25results about How to "Improve audio quality" patented technology

The invention relates to the technical field of audio processing, in particular to an audio optimization method and device based on artificial intelligence, computer equipment and a medium. According to the method, a linear layer is used for mapping a spectrum envelope of a to-be-optimized audio to obtain an envelope feature, an embedding layer is used for embedding a standard audio parameter into a parameter feature, a prediction model is used for predicting a fusion feature of the envelope feature and the parameter feature to obtain a prediction pitch curve, and a noise adding model is used for adding noise to a Mel spectrum of the to-be-optimized audio to obtain a predicted pitch curve. Calculating the noise of the noise addition result by using a noise estimation model to obtain predicted noise, updating the noise estimation model according to the predicted noise, the real noise and the predicted pitch curve, calculating the reference noise of the noise addition result by using the updated noise estimation model, and de-noising the noise addition result according to the reference noise. The noise estimation model is optimized in combination with pitch information, so that the denoising process meets the pitch requirement, and the audio optimization effect is improved.

The invention provides a central control method and system for synchronously playing sound equipment in multiple rooms of a hotel, and is applied to the field of data synchronous processing. According to the invention, the audio stream transmission path diagram is constructed based on the network topology preset by the audio routing table, so that the system can visually check the audio stream transmission path of each room and identify and optimize links possibly causing delay, synchronous playing of cross-room sound equipment can be realized, audio signals can be ensured to be transmitted at the lowest delay, the audio quality is improved, and the user experience is improved. And meanwhile, by dynamically adjusting the compression ratio of audio coding, the system can relieve the network bandwidth pressure on the premise of keeping the audio quality, and audio transmission interruption or quality reduction caused by insufficient network bandwidth is avoided.

This invention provides an audio dynamic range control circuit and an audio device. The audio dynamic range control circuit includes an audio zero-crossing detection circuit, a preprocessing circuit, a first control circuit, a first gain selection circuit, a gain multiplication processing circuit, a second control circuit, a third control circuit, and a second gain selection circuit. The audio dynamic range control circuit provided by this invention achieves an effect similar to the existing RMS algorithm by performing absolute value processing on the linear domain sampling data of the audio signal, while simplifying the hardware circuit implementation. Simultaneously, by detecting the zero-crossing position of the audio signal through the audio zero-crossing detection circuit, and updating the gain value at subsequent zero-crossing positions when a sampling point with high energy is detected, the nonlinearity of the output signal can be greatly improved and auditory discomfort can be eliminated, effectively improving audio quality.

The invention discloses a high-signal-to-noise-ratio directional sound collection processing method, system and platform suitable for remote monitoring and a storage medium, and the method comprises the steps: obtaining to-be-processed audio original signal data in an environment in real time based on remote monitoring, processing the audio original signal data at a time, and generating corresponding first audio signal data; filtering the first audio signal data to generate corresponding second audio signal data, and converting the second audio signal data to generate third audio signal data corresponding to the second audio signal data; in combination with a deep learning model, performing noise reduction processing on the third audio signal data, and generating and backing up fourth audio signal data corresponding to the third audio signal data; according to the invention, noise can be effectively suppressed, the signal-to-noise ratio of an audio signal is improved, and the sound collection quality is improved; data tampering is prevented, and authenticity and integrity of audio data and safety of sound collection are ensured.

The present invention relates to the field of speech recognition processing technology, and is particularly suitable for obtaining training data for machine learning models used in speech recognition transcription. In view of the high cost of data acquisition, high consumption, poor data quality / accuracy, and poor screening effect of existing pseudo-label data in different scenarios, a method, system, device, and medium for screening training data of the present invention are proposed, aiming to solve the technical problem of how to screen high-quality training data for models such as speech recognition, search, and transcription based on the accuracy of pseudo-labels of semi-supervised learning. To this end, the method of the present invention uses the mean number of node links of the decoding path in decoding to sort the decoding results to screen the pseudo-label speech data with the highest sorting as model training data. The screening efficiency and data quality are improved, and the cost and consumption are reduced.

