WO2021135652A1 - 信息处理方法、信息控制中心设备及计算机可读存储介质 - Google Patents

信息处理方法、信息控制中心设备及计算机可读存储介质 Download PDF

Info

Publication number
WO2021135652A1
WO2021135652A1 PCT/CN2020/127639 CN2020127639W WO2021135652A1 WO 2021135652 A1 WO2021135652 A1 WO 2021135652A1 CN 2020127639 W CN2020127639 W CN 2020127639W WO 2021135652 A1 WO2021135652 A1 WO 2021135652A1
Authority
WO
WIPO (PCT)
Prior art keywords
time
specified
date
tentative
specified time
Prior art date
Application number
PCT/CN2020/127639
Other languages
English (en)
French (fr)
Inventor
林永楷
樊帅
杨鹏
徐瑞婷
Original Assignee
思必驰科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 思必驰科技股份有限公司 filed Critical 思必驰科技股份有限公司
Priority to EP20909397.0A priority Critical patent/EP4086895A4/en
Priority to JP2022540600A priority patent/JP2023509651A/ja
Priority to US17/758,051 priority patent/US20230032792A1/en
Publication of WO2021135652A1 publication Critical patent/WO2021135652A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/08—Speech classification or search
    • G10L15/18—Speech classification or search using natural language modelling
    • G10L15/1822—Parsing for meaning understanding
    • G—PHYSICS
    • G04—HOROLOGY
    • G04G—ELECTRONIC TIME-PIECES
    • G04G13/00—Producing acoustic time signals
    • G04G13/02—Producing acoustic time signals at preselected times, e.g. alarm clocks
    • G—PHYSICS
    • G04—HOROLOGY
    • G04G—ELECTRONIC TIME-PIECES
    • G04G13/00—Producing acoustic time signals
    • G04G13/02—Producing acoustic time signals at preselected times, e.g. alarm clocks
    • G04G13/021—Details
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/08—Speech classification or search
    • G10L15/18—Speech classification or search using natural language modelling
    • G10L15/1815—Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/28—Constructional details of speech recognition systems
    • G10L15/34—Adaptation of a single recogniser for parallel processing, e.g. by use of multiple processors or cloud computing
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
    • G10L2015/223—Execution procedure of a spoken command
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network

