Abstract: A method of training an audio separation network is provided. The method includes obtaining a first separation sample set, the first separation sample set including at least two types of audio with dummy labels, obtaining a first sample set by performing interpolation on the first separation sample set based on perturbation data, obtaining a second separation sample set by separating the first sample set using an unsupervised network, determining losses of second separation samples in the second separation sample set, and adjusting network parameters of the unsupervised network based on the losses of the second separation samples, such that a first loss of a first separation result outputted by an adjusted unsupervised network meets a convergence condition.

Abstract: Bone marrow, the primary hematopoietic organ, is more sensitive than other organs when exposed to ionizing radiation or chemotherapeutic drugs due to the fragile hematopoietic stem and progenitor cells (HSPCs). Described herein are novel compounds, compositions, and methods comprising protective drug-based conjugates to achieve the specific protection of HSPCs in the bone marrow.

Abstract: This application provides a method for training a speech synthesis model, a speech synthesis method, an apparatus, an electronic device, a computer-readable storage medium, and a computer program product. The method for training a speech synthesis model includes: obtaining a text sample and a standard speech corresponding to the text sample; performing speech bit stream prediction on the text sample by using the speech synthesis model, to obtain a speech bit stream corresponding to the text sample; decoding the speech bit stream by using the speech synthesis model, to obtain a synthesized speech corresponding to the text sample; and updating a model parameter of the speech synthesis model based on a difference between the synthesized speech and the standard speech.

Abstract: A method of training an audio separation network is provided. The method includes obtaining a first separation sample set, the first separation sample set including at least two types of audio with dummy labels, obtaining a first sample set by performing interpolation on the first separation sample set based on perturbation data, obtaining a second separation sample set by separating the first sample set using an unsupervised network, determining losses of second separation samples in the second separation sample set, and adjusting network parameters of the unsupervised network based on the losses of the second separation samples, such that a first loss of a first separation result outputted by an adjusted unsupervised network meets a convergence condition.

Abstract: This application provides a blood pressure measurement method of a wearable device, and a wearable device. The method provided includes: emitting an optical signal through a first region of a display screen; converting, into a first PPG signal, the optical signal received by the PD; emitting the optical signal through a second region of the display screen if the first PPG signal does not satisfy a signal quality requirement, where the second region and the first region are different in at least one of the following features: a position, luminance, an area size, or a shape on the display screen; converting, into a second PPG signal, the optical signal received by the PD; obtaining a pressure value collected by the PT when the second PPG signal satisfies the signal quality requirement; and determining a blood pressure value of a wearer based on the pressure value and the second PPG signal.

Abstract: Embodiments of the present application relates to a chip heat dissipation structure, a chip structure, a circuit board, and a supercomputing device, and the chip heat dissipation structure includes: a plating layer covering a wafer of the chip; where the plating layer includes a first metal layer, a second metal layer, and a third metal layer sequentially arranged. Three metal layers are added on a top of the chip by physical sputtering, so that a heat sink can be welded on the metal layers by a solder layer, and then the heat sink is fixed on the top of the chip; a main component of the solder layer is metal tin, and the metal layer have a higher thermal conductivity than an epoxy adhesive material mounted on a conventional heat sink.

Abstract: A data processing method based on simultaneous interpretation, applied to a server in a simultaneous interpretation system, including: obtaining audio transmitted by a simultaneous interpretation device; processing the audio by using a simultaneous interpretation model to obtain an initial text; transmitting the initial text to a user terminal; receiving a modified text fed back by the user terminal, the modified text being obtained after the user terminal modifies the initial text; and updating the simultaneous interpretation model according to the initial text and the modified text.

Abstract: An audio processing method and apparatus, including decomposing an audio signal into a low-frequency subband signal and a high-frequency subband signal, obtaining a low-frequency feature of the low-frequency subband signal, obtaining a high-frequency feature of the high-frequency subband signal, feature dimensionality of the high-frequency feature being lower than feature dimensionality of the low-frequency feature, performing quantization encoding on the low-frequency feature to obtain a low-frequency bitstream of the audio signal, and performing quantization encoding on the high-frequency feature to obtain a high-frequency bitstream of the audio signal.

Abstract: This application discloses a multi-sound area-based speech detection method and related apparatus, and a storage medium, which is applied to the field of artificial intelligence. The method includes: obtaining sound area information corresponding to N sound areas including multiple users speaking simultaneously; generating a control signal corresponding to each target detection sound area according to user information corresponding to the target detection sound area; processing multi-user speech input signals by using the control signals, to obtain a speech output signal corresponding to each target detection sound area; generating a speech detection result of the target detection sound area according to the speech output signal corresponding to the target detection sound area; and selecting, among the multiple users, a main speaker based on the user information, the speech output signals and speech detection results of multiple users in the N sound areas.

