Abstract: The accuracy of automatic speech recognition (ASR) tasks is improved using trained models. A speech recognition model is applied in a noisy environment where speech is spoken at a distance from the microphones. The techniques may include extracting speech features, data augmentation by adding feature perturbation, and/or a multi-domain end-to-end speech recognition model. In some implementations, the described technology includes using a teacher-group knowledge distillation strategy to train a deep end-to-end speech recognition model on original speech samples and the sample speech augmentation of the original speech samples, that outputs recognized text transcriptions corresponding to speech detected in the original speech samples and the sample speech augmentation.

Abstract: The application discloses a method of setting a parameter of a multimedia conference. A client may, in response to a multimedia conference setting operation triggered by a user within a first schedule editing interface, obtain a first conference link of the multimedia conference and present a first setting page within the first schedule editing interface, the multimedia conference being a multimedia conference associated with a schedule corresponding to the first schedule editing interface, and the first setting page being for setting a parameter of the multimedia conference, the first schedule editing interface being an editing interface related to a schedule. Then, the client may obtain a first multimedia conference parameter input by the user within the first setting page, and in response to a saving operation triggered by the user within the first setting page, save the first multimedia conference parameter in association with the first conference link.

Abstract: The present disclosure provides a method, an apparatus, an electronic device and a storage medium for processing information. In response to a first predetermined event related to a first object, a first information item related to the first object is fixedly and preferentially displayed in an information flow list of a current user; displaying information related to the first predetermined event in the first information item, wherein the information flow list is used for displaying an information item, and the information item comprises a session item and/or a message item. According to the method of the present disclosure, a first information item is fixedly and preferentially displayed in an information flow list, and thus the experience of reminding a user can be improved.

Abstract: The present disclosure provided a schedule-based information processing method, which may be performed by a client. The client may obtain at least one to-do schedule of a user, and then display, on a first page, reminder information corresponding to a first schedule in the to-do schedule and a quick entry corresponding to the first schedule, where the first page is an instant messaging application interface. The quick entry includes at least one of a chat entry and a multimedia meeting entry, where the chat entry is configured to trigger an operation corresponding to a chat, and the multimedia meeting entry is configured to trigger an operation corresponding to a multimedia meeting.

Abstract: A method, an apparatus, a terminal and a storage medium for display control is provided by the disclosure. The disclosure provides a method of display control which includes: displaying, in a display area of a target application, at least part of pages opened in the target application, if a number of pages opened in the target application exceeds n, displaying n opened pages in the display area of the target application and hiding other pages, and displaying a page identifier of a hidden page, n being not less than 1; and in response to a triggering operation on the page identifier, displaying a page corresponding to the triggered page identifier in the display area of the target application.

Abstract: The embodiments of the present disclosure disclose a schedule sharing method, apparatus, and device, the schedule sharing method includes: determining a schedule to be shared after detecting a trigger operation of a first user; and sending the schedule to be shared to a target object based on a share operation triggered the first user. The target object is a page or webpage accessed by a user having an access authority. When a second user views the schedule information of the schedule to be shared through the target object, the authority information of the second user with respect to the schedule to be shared is obtained, so that an operation control matching authority information of the second user is displayed on the page corresponding to the schedule to be shared, thereby allowing the second user to interact with the schedule to be shared through the operation control.

Abstract: The accuracy of automatic speech recognition (ASR) tasks is improved using trained models. A speech recognition model is applied in a noisy environment where speech is spoken at a distance from the microphones. The techniques may include extracting speech features, data augmentation by adding feature perturbation, and/or a multi-domain end-to-end speech recognition model. In some implementations, the described technology includes using a teacher-group knowledge distillation strategy to train a deep end-to-end speech recognition model on original speech samples and the sample speech augmentation of the original speech samples, that outputs recognized text transcriptions corresponding to speech detected in the original speech samples and the sample speech augmentation.

Abstract: A method for detecting SQL injection interception is provided. The method includes: detecting a received SQL instruction according to a SQL syntax tree rule to determine that the received SQL instruction is a malicious instruction; and analyzing the received SQL instruction by using an analysis model to determine that the received SQL instruction is a potentially malicious instruction, in a case that the received SQL instruction is not determined to be a malicious instruction according to the SQL syntax tree rule. The method can combine the online detection and the offline analysis to intercept the invading malicious SQL commands, and can also find and supplement the loopholes of the rules through offline analysis.

Abstract: A method for detecting SQL injection interception is provided. The method includes: detecting a received SQL instruction according to a SQL syntax tree rule to determine that the received SQL instruction is a malicious instruction; and analyzing the received SQL instruction by using an analysis model to determine that the received SQL instruction is a potentially malicious instruction, in a case that the received SQL instruction is not determined to be a malicious instruction according to the SQL syntax tree rule. The method can combine the online detection and the offline analysis to intercept the invading malicious SQL commands, and can also find and supplement the loopholes of the rules through offline analysis.

Abstract: A photovoltaic intelligent power supply, comprising a plurality of unit modules, and a communication unit (103) and a control unit (106), wherein all the unit modules are connected to the control unit (106) and the communication unit (103); each unit module comprises an input collection unit (101), a data acquisition unit (102), a boost unit (104), an arc isolation unit (105) and an anti-PID unit (107), wherein the input collection unit (101) is connected to a photovoltaic module; the data acquisition unit (102) is configured to acquire voltage and current state signals; the boost unit (104) is configured to perform interleaving chopping and operate in an MPPT mode; the arc isolation unit (105) is configured to receive instructions sent by the control unit (106) to execute opening and closing; and the anti-PID unit (107) is configured to receive instructions sent by the control unit (106) so as to generate proper DC voltages to be applied between a negative electrode of a cell panel and the ground.

Abstract: A photovoltaic intelligent power supply, comprising a plurality of unit modules, and a communication unit (103) and a control unit (106), wherein all the unit modules are connected to the control unit (106) and the communication unit (103); each unit module comprises an input collection unit (101), a data acquisition unit (102), a boost unit (104), an arc isolation unit (105) and an anti-PID unit (107), wherein the input collection unit (101) is connected to a photovoltaic module; the data acquisition unit (102) is configured to acquire voltage and current state signals; the boost unit (104) is configured to perform interleaving chopping and operate in an MPPT mode; the arc isolation unit (105) is configured to receive instructions sent by the control unit (106) to execute opening and closing; and the anti-PID unit (107) is configured to receive instructions sent by the control unit (106) so as to generate proper DC voltages to be applied between a negative electrode of a cell panel and the ground.