27 results about "Quality voice" patented technology

The invention relates to a service compliance dynamic evaluation system based on real-time voice recognition, and belongs to the cross technical field of artificial intelligence and service compliance management. The system comprises a voice signal enhancement acquisition unit, a voice analysis and risk initial judgment unit, a dynamic evaluation iteration unit and a compliance risk disposal unit. The voice signal enhancement acquisition unit processes low-quality voice and generates a voice data set; the voice analysis and risk initial judgment unit is used for transferring voice in real time, extracting a dialogue intention and identifying violation contents and interactive emotions; the dynamic evaluation iteration unit outputs compliance scores and violation details and is in butt joint with a manual labeling platform; and the compliance risk disposal unit generates a risk label, carries out linkage training and appeal review, and forms a compliance risk disposal scheme. According to the invention, full quality inspection of service scenes in multiple industries such as finance, operators and the like is realized, the voice transfer accuracy and compliance rectification rate are improved, the verbal skill violation complaint rate is reduced, and dynamic monitoring and closed-loop disposal of service compliance are achieved.

The invention discloses a voice noise reduction method and device for a multi-person scene, electronic equipment and a storage medium, and relates to the technical field of voice signal processing. The method comprises the following steps: acquiring an audio signal and a video image of a target space scene; determining audio perception information matched with the audio signal and a visual voice activity detection result corresponding to the video image; fusing the audio perception information and the visual voice activity detection result, and determining a target voice sounding object when the voice activity fusion result indicates that the voice activity exists; and positioning a human face corresponding to the target voice sounding object to obtain position change information, updating a pickup direction, controlling beam forming processing, and outputting a denoised target voice signal. According to the scheme provided by the invention, the voice of the current effective speaker can be accurately separated and enhanced from the aliasing audio signals, meanwhile, the interference of other speakers and environmental noise is inhibited, and help is provided for realizing high-quality voice interaction and recognition.

The invention discloses a controllable voice generation method and device based on multi-agent dynamic scheduling, and belongs to the technical field of voice synthesis, and the method comprises the steps: constructing a multi-agent voice generation frame comprising a central scheduling module, an identity agent, an emotion agent and an environment agent; a central scheduling module analyzes a user instruction and outputs a structured task plan to drive each agent to generate primary voice output of identity, emotion and environment dimensions, and a cooperation cost matrix between the agents is constructed according to the primary voice output; the optimal execution path of the multiple agents is solved through an optimization algorithm based on the cost matrix, the agents are executed in a cascading mode based on the optimal execution path, and finally sound mixing is completed and high-quality voice is output. According to the method, the naturalness, the semantic consistency and the overall quality of the synthesized voice in a complex scene can be remarkably improved, and the method is suitable for various man-machine interaction application scenes such as intelligent voice assistants, virtual digital humans, immersive entertainment, barrier-free voice services, personalized content creation and the like.

The invention provides a voice cloning system and method, and the system comprises a data preprocessing module, a feature extraction module, a model training module, and a reasoning generation module, and the data preprocessing module is used for carrying out the human voice accompaniment separation, audio cutting, and automatic marking of an original audio; the feature extraction module is used for extracting a self-encoding feature and a Mel spectrum feature from the audio and pairing the self-encoding feature and the Mel spectrum feature with the text length; the model training module is used for performing fine tuning training on the basis of a small amount of reference audio data by combining a GPT-like model and a VITS and Valle model architecture; and the reasoning generation module is used for generating target audio according to the reference audio and the text provided by the user. According to the method and the device, high-quality voice cloning and text-to-voice conversion can be realized only by a small amount of data sets, so that the user experience is improved, and the development cost is also reduced.

This application provides an earphone case, a wireless communication device, and a method for using the same. The earphone case includes a case body, a cover assembly, and a magnetic component. The cover assembly includes a lid, a microphone module, and a playback module. The lid can be closed and detached from the case body. The playback module is used to play sound. At least a portion of the lid is metal. The magnetic component is detachably mounted to the case body and can fix the lid when the cover assembly is separated from the case body, allowing the cover assembly to function as an independent module for both microphone and playback. In this embodiment, the earphone case not only stores wireless earphones but also allows the cover assembly to function independently as an earphone. During two-way communication, one person wears the wireless earphones while the other wears the cover assembly with the help of the magnetic component, reducing the distance limitation for communication and enabling high-quality voice interaction.