Definitions

  • the present invention relates to the technical field of speech processing, in particular to an information processing method, an information control center device and a computer-readable storage medium.
  • the smart alarm clock can realize the setting of the alarm clock task through voice interaction.
  • the client is difficult to understand the time set by the user's real intention.
  • the user needs to accurately provide the date and time of the time before the client can set the corresponding The task affects the user experience.
  • the embodiments of the present invention provide an information processing method, an information control center device, and a computer-readable storage medium, which can perform time inference on the time in the sound signal and determine the intended time that meets the demand.
  • One aspect of the present invention provides an information processing method, which is applied to an information control center device, and the method includes: obtaining semantic analysis information corresponding to a sound signal, where the semantic analysis information includes a specified time; Time estimation is performed at a specified time, and the intent time is determined; a target instruction corresponding to the sound signal is generated based on the intent time.
  • the time estimation of the specified time based on the current time to determine the intended time includes: judging whether the specified time includes the specified time, and obtaining a first judgment result; when the first When the judgment result judges that the specified time includes the specified time, judge whether the specified time includes the specified date, and obtain the second judgment result; when the second judgment result judges that the specified time includes the specified date, judge the If the specified date is later than the current date, a third judgment result is obtained; when the third judgment result judges that the specified date is later than the current date, the specified time and the specified date are determined as the intended time.
  • the method further includes: when the second judgment result determines that the specified time does not include the specified date, determining the current date as a tentative date; and combining the tentative date and the specified date.
  • the designated time is determined to be a tentative time; it is judged whether the tentative time is not earlier than the current time, and a fourth judgment result is obtained; when the fourth judgment result judges that the tentative time is not earlier than the current time, all The tentative time is determined as the intention time.
  • the method further includes: when the fourth judgment result determines that the tentative time is earlier than the current time, correcting the tentative time based on the principle of time proximity to obtain the corrected time; The correction time is determined as the intended time.
  • the time proximity principle includes at least one of the following principles: a first principle for correcting a provisional time, a second principle for correcting a provisional date, and a third principle for correcting a provisional time.
  • the method further includes: judging whether the specified time includes a specified time period, and obtaining a fifth judgment result; when the fifth judgment result determines that the specified time includes a specified time period, based on a time conversion rule Type conversion at the specified time to obtain the conversion time; the conversion time is used to determine the tentative time, the conversion time does not include the specified time period, and the conversion time and the specified time are used to represent the same time .
  • the time conversion rule includes at least one of the following: a first conversion rule for converting a time type, a second conversion rule for correcting verbal errors, and a third conversion rule for processing critical points of time .
  • the method before judging whether the specified time includes the specified time, the method further includes: obtaining the specified time based on the semantic analysis information; verifying whether the specified time conforms to the law of time, and obtaining the verification result; The verification result verifies that the specified time complies with the time law, and it is determined whether the specified time includes the specified time.
  • an information control center device includes: an obtaining module for obtaining semantic analysis information of a corresponding sound signal, where the semantic analysis information includes a specified time; The designated time performs time estimation to determine the intent time; the generating module is used to generate a target instruction corresponding to the sound signal based on the intent time.
  • the estimation module includes: a first judging sub-module for judging whether the specified time includes a specified time and obtaining the first judgment result; and a second judging sub-module for when the first judging result is obtained.
  • a judgment result judges that the designated time includes a designated time, judge whether the designated time includes a designated date, and obtain a second judgment result;
  • a third judgment sub-module is used to judge whether the second judgment result is the designated time
  • the time includes the specified date, it is determined whether the specified date is later than the current date, and the third judgment result is obtained; the determining sub-module is used for determining whether the specified date is later than the current date by the third judgment result.
  • the specified time and date are determined as the intended time.
  • the determining submodule is further configured to determine the current date as a tentative date when the second judgment result determines that the specified time does not include the specified date; the determining submodule The module is also used to determine the tentative date and the designated time as a tentative time; the inference module further includes: a fourth judging sub-module, which is used to judge whether the tentative time is not earlier than the current time , Obtain a fourth judgment result; the determination sub-module is further configured to determine the tentative time as the intended time when the fourth judgment result determines that the tentative time is not earlier than the current time.
  • the estimation module further includes: a correction sub-module, which is used to determine the tentative time based on the principle of time proximity when the fourth judgment result determines that the tentative time is earlier than the current time. The time is corrected to obtain the correction time; the determining module is also used to determine the correction time as the intended time.
  • a correction sub-module which is used to determine the tentative time based on the principle of time proximity when the fourth judgment result determines that the tentative time is earlier than the current time. The time is corrected to obtain the correction time; the determining module is also used to determine the correction time as the intended time.
  • the speculation module further includes: a fifth judgment sub-module for judging whether the specified time includes a specified time period, and obtaining a fifth judgment result; a conversion sub-module, for when the fifth judgment The judgment result judges that the specified time includes the specified time period, and the specified time is type-converted based on the time conversion rule to obtain the conversion time; the conversion time is used to determine the tentative time, and the conversion time does not include the specified time period , And the conversion time and the designated time are used to represent the same time.
  • the obtaining module is further configured to obtain a specified time based on semantic analysis information; the device further includes: a verification module configured to verify whether the specified time complies with the law of time and obtain a verification result; The verification result verifies that the designated time conforms to the time law, and it is determined whether the designated time includes the designated time.
  • the storage medium includes a set of computer-executable instructions, which are used to execute the information processing method described in any one of the foregoing when the instructions are executed.
  • the information processing method, information control center device, and computer-readable storage medium provided by the embodiments of the present invention.
  • the information control center device uses the information processing method provided by the embodiments of the present invention to process complex and diversified sound signals, and semantic analysis information can be Extracting the specified time from the sound signal, and using the specified time to estimate the time of the semantic analysis information, the specified time provided in the sound signal can be processed more accurately, and the voice interaction process is more accurate.
  • FIG. 1 is a schematic diagram of the implementation process of an information processing method according to an embodiment of the present invention
  • FIG. 2 is a schematic diagram of the implementation process of time inference in an information processing method according to an embodiment of the present invention.
  • FIG. 3 is a schematic diagram of the implementation process of time conversion of an information processing method according to an embodiment of the present invention.
  • FIG. 4 is a schematic diagram of the implementation process of verification of the time law of an information processing method according to an embodiment of the present invention.
  • FIG. 5 is a schematic diagram of a scenario in which an information processing method is applied according to an embodiment of the present invention.
  • FIG. 6 is a schematic flow chart of time inference in a scenario where an information processing method is applied according to an embodiment of the present invention
  • FIG. 7 is a schematic diagram of an implementation module of an information control center device according to an embodiment of the present invention.
  • FIG. 1 is a schematic diagram of the implementation process of an information processing method according to an embodiment of the present invention.
  • an embodiment of the present invention provides an information processing method on the one hand, which is applied to an information control center device.
  • the information control center device may be a cloud server.
  • the method includes: step 101, obtaining a corresponding sound signal Semantic analysis information, the semantic analysis information includes a designated time; step 102, based on the current time, speculate on the designated time to determine the intent time; step 103, generate a target instruction corresponding to the sound signal based on the intent time.
  • the information processing method provided in this embodiment is applied to the information control center equipment and can process complex and diversified sound signals.
  • the sound signals undergo speech recognition and semantic analysis.
  • the semantic analysis information obtained includes designated time and designated tasks.
  • the task is the corresponding target intent in the sound signal;
  • the specified time is the execution time of the corresponding target intent in the sound signal.
  • the information control center equipment can set target instructions based on the intent time and instruct the client to execute the user's instructions at the intent time. Assign tasks to make the results of voice interaction more accurate.
  • the information control center device is selected as a device with data processing capabilities.
  • the information control center device is selected as a cloud server.
  • the cloud server performs voice processing on sound signals, which can greatly reduce the requirements for client hardware, and There is no need to perform complex semantic rule customization on the client, and there is no need to update the client when the cloud server performs semantic upgrades.
  • the client is a terminal that performs signal transmission with an information control center device (for example, a cloud server), and the client has a corresponding function of executing target instructions, that is, the target instructions can be set according to the functions of the client, and the client is selected as the alarm clock .
  • the method includes obtaining semantic analysis information corresponding to the sound signal, and the semantic analysis information includes a specified time.
  • the sound signal can be collected by an audio collecting device.
  • the audio collection device for collecting sound signals selects a microphone array composed of a certain number of microphones, and the microphone array is installed on the client.
  • the semantic analysis information is the information obtained after the voice signal is processed by speech recognition and semantic analysis.
  • the microphone array collects sound signals through a signal processing algorithm, can identify the direction of the sound source, and can also remove background sounds to a certain extent, thereby improving the accuracy of subsequent speech recognition.
  • the voice signal is transmitted to the information control center equipment through the network.
  • the information control center equipment uses ASR voice recognition technology to perform voice recognition on the voice signal.
  • the ASR voice recognition technology can convert the acquired voice signal into text information corresponding to the voice.
  • the ASR speech recognition technology is trained based on the acoustic model and language model in the home environment, which can be more adapted to the home scene, and can also accurately recognize the sound signal in the noisy scene, and obtain accurate text information.
  • the text information is semantically analyzed through the semantic analysis module, and the semantic analysis module can parse the text information into semantic analysis information.
  • the method also includes that the information control center equipment estimates the time based on the current time to determine the intended time.
