting the position and the number of the microphone units until a preset condition is satisfied when it is detected that the microphone unit does not meet the preset condition. The microphone device and the audiodevice are also disclosed in the embodiments of the present invention. Since the microphone matrix in the embodiment of the present invention is divided into two parts of the socket matrix board andthe microphone units, and the microphone units can be flexibly disposed on the socket matrix board, the position, number and relative angles of microphone units can be easily adjusted when detecting the microphone matrix, so as to flexibly cooperate with the voice algorithm, thereby meeting different product requirements, greatly reducing the cost of research and development products, and improving design efficiency.

The invention relates to a method and a system for dynamically packing speech frames; the method comprises the following steps: both sides transmitting speech frames packet the speech frames in terms of single frame when no blockage occurs in the transmission network, and dynamically adjust the number of packing frames with set step-length, and packet the speech frames with the adjusted number when the blockage occurs in the transmission network; and the system comprises a base station controller and a mobile exchange center. The invention adopts single-frame packing when no blockage occurs in the network, thus guaranteeing no time delay of speech; and the invention adopts multi-frame packing when the blockage occurs, and then adopts single-frame packing again after the blockage is relieved, thus overcoming the inherent defects of single single-frame packing or multi-frame packing, and being capable of maximally optimizing the speech effect on the transmission network.

The invention provides a model training method and device, a voice separation method and device and electronic equipment, and the method comprises the steps: the voice features of a sound signal are input into N pre-trained first neural network models, and N output results are obtained, wherein the N output results are voice features of the voice of the speaker corresponding to N pickup areas separated from the sound signal, and N is an integer greater than 1; voice features of the sound signals are input into a second neural network model, the second neural network model is trained, and a loss function used for training the second neural network model is determined based on the N output results. According to the embodiment of the invention, voice separation is carried out by adopting the trained second neural network model, so that the accuracy of voice separation can be improved.

The present invention discloses an analog voice handle for testing. The analog voice handle comprises a voice coding and decoding chip, a permalloy audio transformer, a T-type attenuator and an optocoupler; two sets of transformers are arranged in the permalloy audio transformer; the impedances of the two sets of transformers are 150: 150 and 600: 600 respectively; one terminal of the input end of each of the two sets of transformers is connected with system ground; one terminal of the output end of each of the two sets of transformers is connected with voice handle ground; the non-grounding ends of the two sets of transformers are an MIC lead and a PHONE lead respectively; the voice handle ground and the system ground are connected with each other through a capacitor; the T-type attenuator includes resistors R1, R2 and R3; the first end of the resistor R2 is connected with the AO end of the voice coding and decoding chip; the second end of the resistor R2 is connected with the first end of the resistor R1 and the first end of the resistor R3; the second end of the resistor R3 is grounded; the second end of the resistor R1 is connected with the non-grounding end of the input end of the 150: 150 transformer in the permalloy audio transformer; and the non-grounding end of the input end of the 600: 600 transformer in the permalloy audio transformer is connected with the AI end of the voice coding and decoding chip; and one output end of the optocoupler is a P TT lead.