The invention provides an ultra-low time delay signal sequence conversion method and system, which can be widely applied to various time sequence conversion scenes, including hearing enhancement systems and equipment, such as hearing aid, denoising, simultaneous translation, sound conversion signal sequence, intelligent glasses, brain-computer systems and the like. According to the invention, the rapid high-fidelity noise elimination method is adopted, so that the limitation and difficulty in the traditional technology can be overcome. And for the problems of noise uncertainty and voice distortion in the sound related field, the high-quality voice can be recovered by recognizing the voice content as the intermediate language representation with high generalization, and meanwhile, the voice of the same speaker is synthesized by using a pre-trained artificial intelligence module. And the ultra-low time delay is realized by using a variable multi-partition block and a multi-head prediction mechanism in the process. Therefore, in the specific hearing enhancement field, the voice of the target speaker can be focused by using the voice separation technology, and the uncertainty of background noise is avoided.

This utility model relates to the field of audio acquisition equipment technology and discloses a microphone for classrooms, including a microphone body comprising a housing. This microphone for classrooms constructs a highly efficient dual noise reduction system by incorporating a noise-reducing cover and composite sound-absorbing cotton. The noise-reducing cover tightly wraps around the housing, utilizing its material properties to block most external environmental noise, such as noise from outside the window and footsteps outside the classroom. The composite sound-absorbing cotton, filled between the housing and the noise-reducing cover, further absorbs residual noise penetrating the cover while reducing noise generated by vibrations of internal components. This design allows the microphone to effectively filter out non-target sounds in complex classroom environments, significantly improving the clarity of target sounds such as teachers lecturing and students answering questions. It avoids the audio blurring problem caused by noise interference in traditional microphones, providing higher-quality sound materials for classroom recording, remote teaching, and other scenarios.

A high-quality voice signal processing device through removal of ambient noise based on multi-sensor signal fusion, includes: a voice microphone sensor that senses and outputs a speaker's voice signal; an accelerometer sensor that senses vibration of the speaker's vocal cords and outputs a signal; a noise reduction processing MCU that extracts a voice section according to vocal cord vibration using the output signal of the accelerometer sensor, synthesizes a low-frequency component of the accelerometer sensor and a low-frequency component of the voice microphone sensor at different synthesis ratios based on a level of noise extracted from the output signal of the voice microphone sensor using voice section information, and restores and outputs a voice signal by adding the synthesized low-frequency components and a high-frequency component of the voice microphone sensor; and a wireless communication module that externally outputs the restored voice signal.

The present application realizes a signal generation processing device that realizes a voice synthesis processing or an image signal generation processing that can maintain the speed of the voice synthesis processing or the image signal generation and obtain a high-quality voice signal or an image signal. In the signal generation processing device, first to Nth sub-model sections respectively perform learning processing of learning models included in the first to Nth sub-model sections using noise levels included in different noise level ranges, thereby acquiring learned models. That is, in the signal generation processing device, processing can be performed in parallel for each sub-model section, and as a result, the learning processing can be performed at high speed. In addition, in the signal generation processing device, at the time of prediction processing, a sub-model section to be used for processing can be appropriately selected, and thus a high-precision voice synthesis processing or an image generation processing can be performed.

The application relates to the technical field of audio processing technology, in particular to an embedded audio processing system and method supporting AI language enhancement, which comprises a data acquisition module, an audio analysis module, an AI enhancement module, an audio processing module and an interactive output module. The modules of the system cooperate, the data acquisition module collects audio, the analysis module accurately classifies and calibrates, the enhancement module improves the quality of fuzzy audio, the processing module generates corresponding content, the output module outputs, accurate, efficient and high-quality voice interaction is realized, the speed of vehicle-mounted AI voice answering voice questions is improved, and the accuracy of vehicle-mounted AI voice answering voice questions is improved.

This invention relates to a locomotive CIR host and scheduling communication method supporting 5G broadband voice AMR-WB scheduling communication. The CIR host is implemented by replacing the main control unit on an existing CIR host. The CIR host includes a backplane, and the main control unit includes a base plate and a core board. The core board is mounted on the base plate, which is connected to the backplane. The base plate has an Ethernet port, and the CIR host connects to a 5G communication module via the Ethernet interface to achieve 5G broadband voice AMR-WB scheduling communication. Compared with existing technologies, this invention offers advantages such as providing higher quality voice communication for locomotives, low modification costs, convenience, and strong versatility.