  • the semantic analysis module sends the semantic analysis information to the dialog management system, and the dialog management system is used to speculate the semantic analysis information based on the specified time to obtain the intention time. Not directly returning the semantic analysis information to the client can avoid the client from having to initiate multiple requests to the information control center device, thereby improving the performance and response time of the client.
  • the intent time used to generate the target instruction needs to include date information and time information.
  • the time information since the time information has a 12-hour system and a 24-hour system, the time information also includes time period information and hour information.
  • the method further includes that the information control center device generates a target instruction corresponding to the sound signal based on the intent time. According to the designated task and intent time, the target instruction corresponding to the sound signal can be generated.
  • the target instruction is used to instruct the execution of the specified task in the sound signal. For example, when the sound signal is "7: 00 remind me to buy a train ticket”, in the semantic analysis information obtained by the information control center equipment according to the semantic analysis, the specified time is "7:00”, the task object is "reminder", and the reminder event is "Buy a train ticket”, infer the specified time as "7:00”, based on the principle of time proximity, determine the current time as "December 26, 2019, 8 o'clock", then determine the intent time as "December 2019” On the 27th, 7:00".
  • the task object is "reminder"
  • the reminder event is the target instruction of "buy train ticket”
  • the object of receiving the target instruction is not limited to the client.
  • the receiving object of the target instruction is the information control center device; when the designated task of the target instruction is the right
  • the receiving object of the target instruction may also be the third-party terminal.
  • the method further includes that the information control center device generates task instructions according to the intent time and the designated task, performs speech synthesis processing according to the intent time, and obtains the audio signal corresponding to the designated task; determines the task instruction and the audio signal as the target instruction; The target instruction is sent to the client, so that the client executes the task instruction and broadcasts the audio signal.
  • the intent time and the specified task are processed through the dialogue management module to obtain the target instruction including the audio signal and the task instruction, and then the target instruction is sent to the client, and the client executes the task instruction by parsing the target instruction And broadcast the audio signal to form a voice interaction with the user.
  • the dialogue management system will return different audio response information through speech synthesis technology according to the state of the dialogue.
  • This article reply uses speech synthesis technology to synthesize audio signals Sent to the client, since the alarm clock has been set successfully, the dialog management system will also include the status of the end of the dialog in the returned data.
  • FIG. 2 is a schematic diagram of the implementation process of time inference in an information processing method according to an embodiment of the present invention. This method is applied to information control center equipment.
  • step 102 based on the current time of the designated time to estimate the time to determine the intended time, including: step 1021, determine whether the designated time includes the designated time, and obtain the first judgment result; step 1022, When the first judgment result judges that the specified time includes the specified time, judge whether the specified time includes the specified date, and obtain the second judgment result; step 1023, when the second judgment result judges that the specified time includes the specified date, judge whether the specified date is late On the current date, the third judgment result is obtained; in step 1024, when the third judgment result judges that the specified date is later than the current date, the specified time and the specified date are determined as the intended time.
  • the specified time obtained through semantic analysis may include the following two cases. In one case, the specified time exists in the analytical information; in the other case, the specified time does not exist in the analytical information.
  • the device cannot generate the target instruction with the intent time and the specified task. It needs to conduct multiple rounds of dialogue interaction through speech synthesis technology to obtain the specified time.
  • the information control center device may generate an inquiry instruction and send it to the client.
  • the inquiry instruction is generated by voice synthesis technology and broadcasted through the client to inquire about the designated time.
  • the supplementary time is used to supplement the specified time.
  • the supplementary time can be used Determined to be the designated time.
  • the dialogue management system will generate a text reply "OK, you want to set a few points "Alarm Clock” uses speech synthesis technology to convert the text into audio and send it to the client.
  • the message that the conversation status has not ended is also returned to the client.
  • the client will collect the user’s voice through the microphone array after playing this audio.
  • the sound information determines the specified time of the alarm clock.
  • the specified time may include the specified date and the specified time.
  • the device cannot determine the specified time to perform the specified task. In order to distinguish the above situations, it is necessary to determine whether the specified time includes the specified time.
  • the device needs to perform multiple rounds of dialogue interaction through the speech synthesis technology to obtain the specified time including the specified time.
  • the first judgment result judges that there is a specified time in the semantic analysis information, it is further judged whether the specified time includes the specified date, so as to obtain the second judgment result.
  • the second judgment result is that the specified time includes the specified date; in another case, the specified time does not include the specified date.
  • the third judgment result judges that the specified date is later than the current date, the specified time and the specified date are determined as the intended time.
  • the third judgment result judges that the specified date is not later than the current date, and the specified date is earlier than the current date, the specified time is earlier than the current date, and the device cannot instruct to perform the specified task in the past time, so it can be determined that the semantic analysis information is invalid , Or generate speech through speech synthesis technology, through multiple rounds of dialogue interaction, to get the specified time again.
  • the third judgment result judges that the specified date is not later than the current date, and the specified period and the current date are the same day, it is necessary to further compare the specified time for inference, as shown in step 10211, when the third judgment result is not later On the current date and the same as the current date, the specified date will be determined as a tentative date.
  • the method further includes: step 1025, when the second judgment result determines that the specified time does not include the specified date, determining the current date as the tentative date; step 1026, determining the tentative date and the specified time as Tentative time; step 1027, judge whether the tentative time is not earlier than the current time, and obtain the fourth judgment result; step 1028, when the fourth judgment result judges that the tentative time is not earlier than the current time, determine the tentative time as Intent time.
  • the designated date is the same, that is, the designated period and the current date are the same day
  • the same method can be used for comparison and judgment in the following. That is, when the specified period and the current date are the same day, the current date is determined as the tentative date, and the tentative date and the specified time are determined as the tentative time, and it is judged whether the tentative time is not earlier than the current time, and the fourth critical result. Determine the intent time according to the fourth judgment result. When the fourth judgment result judges that the tentative time is not earlier than the current time.
  • the device can correspond to the target instruction of the tentative time, which is used to instruct the execution of the specified task at the tentative time, and the tentative time can be determined as the intended time.
  • the target instruction can also be generated and instructed to perform the specified task at the intended time, that is, the intended time is the same as the current time.
  • the method further includes: step 1029, when the fourth judgment result judges that the tentative time is earlier than the current moment, correct the tentative time based on the principle of time proximity to obtain the corrected time; step 10210, correct the tentative time The time is determined as the intent time.
  • the fourth judgment result determines that the tentative time is earlier than the current time
  • the tentative time is still determined as the intent time
  • the intent time will be earlier than the current time, and the device cannot instruct the client to perform the specified task at the intent time.
  • the device needs to correct the tentative time so that the corrected time is no earlier than the current time, so that the device can instruct the client to perform the specified task at the intended time.
  • the principle of time proximity is used to determine the intended time as the future time that is closest to the current time and meets the specified time description.
  • FIG. 3 is a schematic diagram of the implementation process of time conversion of an information processing method according to an embodiment of the present invention. This method is applied to information control center equipment.
  • the method further includes: step 301, judging whether the specified time includes the specified time period, and obtaining a fifth judgment result; step 302, when the fifth judgment result judges that the specified time includes the specified time period, based on the time
  • the conversion rule performs type conversion at the specified time to obtain the conversion time; step 303, the conversion time is used to determine the tentative time, the conversion time does not include the specified time period, and the conversion time and the specified time are used to represent the same time.
  • the designated time obtained from the sound signal has two types: a 24-hour system and a 12-hour system. And different people have different ways of understanding the time period. For example, at 0:00 in the evening, some people may express it as 0:00 the next day, some may express it as 24:00, and some people may express it at 12:30. 00:30. In order to facilitate the comparison of the tentative time, the designated time needs to be converted to the 24-hour system. Similarly, the current time obtained is also selected as the 24-hour system. It should be understood that this step can be performed when judging whether the designated time is not earlier than any previous step at the current time. What needs to be added further is that the current time can be the current time collected by the information control center equipment from the client, or it can be the time zone standard time collected by the information control center equipment from the network.
  • the 12-hour format usually includes time period and hour, such as "7:00 in the morning", where the designated time period refers to words such as "morning" and "afternoon” for expressing time periods.
  • time period such as "7:00 in the morning”
  • the designated time period refers to words such as "morning” and "afternoon” for expressing time periods.
  • the conversion rules specifically include: a first conversion rule for turning a designated time into 24 hours, a second conversion rule for correcting verbal errors, and a third conversion rule for processing date critical points. What needs to be added is that, for the convenience of comparison, after the 24-hour system conversion, the time period information is still retained, and the time period information is adjusted according to the hour information.
  • the first conversion rule used to turn the specified time into 24 hours can be:
  • the specified hour is between 1:00 and 12:00, and the specified time period is noon, afternoon, or evening and their synonyms, the specified hour will be added to 12 hours to obtain the conversion time of the 24-hour system. For example, “afternoon, 7:00” is converted to "afternoon, 19:00".
  • the specified hour is between 1:00 and sunrise time (for example, 6:00), and the specified time period is daytime and its synonyms, then the specified hour plus 12 hours will get the conversion time of the 24-hour system.
  • the sunrise time can be Make dynamic adjustments according to different areas. For example, "daytime, 3:00" is converted to "afternoon, 15:00".
  • the second conversion rule for correcting verbal errors can be:
  • the specified hour is between 0:00 and sunrise time, and the specified time period is night and its synonyms, the specified time period is reset to early morning, the specified hour is determined as the conversion hour, and the exception flag is set, and the subsequent passing time is near
  • the principle is re-determined. For example, "evening, 1:00" is converted to "early morning, 1:00, abnormal”.
  • the specified hour is greater than 12:00, and the specified time period is morning, morning, or early morning and its synonyms, the specified time period needs to be reset, the specified hour is determined as the conversion hour, and the abnormal flag is set to avoid verbal errors or misidentification. influences. For example: “AM, 15:00” is converted to "PM, 15:00, abnormal".