The application provides a model training method, a speech conversion method and device, equipment and a storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: obtaining sample audio data of a sample speaker object; inputting the sample audio data into a neural network model comprising an encoding network and a decoding network; performing reconstruction processing on the sample audio data by using the encoding network to obtain initial audio data; performing speech alignment on the initial audio data to obtain a sample audio embedding vector; performing decoupling processing on the sample audio embedding vector, pre-acquired sample pitch parameters and a sample timbre feature vector by using the decoding network to obtain synthesized audio data; performing loss calculation on the synthesized audio data and sample speech data by using a loss function to obtain a model loss value; and performing parameter updating on the neural network model according to the model loss value to train the neural network model and obtain a speech conversion model. The application can improve the speech conversion effect.

The embodiment of the invention relates to the technical field of vehicle-mounted cabins, and discloses a cabin talkback integrated control method, system and device and a medium. When a user carries out talkback operation, the first cabin domain controller issues a PTT instruction for activating the first interphone; after receiving the PPT instruction, the first interphone collects a voice signal and transmits the voice signal to a first cabin area controller through a BLE Bluetooth link, and the first cabin area controller receives and stores the voice signal; when a user carries out receiving operation, the second interphone receives a voice signal of the first interphone and transmits the voice signal to the second cockpit domain controller through the BLE Bluetooth link; and the second cockpit domain controller receives and stores the voice signal and controls a loudspeaker to play the voice signal. The technical problem that a cockpit is low in control capability of intercom equipment can be at least solved.

Embodiments of the present application provide an audio data processing method, a server and a program product. The method is applied to a server, and the method comprises: performing redundant coding processing on to-be-processed audio data to obtain coded audio data, in a case where it is determined based on network state information that the redundant coding process is to be performed; and performing a first redundancy processing operation on the coded audio data to obtain processed audio data. The method significantly improves the quality of audio transmission in a weak network environment, reduces audio delay, freezing and distortion, and improves user experience.

The utility model discloses a sound acquisition device for short video production, and relates to the technical field of sound acquisition devices, the sound acquisition device comprises a bottom plate, the top of the bottom plate is fixedly connected with a support shell, the sound acquisition device also comprises a protection mechanism, the front end of a first threaded rod is fixedly connected with a first knob, the first knob drives the first threaded rod to rotate, and the first threaded rod is fixedly connected with the support shell. A first threaded rod drives a movable shell to move forwards on the inner wall of a fixed shell, the fixed shell drives tow rods on the two sides to move, the tow rods drive sliding plates on the two sides to slide on the inner wall of a limiting shell and move towards the center of the inner wall of the limiting shell, and the two sliding plates drive two protective shells to slide on the inner wall of a supporting shell. The two protective shells are close to each other to form a closed space to protect the sound receiving microphone, so that dust in the air is effectively prevented from being attached to and accumulated on a pickup head of the microphone to cause blockage, and the sound definition of recording and the recording quality are not affected.

The invention discloses an audio transmission method and device, equipment and a storage medium, and relates to the field of digital communication, and the method comprises the steps: obtaining a digital audio signal; encoding the digital audio signal according to a preset audio encoding format to obtain an encoded digital audio signal; partitioning the coded digital audio signal to obtain a first data block; performing interleaving processing on the first data block to obtain a second data block; and sending the second data block to a second terminal, so that the second terminal recovers the second data block to obtain a fourth data block, and performs audio reconstruction based on the fourth data block to obtain a reconstructed digital audio signal.

The invention provides an audio channel parameter adjusting method and device, a medium and a program product, and relates to the technical field of audio engineering. The audio channel parameter adjusting method comprises the following steps: detecting parameters of an audio channel; if it is detected that the parameters change, marshalling corresponding to the audio channels is obtained, the changed parameters are applied to the audio channels in the marshalling, the marshalling comprises at least two audio channels, and the marshalling is created based on a received marshalling instruction. According to the embodiment of the invention, unified adjustment of multiple audio channel parameters can be realized in an audio channel marshalling mode, the adjustment speed is high, the efficiency is high, the consistency of audio signals is effectively ensured, and the audio quality and the user experience are improved.