The invention discloses a voice coding and decoding method based on principal component analysis and multi-scale depth attention, and relates to the technical field of voice signal processing, and the method comprises the steps: carrying out the multi-scale depth attention convolution operation of a voice signal for many times, and obtaining a first feature; wherein after feature extraction and feature fusion are respectively carried out through a plurality of parallel depth separable convolutions with different depth convolution kernels, channel-by-channel multiplication is carried out on fused features according to channel attention weights; performing projection processing on the first feature according to the projection matrix and the feature mean value to obtain a second feature; wherein principal component analysis is carried out on first features of a training set, a feature mean value is determined, and a previous feature vector with the highest cumulative variance contribution rate is selected to form a projection matrix; performing multi-stage vector quantization on the second feature in sequence to obtain a reconstructed vector; and carrying out inverse projection processing on the reconstructed vector, and then decoding and outputting reconstructed voice. And ultra-low-bit-rate and high-quality voice communication is realized.

The invention provides a vehicle-mounted voice endpoint detection method, device and equipment and a storage medium, and belongs to the technical field of vehicle-mounted voice interaction, and the method comprises the steps: collecting a digital voice signal; performing framing processing and adaptive noise elimination on the digital voice signal, and marking a voice frame and a non-voice frame in a current frame; marking a voice starting endpoint and a voice ending endpoint in the voice frame; dividing the continuous voice frames into one or more voice segments, and adjusting the dynamic mute tolerance time and the voice ending endpoint; updating a voice starting endpoint and a voice ending endpoint in the voice frame based on the non-voice frame; and extracting a corresponding voice segment according to the voice starting endpoint and the voice ending endpoint in the updated voice frame. Through the technical scheme in the embodiment of the invention, the starting endpoint and the ending endpoint of the voice can be accurately and stably detected, and high-quality voice segment input is provided for vehicle-mounted voice recognition.

The present invention relates to a high-quality voice signal processing device and method through ambient noise removal based on multi-sensor signal fusion, which enables voice signal processing robust to external noise environments through multi-sensor signal fusion using an accelerometer sensor (ACC) and a voice microphone sensor (MIC). The device comprises: a voice microphone sensor (MIC) that senses and outputs a voice signal of a speaker; an accelerometer sensor (ACC) that detects vocal cord vibration of a speaker and outputs a signal; a noise reduction processing MCU that extracts a vocalization segment based on vocal cord vibration using the output signal of the accelerometer sensor (ACC), synthesizes the low-frequency component of the accelerometer sensor (ACC) and the low-frequency component of the voice microphone sensor (MIC) by varying the synthesis ratio based on the noise level extracted from the output signal of the voice microphone sensor (MIC) using the vocalization segment information, and restores and outputs a voice signal by adding the synthesized low-frequency component and the high-frequency component of the voice microphone sensor (MIC); and a wireless communication module that outputs the restored voice signal externally.

The invention relates to the field of artificial intelligence, is applied to financial and medical scenes, and discloses a text-to-voice method and device, computer equipment and a storage medium, and the method comprises the steps: receiving a target text, a voice prompt and an emotion label; performing semantic coding on the target text to extract text semantic features; performing acoustic coding on the voice prompt to extract voice acoustic features; based on the emotion label and the text semantic feature, generating a dynamic emotion feature through a time sequence emotion model; performing cross-modal fusion on the voice acoustic features and the dynamic emotion features to obtain cross-modal alignment features, and inputting the cross-modal alignment features and the text semantic features into a pre-training language model for feature fusion and voice token prediction to obtain an initial voice token sequence; and performing semantic alignment and spectrum conversion on the initial voice token sequence to generate a target voice waveform. According to the invention, high-quality voice with natural emotional circulation, controllable timbre and high conformity with professional contexts can be synthesized.

The invention provides a mobile device and a self-noise elimination method thereof, and relates to the technical field of denoising. The method comprises the following steps: collecting motion state data of the mobile device in real time; inputting the motion state data into a noise feature pre-judgment model to obtain noise feature parameters output by the noise feature pre-judgment model; the noise characteristic parameters comprise a target amplitude, a target frequency and a target phase of self-noise generated by the mobile device; and based on the noise characteristic parameter, generating a reverse counteracting signal, and controlling a reverse counteracting component of the mobile device to output the reverse counteracting signal. According to the invention, the self-noise is actively offset from the source through hardware, the interference of complex multi-source self-noise is reduced in a real-time and preposed manner, and a solid foundation is laid for high-quality voice interaction of a mobile device in a severe operation environment, that is, high-efficiency and high-quality self-noise elimination is realized, and high-efficiency and high-quality voice interaction is ensured.