  • the third conversion rule for processing the critical point at the time can be:
  • the time proximity principle includes at least one of the following principles: a first principle for correcting a provisional time, a second principle for correcting a provisional date, and a third principle for correcting a provisional time.
  • the provisional time includes three parts: provisional time, provisional date and provisional hour.
  • the correction time obtained by the tentative time correction also includes three parts: the correction time, the correction date, and the correction hour.
  • the first principle for correcting the tentative time can be:
  • the tentative hour does not include the tentative time period, and the tentative hour is less than 12:00
  • the tentative hour is increased by 12 hours to obtain the preset time; to determine whether the preset time is not earlier than the current time, when the preset time Not earlier than the current time, the preset time is determined as the correction time; when the preset time is earlier than the current time, the correction of the tentative time is cancelled.
  • the second principle for correcting tentative dates can be:
  • the second day of the current date is determined as the correction date.
  • the third principle for correcting the tentative time can be:
  • the tentative hour is 12:00 and the tentative time is earlier than the current time
  • the tentative hour is corrected to 0:00, 0:00 is determined as the correction hour, and the next day of the current date is determined as the correction date.
  • the tentative hour is already in the 24-hour system, if the tentative time is still less than the current time, it means that we should further correct the tentative date. For example, when the current time is "seven o'clock in the evening", the sound signal collected is "five in the afternoon”. Even if the time “5 pm” is converted to “17:00” through conversion, it is still necessary to perform +1 day operation on the tentative date based on the principle of the nearest time.
  • FIG. 4 is a schematic diagram of the implementation process of verification of the time law of an information processing method according to an embodiment of the present invention.
  • the method before judging whether the specified time includes the specified time, the method further includes: step 401, obtaining the specified time based on the semantic analysis information; step 402, verifying whether the specified time conforms to the time law, and obtaining the verification result ; Step 403, when the verification result verifies that the specified time conforms to the law of time, it is determined whether the specified time includes the specified time.
  • the method Before verifying whether the designated time includes the designated time, the method also includes verifying whether the designated time complies with the law of time. Specifically, it is verified whether the format content of the specified time is reasonable, and when it is judged that the format content of the specified time is reasonable, it is judged whether the specified time includes the specified time. What needs to be added is that when there is no specified date and/or specified time at the specified time, it is judged that the format content of the specified time is reasonable. When it is judged that the format content of the specified time is unreasonable, the audio signal is synthesized through the dialogue management system, and the client is inquired to obtain a reasonable specified time.
  • the specified task included in the semantic analysis information is used to generate the target instruction, and the type of the specified task includes at least one of the following: One type, the second type used to characterize reminder tasks, the third type used to characterize memo tasks, and the fourth type used to characterize timing tasks.
  • the designated tasks include but are not limited to the above four types, and the designated tasks can also be the fifth type used to characterize deletion, etc., which will not be described in detail below.
  • the information control center equipment estimates the elapsed time. If the current time is after five o'clock in the afternoon, it will generate the target command corresponding to the alarm clock at five o'clock in the morning; if the current time is five in the afternoon Before this point, a target instruction corresponding to the alarm clock at 5 o'clock in the afternoon of the day will be generated.
  • the information control center device determines that the designated task is a normal single task or a periodic task through semantic analysis.
  • Tasks can refer to alarm clocks.
  • W3, object alarm clock ⁇ .
  • Another periodic task can be repeated responses based on the date.
  • the difference between a reminder task and an alarm clock task is judged by whether the semantic analysis information contains a reminder event or an obvious reminder keyword.
  • Such operations can also be performed by voice. For example, delete alarms, delete reminders, delete notes, cancel countdowns, etc.
  • the delete operation can support conditional deletion of specific alarms. For example, there are 3 alarm clocks set for tomorrow and the day after tomorrow, and the user may delete them through the following statement.
  • the device When the device receives these sound signals, it queries the alarm clock records recorded in the cloud, and returns the target instruction with the alarm ID to the device, and the client deletes the alarm, which effectively avoids the need for the client to speak to the user.
  • the alarm clock is in the process of querying, filtering and then deleting.
  • Fig. 5 is a schematic diagram of a scenario where an information processing method is applied in an embodiment of the present invention
  • Fig. 6 is a schematic diagram of a time inference process in a scenario where an information processing method is applied in an embodiment of the present invention.
  • This scenario includes the client 501 and the information control center device 502, and the client 501 and the information control center device 502 are in communication connection.
  • the client 501 is provided with a microphone array 5011 and an execution module 5012 for performing designated tasks.
  • Information control center equipment 502 includes Automatic Speech Recognition (ASR) module 5021, Natural Language Understanding (NLU) module 5022, Dialog Management (DM) module 5023, Text To Speech (Text To Speech) , TTS) module 5024.
  • ASR Automatic Speech Recognition
  • NLU Natural Language Understanding
  • DM Dialog Management
  • the NLP module is used for semantic analysis.
  • the DM module also includes a time estimation sub-module 50231.
  • the client collects the sound signal through the microphone array, and sends the sound signal to the information control center device.
  • the information control center device uses the ASR module for voice recognition to obtain the text information, and then the NLU module of the text information results in semantic analysis to obtain Corresponding to the semantic analysis information of the sound signal.
  • the speech analysis information is estimated by the time estimation sub-module.
  • the specified time from the semantic analysis information, and the specified time includes the specified date and the specified time. Then verify whether the specified date and specified time conform to the law of time. When the time is not in compliance with the law of time, multiple rounds of dialogue are conducted through the DM module to further obtain the specified time in compliance with the law. When it conforms to the time rule, it is judged whether the semantic analysis information includes the specified time. When it is determined that the semantic analysis information does not include the specified time, the DM module conducts multiple rounds of dialogue to further obtain the specified time. When it is determined that the semantic analysis information includes the designated time, it is determined whether the speech analysis information includes the designated date, and when it is determined that the semantic analysis information does not include the designated date, the tentative date is set as the current date.
  • the semantic analysis information includes the specified date
  • the end time is estimated, the result is returned to the client, and the DM module generates an audio signal for voice Broadcast.
  • the specified date is marked to indicate that the date is in the future.
  • no identification is performed. Then, it is judged whether the specified time includes the specified time period and the specified hour.
  • the specified time is converted through time conversion to obtain the conversion time, and the specified date and conversion time are determined as the tentative time
  • the specified date and time are determined as tentative time. Determine whether the tentative time is earlier than the current time.
  • the tentative date is corrected based on the principle of time proximity to obtain the intended time.
  • the intention time period is generated, and the intention time including the intention date, the intention time period and the intention hour are converted through the dialogue management system to obtain the target instruction.
  • the target instruction includes the audio signal and the designated task, and the target instruction is sent to The client, the client executes the specified task through the execution module, and broadcasts the audio corresponding to the audio signal.
  • FIG. 7 is a schematic diagram of an implementation module of an information control center device according to an embodiment of the present invention.
  • an embodiment of the present invention provides an information control center device.
  • the device includes: an obtaining module 701, configured to obtain semantic analysis information of a corresponding sound signal, where the semantic analysis information includes a specified time; and a speculation module 702, configured to specify Time estimation is performed to determine the intent time; the generating module 703 is used to generate a target instruction corresponding to the sound signal based on the intent time.
  • the estimation module 702 includes: a first judgment sub-module 7021, which is used to judge whether the specified time includes the specified time, and obtain the first judgment result; the second judgment sub-module 7022, which is used to obtain the first judgment result When it is judged that the specified time includes the specified time, it is judged whether the specified time includes the specified date, and the second judgment result is obtained; the third judgment sub-module 7023 is used for judging whether the specified date is the specified date when the second judgment result judges that the specified time includes the specified date After the current date, the third judgment result is obtained; the determination sub-module 7024 is used to determine the specified time and the specified date as the intended time when the third judgment result determines that the specified date is later than the current date.
  • the determining sub-module 7024 is also used to determine the current date as a tentative date when the second judgment result determines that the specified time does not include the specified date; the determining sub-module 7024 is also used to determine the tentative date The date and designated time are determined as the tentative time; the inference module also includes: a fourth judgment sub-module 7025, which is used to judge whether the tentative time is not earlier than the current time and obtain the fourth judgment result; the determination sub-module 7024 is also used to When the fourth judgment result judges that the tentative time is not earlier than the current time, the tentative time is determined as the intended time.
  • the estimation module 702 further includes: a correction sub-module 7026, which is used to correct the tentative time based on the principle of time proximity when the fourth judgment result determines that the tentative time is earlier than the current time to obtain a correction Time; the determination sub-module 7024 is also used to determine the correction time as the intended time.
  • a correction sub-module 7026 which is used to correct the tentative time based on the principle of time proximity when the fourth judgment result determines that the tentative time is earlier than the current time to obtain a correction Time
  • the determination sub-module 7024 is also used to determine the correction time as the intended time.
  • the estimation module 702 further includes: a fifth judgment sub-module 7027, which is used to judge whether the specified time includes the specified time period, and obtain the fifth judgment result; the conversion sub-module 7028, which is used to judge the fifth judgment result In order that the specified time includes the specified time period, the specified time is typed based on the time conversion rule to obtain the conversion time; the conversion time is used to determine the tentative time, the conversion time does not include the specified time period, and the conversion time and the specified time are used to represent the same time.
  • the obtaining module 701 is further configured to obtain a specified time based on semantic analysis information; the device further includes: a verification module 704, configured to verify whether the specified time complies with the time law, and obtain the verification result; when the verification result is verified as the specified time The time conforms to the law of time, and it is judged whether the specified time includes the specified time.
  • the storage medium includes a set of computer-executable instructions, which are used to execute the information processing method described in any one of the foregoing when the instructions are executed.
  • first and second are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with “first” and “second” may explicitly or implicitly include at least one of the features.
  • “plurality” means two or more than two, unless otherwise specifically defined.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Theoretical Computer Science (AREA)
  • Electric Clocks (AREA)
  • Telephonic Communication Services (AREA)
  • User Interface Of Digital Computer (AREA)
  • Machine Translation (AREA)