The embodiment of the invention provides an audio correction method and device, electronic equipment and a storage medium, and belongs to the technical field of voice recognition. The method comprises the following steps: acquiring a sample audio, and acquiring an original audio correction model; performing pitch feature extraction on the sample audio according to a pitch encoder; performing pitch correction processing on the original pitch vector according to the generator; performing content feature extraction on the sample audio according to a content encoder; performing content correction processing on the original content vector according to the generator; performing audio fusion processing on the corrected pitch vector and the corrected content vector according to a generator; performing audio quality discrimination on the corrected audio according to a target discriminator; performing parameter adjustment on the original audio correction model according to the target discrimination data to obtain a target audio correction model; and performing audio correction on a pre-acquired target audio according to the target audio correction model. The embodiment of the invention can be applied to financial science and technology, the audio quality is improved, and the accuracy of speech recognition is improved.

The application discloses a phoneme duration prediction method based on VITS neural spline flow improvement, introduces a method based on neural spline flow by improving the random duration predictor in the VITS model, effectively improves the accuracy of phoneme duration prediction, through detailed analysis of input data distribution and block interval division, the model can perform more effective coupling transformation in each block interval, which enables the model to predict the duration of phonemes with higher accuracy in actual speech synthesis applications. In addition, the application can speed up the training process and focus on optimizing the performance of the duration prediction module by freezing the model, this method not only improves the training efficiency, but also ensures the accuracy of the model in predicting the duration of phonemes, and the mean and variance of the error are used as evaluation indexes to further verify the prediction accuracy of the model, and the transformation interval boundary value is optimized by adjusting the hyperparameter B, realizing the optimization of the performance of the model.

The utility model discloses a sounding unit, which comprises a vibration assembly, a basin stand, and a magnetic circuit assembly and a shaping support chip which are arranged on the basin stand, the vibration assembly comprises a voice coil, the magnetic circuit assembly comprises a central magnetic steel and a plurality of side magnetic steels, the central magnetic steel is located in an area surrounded by the voice coil, and the plurality of side magnetic steels are wound on the periphery of the voice coil; the multiple pieces of edge magnetic steel comprise two pieces of first edge magnetic steel and two pieces of second edge magnetic steel, a gap is formed between the adjacent edges of the two pieces of second edge magnetic steel, the gap and the adjacent edges of the two pieces of second edge magnetic steel define a containing area, the two pieces of first edge magnetic steel are arranged in the containing area, and the two pieces of second edge magnetic steel are arranged in the containing area. First gaps are formed between the two ends of the first side magnetic steel and the adjacent second side magnetic steel; the shaping support chip is provided with a connecting arm, one part of the connecting arm is bonded with the bottom of the voice coil, and the other part extends outwards from the first gap. The side magnetic steel is of a surrounding voice coil structure, the whole voice coil is almost completely surrounded by a magnetic circuit, and the performance is improved.

The utility model relates to the technical field of wireless earphones, in particular to an ear clip type wireless earphone, which comprises a sounding piece provided with a sounding hole. The ear hook is connected with the sound production part; the battery compartment is arranged at one end, far away from the sound production part, of the ear hook; in a wearing state, the ear hooks are configured to bypass auricles from the auricular conchae of a user, so that the battery bins are located between the head of the user and the auricles, and the sound production holes face the ear canals of the user; one end, far away from the ear hook, of the sounding piece is provided with a first tangent plane, one end, far away from the ear hook, of the battery compartment is provided with a second tangent plane, the first tangent plane is parallel to the second tangent plane, and the distance between the first tangent plane and the second tangent plane is 1-3mm. The earphone provided by the utility model improves the sound quality definition and immersion under the condition that the wearing comfort is ensured, and reduces the sound leakage of the earphone.