The application discloses a kind of based on AI speech synthesis and animation driving's lecturer video automatic generation method, comprising: by improved inner point method and incremental shortest path algorithm to PPT or text script is structured and analyzed;Application full dynamic parallel single link clustering algorithm carries out semantic grouping and generates enhanced script with expressive mark;High-quality voice data stream is generated by using CosyVoice technology combined with low-rank approximation method;The mapping relationship between content and action expression is established by semantic analysis, and a complete set of action expression instructions is generated;The museTalk technology is used to drive the digital human model, and the final lecturer teaching video is generated by using parallel rendering algorithm.The application realizes the efficient and automatic generation of educational videos, significantly improves the content production efficiency, reduces the production cost, and ensures the professionalism and expressiveness of the teaching video.

An object of a system according to an embodiment is to quickly generate a high-quality voice clone and realize efficient advertisement production.SOLUTION: A system includes a learning unit, a generation unit, and an advertisement creation unit. The learning unit learns voice data. The generation unit generates a voice clone based on the data learned by the learning unit. The advertisement production unit produces an advertisement using the voice clone generated by the generation unit.SELECTED DRAWING: Figure 1

The application provides a mobile device and a self-noise elimination method thereof, and relates to the technical field of noise elimination. The method comprises the following steps: collecting motion state data of the mobile device in real time; inputting the motion state data into a noise characteristic pre-judgment model to obtain noise characteristic parameters output by the noise characteristic pre-judgment model; the noise characteristic parameters comprise a target amplitude, a target frequency and a target phase of self-noise generated by the mobile device; based on the noise characteristic parameters, a reverse cancellation signal is generated, and a reverse cancellation component of the mobile device is controlled to output the reverse cancellation signal. The application actively cancels self-noise from the source by hardware, reduces the interference of complex multi-source self-noise in real time and in advance, lays a solid foundation for high-quality voice interaction of the mobile device in a harsh operating environment, realizes efficient and high-quality self-noise elimination, and ensures efficient and high-quality voice interaction.

An object of a system according to an embodiment is to realize low-cost and high-quality voice conversion in real time.SOLUTION: A system includes an input unit, an analysis unit, a conversion unit, and an output unit. The input unit inputs voice. The analysis unit analyzes the voice input by the input unit. The conversion unit converts the voice based on the parameter generated by the analysis unit. The output unit outputs the voice converted by the conversion unit in real time.SELECTED DRAWING: Figure 1

The invention discloses a multi-user audio inflexion control method and system. The method comprises the following steps: acquiring original audio signals of at least two users; performing AI inflexion processing on the original audio signal of each user to generate an inflexion audio stream corresponding to each user; distributing the inflexion audio stream of each user to different audio channels of the multi-channel audio output interface to form a multi-channel composite audio stream, and outputting the multi-channel composite audio stream to the PC end through the multi-channel audio output interface; and the PC terminal identifies the multi-channel audio and configures a corresponding input source for each inflexion audio stream in the multi-channel audio, so that each inflexion audio is called. Through the scheme of the invention, on the premise that the user does not need to operate the audio setting of the PC end, the real-time independent voice changing of multiple persons is realized, each user can synchronously enjoy the customized and high-quality voice changing effect, the low delay, high compatibility and the flexible expandability of the system architecture are ensured, and the experience and convenience of multi-person online voice interaction are greatly improved.

The application provides a method and system for generating a high-quality voice database, the method comprising: collecting a certain amount of seed sound data; generating target text data using the seed sound data, preparing text that needs to be synthesized; using a sound cloning technology based on an AI large model, training a deep learning algorithm and a large amount of data to achieve high-precision imitation and replication of the voice of a specific speaker, and generating voice data; verifying the quality of the generated voice data, removing unqualified voice data, ensuring the quality of the voice database, and obtaining a high-quality voice database. The application greatly reduces the cost of voice data collection, can quickly generate a large amount of high-quality voice data, significantly shortens the production cycle of the voice database, improves the efficiency of voice data collection, improves the consistency of the generated voice data and the corresponding text, ensures the accuracy and naturalness of the voice data, and improves the overall quality of the voice database.