Abstract

一种信息处理方法、信息控制中心设备(502)及计算机可读存储介质，信息处理方法包括：获得对应声音信号的语义解析信息，语义解析信息包含指定时间(101)；基于当前时间对指定时间进行时间推测，确定意图时间(102)；基于意图时间生成对应声音信号的目标指令(103)；应用信息处理方法，信息控制中心设备(502)能够处理复杂且多样化的声音信号，语义解析信息能够提取到声音信号中的指定时间和目标意图，通过指定时间对语义解析信息进行时间推测，能够更准确地处理声音信号中提供的指定时间，使语音交互过程更加准确。

Description

信息处理方法、信息控制中心设备及计算机可读存储介质 技术领域
本发明涉及语音处理技术领域，尤其涉及一种信息处理方法、信息控制中心设备及计算机可读存储介质。
背景技术
区别于传统只能手动设置闹铃和时间的闹钟，智能闹钟能够通过语音交互的方式实现闹钟任务的设置。但目前，由于客户端本身数据处理能力的限制，客户端难以理解用户的真正意图设置的时间，用户在通过语音交互设置时间时，需要准确地提供时间的日期和时刻，客户端才能设置对应的任务，影响用户体验。
发明内容
本发明实施例提供了一种信息处理方法、信息控制中心设备及计算机可读存储介质，能够对声音信号中的时间进行时间推断，确定符合需求的意图时间。
本发明一方面提供一种信息处理方法，所述方法应用于信息控制中心设备，所述方法包括：获得对应声音信号的语义解析信息，所述语义解析信息包含指定时间；基于当前时间对所述指定时间进行时间推测，确定意图时间；基于所述意图时间生成对应声音信号的目标指令。
在一可实施方式中，所述基于所述当前时间对所述指定时间进行时间推测，确定意图时间，包括：判断所述指定时间是否包括指定时刻，获得第一判断结果；当所述第一判断结果判断为所述指定时间包括指定时刻时，判断所述指定时间是否包括指定日期，获得第二判断结果；当 所述第二判断结果判断为所述指定时间包括指定日期时，判断所述指定日期是否晚于当前日期，获得第三判断结果；当所述第三判断结果判断为所述指定日期晚于当前日期时，将所述指定时刻和指定日期确定为意图时间。
在一可实施方式中，所述方法还包括：当所述第二判断结果判断为所述指定时间不包括指定日期时，将所述当前日期确定为暂定日期；将所述暂定日期和所述指定时刻确定为暂定时间；判断所述暂定时间是否不早于当前时间，获得第四判断结果；当第四判断结果判断为所述暂定时间不早于当前时间时，将所述暂定时间确定为意图时间。
在一可实施方式中，所述方法还包括：当所述第四判断结果判断为所述暂定时间早于当前时刻时，基于时间就近原则对所述暂定时间进行校正，获得校正时间；将所述校正时间确定为意图时间。
在一可实施方式中，时间就近原则包括如下原则至少之一：用于校正暂定时刻的第一原则、用于校正暂定日期的第二原则、用于校正暂定时间的第三原则。
在一可实施方式中，所述方法还包括：判断所述指定时刻是否包括指定时段，获得第五判断结果；当所述第五判断结果判断为所述指定时刻包括指定时段，基于时刻转换规则将所述指定时刻进行类型转换，获得转换时刻；所述转换时刻用于确定所述暂定时间，所述转换时刻不包括指定时段，且所述转换时刻与所述指定时刻用于表征同一时间。
在一可实施方式中，所述时刻转换规则包括如下至少之一：用于转换时刻类型的第一转换规则、用于校正口误的第二转换规则和用于处理时刻临界点的第三转换规则。
在一可实施方式中，在判断所述指定时间是否包括指定时刻之前，所述方法还包括：基于语义解析信息获得指定时间；验证所述指定时间是否符合时间规律，获得验证结果；当所述验证结果验证为所述指定时 间符合时间规律，判断所述指定时间是否包括指定时刻。
本发明一方面提供一种信息控制中心设备，所述设备包括：获得模块，用于获得对应声音信号的语义解析信息，所述语义解析信息包含指定时间；推测模块，用于基于当前时间对所述指定时间进行时间推测，确定意图时间；生成模块，用于基于所述意图时间生成对应声音信号的目标指令。
在一可实施方式中，所述推测模块，包括：第一判断子模块，用于判断所述指定时间是否包括指定时刻，获得第一判断结果；第二判断子模块，用于当所述第一判断结果判断为所述指定时间包括指定时刻时，判断所述指定时间是否包括指定日期，获得第二判断结果；第三判断子模块，用于当所述第二判断结果判断为所述指定时间包括指定日期时，判断所述指定日期是否晚于当前日期，获得第三判断结果；确定子模块，用于当所述第三判断结果判断为所述指定日期晚于当前日期时，将所述指定时刻和指定日期确定为意图时间。
在一可实施方式中，所述确定子模块，还用于当所述第二判断结果判断为所述指定时间不包括指定日期时，将所述当前日期确定为暂定日期；所述确定子模块，还用于将所述暂定日期和所述指定时刻确定为暂定时间；所述推测模块，还包括：第四判断子模块，用于判断所述暂定时间是否不早于当前时间，获得第四判断结果；所述确定子模块，还用于当第四判断结果判断为所述暂定时间不早于当前时间时，将所述暂定时间确定为意图时间。
在一可实施方式中，所述推测模块，还包括：校正子模块，用于当所述第四判断结果判断为所述暂定时间早于当前时刻时，基于时间就近原则对所述暂定时间进行校正，获得校正时间；所述确定模块，还用于将所述校正时间确定为意图时间。
在一可实施方式中，所述推测模块，还包括：第五判断子模块，用 于判断所述指定时刻是否包括指定时段，获得第五判断结果；转换子模块，用于当所述第五判断结果判断为所述指定时刻包括指定时段，基于时刻转换规则将所述指定时刻进行类型转换，获得转换时刻；所述转换时刻用于确定所述暂定时间，所述转换时刻不包括指定时段，且所述转换时刻与所述指定时刻用于表征同一时间。
在一可实施方式中，所述获得模块，还用于基于语义解析信息获得指定时间；所述设备还包括：验证模块，用于验证所述指定时间是否符合时间规律，获得验证结果；当所述验证结果验证为所述指定时间符合时间规律，判断所述指定时间是否包括指定时刻。
本发明另一方面提供一种计算机可读存储介质，所述存储介质包括一组计算机可执行指令，当所述指令被执行时用于执行上述任一项所述的信息处理方法。
本发明实施例提供的信息处理方法、信息控制中心设备及计算机可读存储介质，信息控制中心设备应用本发明实施例提供的信息处理方法，能够处理复杂且多样化的声音信号，语义解析信息能够提取到声音信号中的指定时间，通过指定时间对语义解析信息进行时间推测，能够更准确地处理声音信号中提供的指定时间，使语音交互过程更加准确。
附图说明
通过参考附图阅读下文的详细描述，本发明示例性实施方式的上述以及其他目的、特征和优点将变得易于理解。在附图中，以示例性而非限制性的方式示出了本发明的若干实施方式，其中：
在附图中，相同或对应的标号表示相同或对应的部分。
图1为本发明实施例一种信息处理方法的实现流程示意图；
图2为本发明实施例一种信息处理方法时间推断的实现流程示意图。
图3为本发明实施例一种信息处理方法时刻转换的实现流程示意图。
图4为本发明实施例一种信息处理方法时间规律验证的实现流程示意图。
图5为本发明实施例应用信息处理方法的场景示意图；
图6为本发明实施例应用信息处理方法的场景中时间推断的流程示意图；
图7为本发明实施例一种信息控制中心设备的实现模块示意图。
具体实施方式
为使本发明的目的、特征、优点能够更加的明显和易懂，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而非全部实施例。基于本发明中的实施例，本领域技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
图1为本发明实施例一种信息处理方法的实现流程示意图。
参见图1，本发明实施例一方面提供一种信息处理方法，方法应用于信息控制中心设备，示例性地，信息控制中心设备可以为云服务器，该方法包括：步骤101，获得对应声音信号的语义解析信息，语义解析信息包含指定时间；步骤102，基于当前时间对指定时间进行时间推测，确定意图时间；步骤103，基于意图时间生成对应声音信号的目标指令。
本实施例提供的信息处理方法应用在信息控制中心设备，能够处理复杂、多样化的声音信号，声音信号经过语音识别和语义解析，获得的语义解析信息中包括指定时间和指定任务，其中，指定任务为声音信号中对应的目标意图；指定时间为声音信号中对应执行目标意图的执行时间。当指定时间不准确时，通过当前时间对指定时间进行时间推测，能够准确地确定指定任务对应的意图时间，信息控制中心设备能够依据意图时间设定目标指令，指示客户端在意图时间执行用户的指定任务，使 语音交互对应的结果更加准确。
信息控制中心设备选为具有数据处理能力的设备，本实施例中，信息控制中心设备选为云服务器，通过云服务器对声音信号进行语音处理，可以较大程度降低对客户端硬件的要求，并且不需要在客户端进行复杂的语义规则定制工作，在云服务器进行语义升级时，也不需要更新客户端。本实施例客户端为与信息控制中心设备(例如，云服务器)进行信号传输的终端，且客户端具有执行目标指令的对应功能，即目标指令可以根据客户端的功能进行设置，客户端选为闹钟。
本方法包括获得对应声音信号的语义解析信息，语义解析信息包含指定时间。声音信号可以通过音频采集装置进行采集。本实施例中，进行声音信号采集的音频采集装置选用由一定数量的麦克风组成的麦克风阵列，麦克风阵列装设在客户端上。语义解析信息为声音信号经过语音识别处理和语义解析处理后得到的信息。进一步的，麦克风阵列通过信号处理算法进行声音信号的采集，能够识别声源的方向，同时还能够一定程度的将背景声音清除掉，进而提升后续的语音识别的准确性。利用网络将声音信号传送到信息控制中心设备，信息控制中心设备通过ASR语音识别技术对声音信号进行语音识别，ASR语音识别技术能够将获取到的声音信号转化为语音对应的文本信息。本方法中，ASR语音识别技术基于家居环境下的声学模型与语言模型进行训练，能够更适应家居场景，在嘈杂场景下也能够对声音信号进行准确识别，获得准确的文本信息。文本信息通过语义解析模块进行语义解析，语义解析模块能够将文本信息解析为语义解析信息。比如，在一种情况下，文本信息为“七点提醒我买火车票”将会被解析为以下语义解析信息，语义解析信息以实体结构进行体现{时间＝07：00，提醒事件＝买火车票，任务对象＝提醒}；在另一种情况下，文本信息为“五点的闹钟”，将会被解析为以下结构{时间＝05：00，任务对象＝闹钟}。
本方法还包括信息控制中心设备基于当前时间对指定时间进行时间推测，确定意图时间。语义解析模块将语义解析信息发送至对话管理系统，对话管理系统用于基于指定时间对语义解析信息进行时间推测，获得意图时间。不直接将语义解析信息返回给客户端，能够避免客户端需要多次向信息控制中心设备发起请求，进而提升客户端的性能与响应时间。
可以理解的是，用于生成目标指令的意图时间需要包含日期信息、时刻信息。其中，由于时刻信息存在12小时制和24小时制，时刻信息还包括时段信息和小时信息。例如，一个满足意图时间准确性要求的时间信息包含{日期＝20190305，时段＝下午，小时＝5：00}。
而由声音信号解析的指定时间通常只有小时信息，不会完全满足上述要求。因此，需要基于当前时间对指定时间进行时间推断，以获得满足意图时间需要的时间信息。例如：当语义解析信息中获得的指定时间为{小时＝5：00}，基于当前时间对指定时间进行时间推断，当前时间为{日期＝20190305，时段＝下午，小时＝4：00},则获得的意图时间为{日期＝20190305，时段＝下午，小时＝17：00}。即选取离当前时间最近、且能够符合指定时间描述的未来时间作为意图时间。