The utility model discloses a sounding unit, which comprises a vibration assembly, a basin stand, and a shaping support chip and a magnetic circuit assembly which are arranged on the basin stand, the vibration assembly comprises a voice coil, the magnetic circuit assembly comprises a central magnetic steel and two side magnetic steels, the central magnetic steel is located in an area surrounded by the voice coil, and the two side magnetic steels are located in an area surrounded by the voice coil. The voice coil is located in an area defined by the two pieces of side magnetic steel, and a first gap is formed between the two pieces of side magnetic steel; the shaping support chip is provided with a connecting arm, one part of the connecting arm is bonded with the voice coil, and the other part of the connecting arm extends outwards from the first gap so as to extend out of an area defined by the two pieces of edge magnetic steel. According to the utility model, the voice coil is arranged in the area encircled by the two pieces of side magnetic steel, only the first gap is arranged between the two pieces of side magnetic steel to avoid the connecting arm of the shaped support chip, and the avoiding space is small, so that the side magnetic steel can be of a structure encircling the voice coil, the whole voice coil is almost fully surrounded by a magnetic circuit, and the performance is improved.

The present invention relates to a noise reduction type environmentally friendly Bluetooth headset and a noise reduction method thereof, the noise reduction type environmentally friendly Bluetooth headset comprises a main body shell, an ear hook, an earphone cover plate and an earplug, the earplug has a sponge ring, a silicone ring is embedded in the sponge ring, and the silicone ring is sleeved on the outside of a speaker housing, characterized in that a control panel, a power supply device, a data storage device, a Bluetooth communication device and a Beidou positioning device are arranged on the inner side of the main body shell, an earphone switch button, a volume increase button, a volume decrease button and a positioning button are arranged on the side of the main body shell, a display screen is arranged on one side wall of the main body shell, a noise reduction device and an audio collection device are arranged in the cavity of the earphone cover plate, a speaker is arranged on the inner side of the speaker housing, the noise reduction device, the audio collection device and the speaker are respectively connected to the control panel, and the main body shell, the earphone cover plate and the speaker housing are all made of degradable materials. The Bluetooth headset can not only effectively reduce noise, but also has rich functions and meets the requirements for environmental protection.

The present disclosure relates to an audio recording method, device, electronic equipment, storage medium and program product. The audio recording method comprises: determining at least one target channel satisfying a preset condition during audio recording, the preset condition representing that the audio quality of the audio data obtained by recording through the target channel is higher than a preset audio quality; and recording audio through the at least one target channel to obtain target audio data. The method uses the target channel with audio quality higher than the preset audio quality to record and obtain target audio data, which can obtain high-quality input speech, improve the audio quality of the input speech, especially in the speech-to-text input scene, which can effectively improve the accuracy of speech-to-text, and thus improve the user experience of the speech input function.

The invention discloses a digital conference audio interaction system and method and a storage medium, and particularly relates to the technical field of audio interaction, and the system comprises a conference starting and authorization module which is used for starting a conference and managing the authority of participants; the conference scene virtualization module can be used for setting a conference scene and related elements; the participant identity verification module verifies the identity through face recognition and the like and provides personalized setting; the voice separation and recognition module can separate and extract personal audios when multiple persons speak; the interactive operation matching module is used for matching an operation instruction with a virtual object and processing an unidentified object; the system further comprises an audio processing module for optimizing the audio effect. According to the invention, the defects of the existing digital conference system in the aspects of organization management, scene setting, identity verification, voice processing, interactive operation and the like are overcome, the efficiency, quality and participating experience of the digital conference are improved, and the method is suitable for various business, academic and other conference scenes.

Embodiments of the present application provide an audio processing method and device, electronic equipment and storage medium, belonging to the field of financial technology. The method comprises: generating a first latent variable based on a pre-configured audio encoder according to a first timbre vector, a first content vector and a first pitch vector; performing latent relationship mapping processing on the first latent variable to obtain a second latent variable; performing alignment processing on the first content vector and the obtained second pitch vector to obtain a second content vector corresponding to the to-be-processed audio; performing spectral optimization processing on the initial mel spectrum based on the audio encoder according to the first timbre vector, the second content vector, the second pitch vector and the second latent variable to obtain an optimized mel spectrum. Embodiments of the present application not only improve the sound quality and pitch of the to-be-processed audio, but also optimize the initial mel spectrum based on the improved audio parameters to obtain a beautified mel spectrum, which is conducive to improving the audio quality of sound beautification.