The invention discloses a TTS speech synthesis method and device based on speech prompt noise reduction, equipment and a medium, and relates to the field of speech synthesis in the fields of finance, medical treatment, insurance, banking and the like, and the method comprises the steps: collecting an original speech prompt which comprises background noise; performing noise suppression processing on the original voice prompt to obtain a preliminary pure voice prompt; performing voice definition enhancement processing on the preliminary pure voice prompt to obtain a high-quality voice prompt; and performing TTS training or reasoning based on the high-quality voice prompt. According to the method, the definition, the truth and the anti-interference capability of the synthesized voice can be remarkably improved, the problem of synthesized voice distortion caused by voice prompt quality is fundamentally solved, and the applicability and the user experience of the system in a real environment are enhanced.

The invention provides electronic equipment, a method for collecting an audio signal and a computer readable storage medium. The electronic equipment comprises a body, a circuit board, keys and a first microphone, the body comprises a first surface and a second surface different from the first surface, the second surface is provided with a first sound receiving hole, the circuit board is arranged in the body, the keys are arranged on the first surface of the body and electrically connected with the circuit board, and the first microphone is arranged on the body and electrically connected with the circuit board. The key responds to the trigger operation and controls the first microphone to collect a first audio signal input from the outside of the body. When the user holds the electronic equipment by hand, the mouth of the user approaches and directly faces the first sound receiving hole to output the voice signal, so that the signal-to-noise ratio of the first audio signal which is acquired by the first microphone through the first sound receiving hole and comprises the voice signal and the environmental noise is higher, and compared with the related technology, the signal-to-noise ratio of the acquired audio signal is improved; the electronic equipment can collect the voice signal output by the user in a noise environment with high quality.

The invention discloses a laser monitoring automatic detection method and system based on mechanical vibration response, and the method comprises the steps: modulating reflected light through the micro-amplitude mechanical vibration of an object driven by voice, and precisely capturing a modulation signal embedded with vibration information through an interference effect and a high-sensitivity photoelectric detector; and noise interference is effectively removed in combination with adaptive filtering and wavelet denoising algorithms, pure vibration features are extracted, and a high-quality voice sequence is efficiently restored based on a Doppler frequency shift model. And an intelligent load sorting replacement mechanism is integrated to carry out priority sorting on key audio segments, and a storage path is automatically calibrated by using a virtual instrument technology, so that optimized storage of high-priority segments is realized, and finally, an efficient and searchable voice storage structure is formed. According to the invention, long-distance, non-contact and high-concealment voice acquisition and intelligent storage are realized, and the distance, definition and subsequent utilization efficiency of voice information acquisition are remarkably improved.

The invention discloses a visual guidance emotional speech synthesis method based on a multi-modal diffusion model, and relates to the crossing field of artificial intelligence and speech synthesis. The method aims to solve the problems of insufficient multi-modal fusion, poor spectrum quality, insufficient style similarity and the like in the existing visual guidance speech synthesis, and the core scheme comprises the steps of constructing a multi-modal fusion mechanism taking phoneme features as a core, and constructing a multi-modal fusion mechanism in a mode of combining cross-modal attention and self-attention; deeply fusing text semantics with identity and emotion features in the figure image; a multi-mode duration predictor is designed, bimodal information is integrated to accurately predict phoneme duration, and time sequence alignment of a text and a voice sequence is achieved; a diffusion model is adopted to carry out spectrum decoding, Mel-spectrogram features are optimized through forward noise adding and reverse denoising processes, and the problem of spectrum over-smoothing is solved. Experimental results show that compared with a traditional TTS Diffusion model, the method can generate voice which is high in naturalness and emotional and accurately matched with visual features, and is suitable for scenes of animation dubbing, virtual human interaction, low-quality voice enhancement and the like.

The invention discloses a high-quality voice communication processing system based on numerical control attenuation and automatic level control, which belongs to the technical field of wireless communication networks, and comprises the following steps: an audio excitation signal generation module used for obtaining an audio excitation signal; the numerical control attenuation adjustment module is used for obtaining an initial adjustment audio excitation signal and transmitting the initial adjustment audio excitation signal to the automatic level control module; the automatic level control module is used for obtaining a fine tuning audio excitation signal and transmitting the fine tuning audio excitation signal to the radio frequency front-end control module; the radio frequency front-end control module is used for obtaining a high-quality voice signal and transmitting the high-quality voice signal to the signal transmitting module; and the signal transmitting module is used for transmitting the high-quality voice signal to a preset position, so that the technical problem of voice signal distortion caused by numerical control attenuation and unstable automatic level control adjustment due to electromagnetic interference and transient amplitude change in the prior art is solved.