本方法还包括信息控制中心设备基于意图时间生成对应声音信号的目标指令。根据指定任务和意图时间，能够生成对应声音信号的目标指令。目标指令用于指示执行声音信号中的指定任务。例如，当声音信号为“7：00提醒我买火车票”，信息控制中心设备根据语义解析获得的语义解析信息中，指定时间为“7：00”，任务对象为“提醒”，提醒事件为“买火车票”，对指定时间为“7：00”进行时间推断，基于时间就近原则，确定当前时间为“2019年12月26日，8点”，则确定意图时间为“2019年12月27日，7：00”。然后生成对应意图时间为“2019年12月27日，7：00”，任务对象为“提醒”，提醒事件为“买火车票”的目 标指令，并将目标指令发送至客户端，以使客户端在2019年12月27日7：00，通过语音播报提醒购买火车票。需要说明的是，接收目标指令的对象不限于客户端，当目标指令的指定任务为对信息控制中心设备进行更改时，目标指令的接收对象为信息控制中心设备；当目标指令的指定任务为对第三方终端进行控制时，目标指令的接收对象还可以为第三方终端。
具体的，本方法还包括，信息控制中心设备根据意图时间和指定任务生成任务指令，根据意图时间进行语音合成处理，得到对应指定任务的音频信号；将任务指令和音频信号确定成目标指令；将目标指令发送至客户端，以使客户端执行任务指令和播报音频信号。
在获得意图时间后，通过对话管理模块对意图时间和指定任务进行处理，从而获得包括音频信号和任务指令的目标指令，然后将目标指令发送至客户端，客户端通过解析目标指令，执行任务指令和播报音频信号，形成与用户的语音交互。具体的，对话管理系统在收到意图时间后，将会根据对话的状态，通过语音合成技术返回不一样的音频回复信息，比如对话管理系统收到下意图时间为{日期＝20190305，时段＝下午，小时＝05：00，任务对象＝闹钟}时，将会下放设置闹钟的任务指令给客户端，同时生成一条文本回复“五点的闹钟设置成功”，该本文回复通过语音合成技术合成音频信号发送给客户端，由于闹钟已经设置成功，对话管理系统也会在返回的数据中包含对话结束的状态。
图2为本发明实施例一种信息处理方法时间推断的实现流程示意图。该方法应用于信息控制中心设备。
参见图2，在本发明实施例中，步骤102，基于当前时间对指定时间进行时间推测，确定意图时间，包括：步骤1021，判断指定时间是否包括指定时刻，获得第一判断结果；步骤1022，当第一判断结果判断为指定时间包括指定时刻时，判断指定时间是否包括指定日期，获得第二判 断结果；步骤1023，当第二判断结果判断为指定时间包括指定日期时，判断指定日期是否晚于当前日期，获得第三判断结果；步骤1024，当第三判断结果判断为指定日期晚于当前日期时，将指定时刻和指定日期确定为意图时间。
可以理解的是，通过语义解析得到的指定时间可能包括以下两种情况，一种情况下，解析信息中存在指定时间；另一种情况下，解析信息中不存在指定时间。当解析信息中不存在指定时间时，设备无法生成具有意图时间和指定任务的目标指令，需要通过语音合成技术进行多轮对话交互，以获取指定时间。具体的，当解析信息中不存在指定时间时，信息控制中心设备可以生成询问指令，并发送至客户端，询问指令以语音合成技术生成，通过客户端进行播报，用于询问指定时间。例如“您希望过多久提醒您呢？”然后再次采集声音信号，通过语音识别和语义分析该声音信号，获得该声音信号中的补充时间，补充时间用于对指定时间进行补充，可以将补充时间确定为指定时间。
例如，当声音信号为”设置闹钟“时，语义解析信息为{任务对象＝闹钟，操作＝设置}，不存在指定时间，对话管理系统将会生成一条文本回复“好的，你想设置几点的闹钟”经过语音合成技术将文本转成音频后发送给客户端，同时将对话状态还未结束的信息也返回给客户端，客户端将会在播放这条音频后，通过麦克风阵列收集用户的声音信息以确定闹钟的指定时间。
进一步的，当解析信息中存在指定时间时，指定时间中可能包括指定日期和指定时刻。当语义解析信息中不存在指定时刻时，设备无法确定执行指定任务的指定时间，为将上述情况进行区分，需要判断指定时间是否包括指定时刻。当第一判断结果判断为语义解析信息中不存在指定时刻时，设备需要通过语音合成技术进行多轮对话交互，以获取包含指定时间的指定时间。
当第一判断结果判断为语义解析信息中存在指定时刻时，进一步判断指定时间是否包括指定日期，以获得第二判断结果。一种情况下，第二判断结果为指定时间包括指定日期；另一种情况下，指定时间不包括指定日期。当判断为指定时间包括指定日期时，需要判断指定日期是否晚于当前日期，获得第三判断结果。当第三判断结果判断为指定日期晚于当前日期，将指定时刻和指定日期确定为意图时间。
当第三判断结果判断为指定日期不晚于当前日期，且指定日期早于当前日期，则指定时间早于当前日期，设备无法在过去的时间指示执行指定任务，因此可以判定该语义解析信息无效，或通过语音合成技术生成语音，通过进行多轮对话交互，以再次获取指定时间。当第三判断结果判断为指定日期不晚于当前日期，且指定时期和当前日期为同一日，则需要对指定时刻进行进一步比较以进行推测，如步骤10211所示，当第三判断结果不晚于当前日期，且与当前日期相同，将指定日期确定为暂定日期。
在本发明实施例中，方法还包括：步骤1025，当第二判断结果判断为指定时间不包括指定日期时，将当前日期确定为暂定日期；步骤1026，将暂定日期和指定时刻确定为暂定时间；步骤1027，判断暂定时间是否不早于当前时间，获得第四判断结果；步骤1028，当第四判断结果判断为暂定时间不早于当前时间时，将暂定时间确定为意图时间。
将当前日期确定为指定日期后，此处的指定日期与上述第三判断结果判断为指定时期和当前日期为同一日，该情况下的指定日期相同，即均是指定时期和当前日期为同一日，以下可以采用相同方法进行比较判断。即，在指定时期和当前日期为同一日时，将当前日期确定为暂定日期，并将暂定日期和指定时刻确定为暂定时间，判断暂定时间是否不早于当前时间，获得第四判断结果。根据第四判断结果确定意图时间。当第四判断结果判断为暂定时间不早于当前时间时。即设备能够对应暂定 时间的目标指令，用于指示在暂定时间执行指定任务，即可将暂定时间确定为意图时间。需要补充的是，在一种特殊情况下，暂定时间和当前时间相同，此时同样可以生成目标指令并指示在意图时间执行指定任务，即此时意图时间与当前时间相同。
在本发明实施例中，方法还包括：步骤1029，当第四判断结果判断为暂定时间早于当前时刻时，基于时间就近原则对暂定时间进行校正，获得校正时间；步骤10210，将校正时间确定为意图时间。
当第四判断结果判断为暂定时间早于当前时刻时，如果仍然将暂定时间确定为意图时间，会导致意图时间早于当前时间，设备无法指示客户端在意图时间指示执行指定任务，不符合时间规律。因此，设备需要对暂定时间进行校正，使校正时间不早于当前时间，进而使设备能够指示客户端在意图时间执行指定任务。时间就近原则用于将意图时间确定为最接近当前时间且满足指定时间描述的未来时间。
图3为本发明实施例一种信息处理方法时刻转换的实现流程示意图。该方法应用于信息控制中心设备。
参见图3，在本发明实施例中，方法还包括：步骤301，判断指定时刻是否包括指定时段，获得第五判断结果；步骤302，当第五判断结果判断为指定时刻包括指定时段，基于时刻转换规则将指定时刻进行类型转换，获得转换时刻；步骤303，转换时刻用于确定暂定时间，转换时刻不包括指定时段，且转换时刻与指定时刻用于表征同一时间。
由于从声音信号中获得的指定时刻存在有24小时制和12小时制两种类型。且不同的人对于时段的理解方式是不同的，比如晚上0:00有的人可能是表示为第二天0:00，有的人可能表示为24：00，12：30有的人可能表示00：30。为方便暂定时间的比较，需要将指定时刻转换为24小时制，同理，获取的当前时刻也选为24小时制。需要理解的是，该步骤可以在判断指定时刻是否不早于当前时刻的任一在先步骤时进行。进 一步需要补充的是，当前时刻可以为信息控制中心设备从客户端采集的当前时刻，也可以为信息控制中心设备从网络上采集的时区标准时间。
具体的，在12小时制的表述中通常包括时段和小时，如“上午7:00”，其中指定时段指代如“上午”、“下午”等用于进行时段表示的词。当获得指定时刻后，判断指定时刻中是否包括指定时段，如果包括指定时段，则认为指定时刻采用的是12小时制的表述方式，需要对指定时刻进行转换，以获得用24小时制进行表述的指定时刻。
转换规则具体包括：用于将指定时刻24小时化的第一转换规则、用于校正口误的第二转换规则和用于处理日期临界点的第三转换规则。需要补充的是，为了方便比较，在进行24小时制转换后，仍然保留时段信息，且时段信息根据小时信息进行调整。
其中，用于将指定时刻24小时化的第一转换规则可以为：
1、如果指定小时在1:00与12：00之间，并且指定时段为中午、下午或晚上及其同义词，则将指定小时加12小时，获得24小时制的转换时刻。例如“下午，7：00”转换为“下午，19：00”。
2、如果指定小时在1:00到日出时间之间(比如6：00)，并且指定时段为白天及其同义词，则指定小时加12小时，获得24小时制的转换时刻，日出时间可以根据不同区域的进行动态调整。例如“白天，3：00”转换为“下午，15：00”。
3、如果指定小时在日落时间到12：00之间，并且指定时段为晚上及其同义词，则需要将指定小时数12小时，获得24小时制的转换时刻，日落时间可以根据不同区域的进行动态调整。例如“晚上，10：00”转换为“晚上，22：00”。
用于校正口误的第二转换规则可以为：
1、如果指定小时在0：00到日出时间之间，并且指定时段为晚上及其同义词，则重置指定时段为凌晨，将指定小时确定为转换小时，并且 设置异常标识，后续通过时间就近原则进行重新确定。例如“晚上，1：00”转换为“凌晨，1：00，异常”。
2、如果指定小时大于12：00，并且指定时段为早上、上午或者凌晨及其同义词，则需要将指定时段重置，将指定小时确定为转换小时，并且设置异常标识，避免口误或者误识别产生影响。例如：“上午，15：00”转换为“下午，15：00，异常”。
用于处理时刻临界点的第三转换规则可以为：
1、如果指定小时等于24：00则需要将指定小时设置0：00，重置指定时段为凌晨，并且设置异常标识。例如“24:00”转换为“凌晨，0：00，异常”。
2、如果指定小时等于12：00，并且指定时段为晚上或者凌晨及其同义词，需要将小时数设置0：00，重置指定时段为凌晨，并且设置异常标识。例如“晚上，12:00”转换为“凌晨，0：00，异常”。
在本发明实施例中，时间就近原则包括如下原则至少之一：用于校正暂定时刻的第一原则、用于校正暂定日期的第二原则、用于校正暂定时间的第三原则。需要理解的是，暂定时间包括暂定时刻、暂定日期和暂定小时三个部分。由暂定时间进行校正获得的校正时间同样包括校正时刻、校正日期和校正小时三个部分。
用于校正暂定时刻的第一原则可以为：
1、当暂定小时不包含暂定时段，且暂定小时小于12：00时，将暂定小时增加12小时，获得预设时间；判断预设时间是否不早于当前时间，当预设时间不早于当前时间，将预设时间确定为校正时间；当预设时间早于当前时间，取消对暂定时间的校正。
用于校正暂定日期的第二原则可以为：
1、当暂定时间有异常标识，并且暂定小时小于或等于6：00时，将当前日期的第二天确定为校正日期。
2、当暂定小时不包含暂定时段，且暂定小时小于12：00时，将当前日期的第二天确定为校正日期。