The utility model discloses a kind of ear support of head-mounted intelligent equipment and head-mounted intelligent equipment, the ear support of this head-mounted intelligent equipment includes shell, sound guide cavity, sound outlet, loudspeaker, the loudspeaker is set in the position corresponding to the rear of the pinna of wearer of the shell, the sound outlet is opened in the position corresponding to the front of the pinna of wearer on the shell, the sound guide cavity is located between the loudspeaker and the sound outlet to connect the loudspeaker and the sound outlet, and at least one acoustic wave reflection member is protruded on the inner wall of the sound guide cavity.The ear support of head-mounted intelligent equipment and head-mounted intelligent equipment disclosed by the utility model can effectively improve audio quality, and is applied to head-mounted intelligent equipment such as intelligent glasses.

The present invention relates to the field of data processing, and in particular to a method for dynamic transmission and processing of audio data and an audio playback device. The method comprises the following steps: identifying key transmission nodes in an audio transmission network, marking multiple transmission network nodes; performing multi-dimensional state perception on multiple transmission network nodes, and reconstructing multi-dimensional environmental indicators to build a dynamic twin network environment; obtaining audio data to be transmitted; performing low-frequency filtering optimization processing on the audio data to be transmitted, and performing feature segmentation processing to obtain multiple audio feature segments; performing data stream encapsulation processing on multiple audio feature segments, and performing lossless compression encoding to generate multiple audio lossless compression codes; performing intelligent segmented transmission processing on multiple audio lossless compression codes based on a dynamic twin network environment, and extracting audio segment transmission response data. The present invention achieves efficient audio data transmission and low audio errors.

The invention provides an echo time delay detection and elimination system and method, and the system comprises an initialization and audio playing module which is used for initializing equipment and controlling the equipment to play recorded audio data based on an initialization result; the time delay detection module is used for sending the recorded audio data to the sound mixing equipment based on the playing result, controlling the drive of the audio loudspeaker to play the sound mixing data of the sound mixing equipment based on the sending result, and meanwhile, collecting the playing audio data based on a preset microphone and sending the playing audio data to the sound mixing equipment; sending the played audio data and recorded audio data in the audio mixing equipment to echo cancellation equipment for time delay detection; and the echo cancellation module is used for carrying out parameter configuration on echo cancellation equipment based on the time delay detection result, and controlling the echo cancellation equipment to carry out echo cancellation operation based on the parameter configuration result. The method can cope with complex and changeable practical application environments, so as to improve the accuracy and reliability of echo cancellation, improve the audio quality, and meet the requirements of people for high-quality audio experience.

The invention relates to the technical field of Bluetooth sound boxes, in particular to a Bluetooth sound box data processing method and system and a storage medium. The method comprises the following steps: performing multi-modal data capture on a client to obtain original multi-modal client data; carrying out self-adaptive multi-resolution compression on the original multi-modal client data to obtain compressed multi-modal client data; transmitting the compressed multi-modal client data to the Bluetooth sound box through the client to obtain received compressed multi-modal data; performing multi-modal data decoupling on the received compressed multi-modal data to obtain separated audio data; performing dynamic spectrum reconstruction optimization on the separated audio data to obtain reconstructed audio data; and performing music preference type determination on the user to obtain a user music preference type identifier. According to the invention, the limitation of a traditional Bluetooth sound box in the aspects of personalized service and environmental adaptability is effectively solved.

The invention relates to an audio enhancement method and device and an earphone. The audio enhancement method comprises the following steps: acquiring an air conduction audio signal and a bone conduction audio signal; acquiring a first audio feature of the air conduction audio signal and a second audio feature of the bone conduction audio signal; performing feature extraction on the bone conduction audio signal through a trained feature extraction model to obtain a voiceprint feature of the bone conduction audio signal; inputting the voiceprint feature of the bone conduction audio signal, the first audio feature and the second audio feature into a trained amplitude prediction model to obtain a predicted amplitude; and enhancing the audio based on the predicted amplitude to obtain a target audio signal. According to the invention, the dual model mechanism fuses the advantages of bone conduction and air conduction, so that the obtained target audio signal does not lose the frequency band, the noise reduction is realized, and the audio quality is improved.