Abstract: A method for speech recognition based on language adaptivity comprises obtaining voice data of a user. The method also comprises extracting, based on the obtained voice data, a phoneme feature representing pronunciation phoneme information. The phoneme feature is input to a pre-trained language discrimination model that is pre-trained based on a multilingual corpus. A language discrimination result corresponding to the phoneme feature and in accordance with the language discrimination model is obtained. The method also comprises obtaining a speech recognition result of the voice data based on a language acoustic model of a language corresponding to the language discrimination result. The method further comprises determining a speech recognition result of the voice data based on a language acoustic model of a language corresponding to the language discrimination result.

Abstract: This application relates to a speech synthesis method and apparatus, a model training method and apparatus, and a computer device. The method includes: obtaining to-be-processed linguistic data; encoding the linguistic data, to obtain encoded linguistic data; obtaining an embedded vector for speech feature conversion, the embedded vector being generated according to a residual between synthesized reference speech data and reference speech data that correspond to the same reference linguistic data; and decoding the encoded linguistic data according to the embedded vector, to obtain target synthesized speech data on which the speech feature conversion is performed. The solution provided in this application can prevent quality of a synthesized speech from being affected by a semantic feature in a mel-frequency cepstrum.

Abstract: Embodiments of this application disclose a speech-driven animation method and apparatus based on artificial intelligence (AI). The method includes obtaining a first speech, the first speech comprising a plurality of speech frames; determining linguistics information corresponding to a speech frame in the first speech, the linguistics information being used for identifying a distribution possibility that the speech frame in the first speech pertains to phonemes; determining an expression parameter corresponding to the speech frame in the first speech according to the linguistics information; and enabling, according to the expression parameter, an animation character to make an expression corresponding to the first speech.

Abstract: A mixed speech recognition method, a mixed speech recognition apparatus, and a computer-readable storage medium are provided. The mixed speech recognition method includes: monitoring an input of speech input and detecting an enrollment speech and a mixed speech; acquiring speech features of a target speaker based on the enrollment speech; and determining speech belonging to the target speaker in the mixed speech based on the speech features of the target speaker. The enrollment speech includes preset speech information, and the mixed speech is non-enrollment speech inputted after the enrollment speech.

Abstract: Methods and apparatuses are provided for performing sequence to sequence (Seq2Seq) speech recognition training performed by at least one processor. The method includes acquiring a training set comprising a plurality of pairs of input data and target data corresponding to the input data, encoding the input data into a sequence of hidden states, performing a connectionist temporal classification (CTC) model training based on the sequence of hidden states, performing an attention model training based on the sequence of hidden states, and decoding the sequence of hidden states to generate target labels by independently performing the CTC model training and the attention model training.

Abstract: A speech separation model training method and apparatus, a computer-readable storage medium, and a computer device are provided, the method including: obtaining first audio and second audio, the first audio including target audio and having corresponding labeled audio, and the second audio including noise audio.

Abstract: This application relates to a method of extracting an inter channel feature from a multi-channel multi-sound source mixed audio signal performed at a computing device.

Abstract: A neural network training method is provided. The method includes obtaining an audio data stream, performing, for different audio data of each time frame in the audio data stream, feature extraction in each layer of a neural network, to obtain a depth feature outputted by a corresponding time frame, fusing, for a given label in labeling data, an inter-class confusion measurement index and an intra-class distance penalty value relative to the given label in a set loss function for the audio data stream through the depth feature, and updating a parameter in the neural network by using a loss function value obtained through fusion.

Abstract: A far-field pickup device including a device body and a microphone pickup unit is provided. The microphone pickup unit is configured to collect user speech and an echo of a first sound signal output by the device body, and transmit, to the device body, a signal obtained through digital conversion of the collected user speech and the echo. The device body includes a signal playback source, a synchronizing signal generator, a horn, a delay determining unit, and an echo cancellation unit configured to perform echo cancellation on the signal transmitted by the microphone pickup unit to obtain a collected human voice signal.

Abstract: This application discloses an artificial intelligence-based (AI-based) wakeup word detection method performed by a computing device. The method includes: constructing, by using a preset pronunciation dictionary, at least one syllable combination sequence for self-defined wakeup word text inputted by a user; obtaining to-be-recognized speech data, and extracting speech features of speech frames in the speech data; inputting the speech features into a pre-constructed deep neural network (DNN) model, to output posterior probability vectors of the speech features corresponding to syllable identifiers; determine a target probability vector from the posterior probability vectors according to the syllable combination sequence; and calculate a confidence according to the target probability vector, and determine that the speech frames include the wakeup word text when the confidence is greater than or equal to a threshold.