3、当暂定小时位于12：00到24点之间，并且对应的声音信号中不包括指定日期，将当前日期的第二天确定为校正日期。
4、当暂定时段和暂定小时均明确，并且对应的声音信号中不包括指定日期，将当前日期的第二天确定为校正日期。
5、当暂定日期晚于当前日期，并且具有异常标识，将当前日期的第二天确定为校正日期。
用于校正暂定时间的第三原则可以为：
1、当暂定小时是12：00，并且暂定时间早于当前时间时，将暂定小时校正为0：00，将0:00确定为校正小时，将当前日期的第二天确定为校正日期。
当暂定小时已经是24小时制时，如果暂定时间仍然小于当前时间，则说明我们应该进一步校正暂定日期，比如在当前时间为“晚上七点”时，采集的声音信号为“下午五点的闹钟”，即便通过转换将时间“下午五点”转换为“17:00”，仍然需要依据时间就近原则对暂定日期进行+1天的操作。
图4为本发明实施例一种信息处理方法时间规律验证的实现流程示意图。
参见图4，在本发明实施例中，在判断指定时间是否包括指定时刻之前，方法还包括：步骤401，基于语义解析信息获得指定时间；步骤402，验证指定时间是否符合时间规律，获得验证结果；步骤403，当验证结果验证为指定时间符合时间规律，判断指定时间是否包括指定时刻。
为了对于不合理的时间给出提示，例如“二月三十五号四十六点的闹钟”，包含不存在的指定日期与指定时刻。在验证指定时间是否包括指定时刻之前，本方法还包括对指定时间进行是否符合时间规律的验证。 具体的，验证指定时间的格式内容是否合理，当判断为指定时间的格式内容合理时，判断指定时间是否包括指定时刻。需要补充的是，当指定时间不存在指定日期和/或指定时刻，判断为指定时间的格式内容合理。当判断为指定时间的格式内容不合理时，通过对话管理系统合成音频信号，通过客户端进行询问，以获得合理的指定时间。
进一步进行补充的是，当客户端为闹钟时，根据闹钟的功能，语义解析信息中包括的指定任务用于生成目标指令，指定任务的类型包括如下至少之一：用于表征闹铃任务的第一类型、用于表征提醒任务的第二类型、用于表征备忘任务的第三类型、用于表征计时任务的第四类型。根据闹钟的功能，指定任务包括但不限于以上四种，指定任务还可以是用于表征删除的第五类型等等，以下不做赘述。
以下提供几种具体实施场景进行说明。
当采集到声音信号为“设置五点的闹钟”，信息控制中心设备经过时间推测，如果当前时间在下午五点以后，会生成对应明天早上五点的闹钟的目标指令；如果当前时间在下午五点以前，将生成对应当天下午五点的闹钟的目标指令。
对于用于表征闹铃任务的第一类型任务，为了进一步方便用户设置多个任务，信息控制中心设备通过语义解析，确定指定任务为普通单次任务或周期型任务。任务可以指代闹钟。一种周期型任务可以是根据星期进行重复响铃，比如当用户说“每周一到周三早上八点叫我起床”，设备会生成一条重复周期型的闹钟设置命令{TIME＝08:00，REPEAT＝W1|W2|W3，对象＝闹钟}。另一种周期型任务可以是根据日期进行重复响应，比如当用户说“八月一号到十号每天早上八点叫我起床”时，设备会生成对应的闹钟设置指令{TIME＝08：00，REPEAT＝20190801<20190810，对象＝闹钟}。
对于用于表征备忘任务的第三类型任务。用户在说“设置备忘”后， 设备会提示用户“请告诉我备忘的内容”，接着用户就可以把备忘的内容记录在设备或客户端中。
对应用于表征计时任务的第四类型任务。比如当指定任务为“用倒计时设置跨年时间”时，采集到声音信号为“零点的倒计时”，或者说“设置新年倒计时”，就能够生成对于零点或新年的目标指令。
提醒任务与闹钟任务的区别通过语义解析信息是否带有提醒事件或者明显的提醒关键词来判断的，比如“五点叫我”语义解析的结果是{时间＝05：00，对象＝闹钟}，“五点提醒我开会”的语义解析则是{时间＝05：00，对象＝日程，事项＝开会}，通过对解析对象的区别，信息控制中心设备能够同时理解闹钟、提醒、备忘、倒计时四种时间相关的功能所涉及到的语义槽，并根据对话状态与策略返回不同的对话状态以及回复文本。
设置过的闹钟与提醒，尤其是周期型的闹钟，在一些场景用户需要可以删除特定的闹钟，这样的操作同样可以通过语音进行操作。比如删除闹钟，删除提醒，删除备忘，取消倒计时等。删除的操作能够支持带条件的删除特定闹钟。比如原来有设置了明天后天的闹钟各3个，用户可能通过以下说法进行删除。语音输入“删除明天的闹钟”对应语义{操作＝删除，日期＝20190817}。或者“删除明天早上的闹钟”对应语义{操作＝删除，日期＝20190817，时间＝06:00<12:00}。又或者“删除明天早上八点到十点之间的闹钟”对应语义{操作＝删除，日期＝20190817，时间＝08:00<08:00}。为了方便对闹钟或其他任务进行区分，在设定任务时，对任务进行特定ID的命名，并进行存储。
设备在收到这些声音信号时，通过查询云端记录的闹钟记录，返回带有闹钟ID的目标指令给设备端，由客户端删除掉闹钟，这样就有效的避免了客户端需要针对用户的说法对闹钟进行查询过滤再删除的过程。
图5为本发明实施例应用信息处理方法的场景示意图；图6为本发 明实施例应用信息处理方法的场景中时间推断的流程示意图。
参见图5和图6，为方便上述实施例的理解，以下提供一个整体流程场景进行解释。
该场景包括客户端501和信息控制中心设备502，客户端501和信息控制中心设备502进行通信连接。客户端501上设置有麦克风阵列5011和用于执行指定任务的执行模块5012。信息控制中心设备502包括自动语音识别(Automatic Speech Recognition，ASR)模块5021、自然语言理解(Natural Language Understanding,NLU)模块5022、对话管理(Dialog Management,DM)模块5023、文本转音频(Text To Speech，TTS)模块5024。NLP模块用于进行语义分析。DM模块还包括时间推测子模块50231。
客户端通过麦克风阵列采集声音信号，并将声音信号发送至信息控制中心设备，信息控制中心设备将声音信号通过ASR模块进行语音识别，获得文本信息，再将文本信息结果NLU模块进行语义解析，获得对应声音信号的语义解析信息。将语音解析信息通过时间推测子模块进行推测。
首先，从语义解析信息中获取指定时间，指定时间包括指定日期和指定时刻。然后对指定日期和指定时刻是否符合时间规律进行验证。当不符合时间规律时，通过DM模块进行多轮对话进一步获取符合规律的指定时间。当符合时间规律时，判断语义解析信息是否包括了指定时刻。当判断为语义解析信息没有包括指定时刻，通过DM模块进行多轮对话进一步获取指定时刻。当判断为语义解析信息包括指定时刻，判断语音解析信息是否包括指定日期，当判断为语义解析信息不包括指定日期，则将暂定日期设置为当前日期。当判断为语义解析信息包括指定日期，判断指定日期是否早于当前日期，当判断为指定日期早于当前日期，结束时间推测，将该结果返回至客户端，并通过DM模块生成音频信号进行语音播报。当判断为指定日期不早于当前日期，判断指定日期是否晚 于当前日期，当判断为指定日期晚于当前日期，对指定日期进行标识，以注明该日期属于未来日期。当判断为指定日期不晚于当前日期，不进行标识。然后，判断指定时刻是否包括指定时段和指定小时，当判断为指定时刻包括指定时段和指定小时，通过时刻转换对指定时刻进行转换，获得转换时刻，将指定日期和转换时刻确定为暂定时间，当判断为指定时刻不包指定时段，将指定日期和指定时刻确定为暂定时间。判断暂定时间是否早于当前时间。当判断为暂定时间早于当前时间，基于时间就近原则对暂定日期进行校正，获得意图时间。然后基于意图时间中的意图时刻生成意图时段，将包括意图日期、意图时段和意图小时的意图时间通过对话管理系统进行转换，获得目标指令，目标指令包括音频信号和指定任务，将目标指令发送至客户端，客户端通过执行模块执行指定任务，并播报对应音频信号的音频。
图7为本发明实施例一种信息控制中心设备的实现模块示意图。
本发明实施例一方面提供一种信息控制中心设备，设备包括：获得模块701，用于获得对应声音信号的语义解析信息，语义解析信息包含指定时间；推测模块702，用于基于当前时间对指定时间进行时间推测，确定意图时间；生成模块703，用于基于意图时间生成对应声音信号的目标指令。
在本发明实施例中，推测模块702，包括：第一判断子模块7021，用于判断指定时间是否包括指定时刻，获得第一判断结果；第二判断子模块7022，用于当第一判断结果判断为指定时间包括指定时刻时，判断指定时间是否包括指定日期，获得第二判断结果；第三判断子模块7023，用于当第二判断结果判断为指定时间包括指定日期时，判断指定日期是否晚于当前日期，获得第三判断结果；确定子模块7024，用于当第三判断结果判断为指定日期晚于当前日期时，将指定时刻和指定日期确定为意图时间。
在本发明实施例中，确定子模块7024，还用于当第二判断结果判断为指定时间不包括指定日期时，将当前日期确定为暂定日期；确定子模块7024，还用于将暂定日期和指定时刻确定为暂定时间；推测模块，还包括：第四判断子模块7025，用于判断暂定时间是否不早于当前时间，获得第四判断结果；确定子模块7024，还用于当第四判断结果判断为暂定时间不早于当前时刻时，将暂定时间确定为意图时间。
在本发明实施例中，推测模块702，还包括：校正子模块7026，用于当第四判断结果判断为暂定时间早于当前时刻时，基于时间就近原则对暂定时间进行校正，获得校正时间；确定子模块7024，还用于将校正时间确定为意图时间。
在本发明实施例中，推测模块702，还包括：第五判断子模块7027，用于判断指定时刻是否包括指定时段，获得第五判断结果；转换子模块7028，用于当第五判断结果判断为指定时刻包括指定时段，基于时刻转换规则将指定时刻进行类型转换，获得转换时刻；转换时刻用于确定暂定时间，转换时刻不包括指定时段，且转换时刻与指定时刻用于表征同一时间。
在本发明实施例中，获得模块701，还用于基于语义解析信息获得指定时间；设备还包括：验证模块704，用于验证指定时间是否符合时间规律，获得验证结果；当验证结果验证为指定时间符合时间规律，判断指定时间是否包括指定时刻。
本发明另一方面提供一种计算机可读存储介质，所述存储介质包括一组计算机可执行指令，当所述指令被执行时用于执行上述任一项所述的信息处理方法。
在本说明书的描述中，参考术语“一个实施例”、“一些实施例”、“示例”、“具体示例”、或“一些示例”等的描述意指结合该实施例或示例描述的具体特征、结构、材料或者特点包含于本发明的至少一个 实施例或示例中。而且，描述的具体特征、结构、材料或者特点可以在任一个或多个实施例或示例中以合适的方式结合。此外，在不相互矛盾的情况下，本领域的技术人员可以将本说明书中描述的不同实施例或示例以及不同实施例或示例的特征进行结合和组合。
此外，术语“第一”、“第二”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或隐含地包括至少一个该特征。在本发明的描述中，“多个”的含义是两个或两个以上，除非另有明确具体的限定。
以上，仅为本发明的具体实施方式，但本发明的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本发明揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本发明的保护范围之内。因此，本发明的保护范围应以权利要求的保护范围为准。

Claims (10)

  1. 一种信息处理方法，其特征在于，所述方法应用于信息控制中心设备，所述方法包括：
    获得对应声音信号的语义解析信息，所述语义解析信息包含指定时间；
    基于当前时间对所述指定时间进行时间推测，确定意图时间；
    基于所述意图时间生成对应声音信号的目标指令。
  2. 根据权利要求1所述的方法，其特征在于，所述基于所述当前时间对所述指定时间进行时间推测，确定意图时间，包括：
    判断所述指定时间是否包括指定时刻，获得第一判断结果；
    当所述第一判断结果判断为所述指定时间包括指定时刻时，判断所述指定时间是否包括指定日期，获得第二判断结果；
    当所述第二判断结果判断为所述指定时间包括指定日期时，判断所述指定日期是否晚于当前日期，获得第三判断结果；
    当所述第三判断结果判断为所述指定日期晚于当前日期时，将所述指定时刻和指定日期确定为意图时间。
  3. 根据权利要求2所述的方法，其特征在于，所述方法还包括：
    当所述第二判断结果判断为所述指定时间不包括指定日期时，将所述当前日期确定为暂定日期；
    将所述暂定日期和所述指定时刻确定为暂定时间；
    判断所述暂定时间是否不早于当前时间，获得第四判断结果；
    当第四判断结果判断为所述暂定时间不早于当前时间时，将所述暂定时间确定为意图时间。
  4. 根据权利要求3所述的方法，其特征在于，所述方法还包括：
    当所述第四判断结果判断为所述暂定时间早于当前时刻时，基于时 间就近原则对所述暂定时间进行校正，获得校正时间；
    将所述校正时间确定为意图时间。
  5. 根据权利要求4所述的方法，其特征在于，所述时间就近原则包括如下原则至少之一：用于校正暂定时刻的第一原则、用于校正暂定日期的第二原则、用于校正暂定时间的第三原则。
  6. 根据权利要求4所述的方法，其特征在于，所述方法还包括：
    判断所述指定时刻是否包括指定时段，获得第五判断结果；
    当所述第五判断结果判断为所述指定时刻包括指定时段，基于时刻转换规则将所述指定时刻进行类型转换，获得转换时刻；
    所述转换时刻用于确定所述暂定时间，所述转换时刻不包括指定时段，且所述转换时刻与所述指定时刻用于表征同一时间。
  7. 根据权利要求1所述的方法，其特征在于，所述时刻转换规则包括如下至少之一：用于转换时刻类型的第一转换规则、用于校正口误的第二转换规则和用于处理时刻临界点的第三转换规则。
  8. 根据权利要求2所述的方法，其特征在于，在判断所述指定时间是否包括指定时刻之前，所述方法还包括：
    基于语义解析信息获得指定时间；
    验证所述指定时间是否符合时间规律，获得验证结果；
    当所述验证结果验证为所述指定时间符合时间规律，判断所述指定时间是否包括指定时刻。
  9. 一种信息控制中心设备，其特征在于，所述设备包括：
    获得模块，用于获得对应声音信号的语义解析信息，所述语义解析信息包含指定时间；
    推测模块，用于基于当前时间对所述指定时间进行时间推测，确定意图时间；
    生成模块，用于基于所述意图时间生成对应声音信号的目标指令。
  10. 一种计算机可读存储介质，所述存储介质包括一组计算机可执行指令，当所述指令被执行时用于执行权利要求1-8任一项所述的信息处理方法。
PCT/CN2020/127639 2019-12-30 2020-11-09 信息处理方法、信息控制中心设备及计算机可读存储介质 WO2021135652A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP20909397.0A EP4086895A4 (en) 2019-12-30 2020-11-09 INFORMATION PROCESSING PROCEDURES, INFORMATION CONTROL CENTER AND COMPUTER READABLE STORAGE MEDIUM
JP2022540600A JP2023509651A (ja) 2019-12-30 2020-11-09 情報処理方法、情報制御センター装置及びコンピュータ読み取り可能な記憶媒体
US17/758,051 US20230032792A1 (en) 2019-12-30 2020-11-09 Information processing method, information control center device, and computer-readable storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911396185.1 2019-12-30
CN201911396185.1A CN111192579B (zh) 2019-12-30 2019-12-30 信息处理方法、信息控制中心设备及计算机可读存储介质

Publications (1)

Publication Number Publication Date
WO2021135652A1 true WO2021135652A1 (zh) 2021-07-08

Family

ID=70707791

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/127639 WO2021135652A1 (zh) 2019-12-30 2020-11-09 信息处理方法、信息控制中心设备及计算机可读存储介质

Country Status (5)

Country Link
US (1) US20230032792A1 (zh)
EP (1) EP4086895A4 (zh)
JP (1) JP2023509651A (zh)
CN (1) CN111192579B (zh)
WO (1) WO2021135652A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111192579B (zh) * 2019-12-30 2022-09-23 思必驰科技股份有限公司 信息处理方法、信息控制中心设备及计算机可读存储介质

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090048832A1 (en) * 2005-11-08 2009-02-19 Nec Corporation Speech-to-text system, speech-to-text method, and speech-to-text program
CN103440866A (zh) * 2013-07-30 2013-12-11 广东明创软件科技有限公司 根据通话信息执行任务的方法及移动终端
WO2014176750A1 (en) * 2013-04-28 2014-11-06 Tencent Technology (Shenzhen) Company Limited Reminder setting method, apparatus and system
CN106020953A (zh) * 2016-05-12 2016-10-12 青岛海信移动通信技术股份有限公司 一种在电子日历中建立日程的方法和装置
CN106941619A (zh) * 2017-03-16 2017-07-11 百度在线网络技术（北京）有限公司 基于人工智能的节目提醒方法、装置以及系统
CN107465599A (zh) * 2017-08-15 2017-12-12 竞技世界（北京）网络技术有限公司 一种即时通讯中的日程设置方法及装置
CN111192579A (zh) * 2019-12-30 2020-05-22 苏州思必驰信息科技有限公司 信息处理方法、信息控制中心设备及计算机可读存储介质

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090048832A1 (en) * 2005-11-08 2009-02-19 Nec Corporation Speech-to-text system, speech-to-text method, and speech-to-text program
WO2014176750A1 (en) * 2013-04-28 2014-11-06 Tencent Technology (Shenzhen) Company Limited Reminder setting method, apparatus and system
CN103440866A (zh) * 2013-07-30 2013-12-11 广东明创软件科技有限公司 根据通话信息执行任务的方法及移动终端
CN106020953A (zh) * 2016-05-12 2016-10-12 青岛海信移动通信技术股份有限公司 一种在电子日历中建立日程的方法和装置
CN106941619A (zh) * 2017-03-16 2017-07-11 百度在线网络技术（北京）有限公司 基于人工智能的节目提醒方法、装置以及系统
CN107465599A (zh) * 2017-08-15 2017-12-12 竞技世界（北京）网络技术有限公司 一种即时通讯中的日程设置方法及装置
CN111192579A (zh) * 2019-12-30 2020-05-22 苏州思必驰信息科技有限公司 信息处理方法、信息控制中心设备及计算机可读存储介质

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4086895A4 *

Also Published As

Publication number Publication date
US20230032792A1 (en) 2023-02-02
EP4086895A1 (en) 2022-11-09
JP2023509651A (ja) 2023-03-09
EP4086895A4 (en) 2023-06-14
CN111192579B (zh) 2022-09-23
CN111192579A (zh) 2020-05-22

Similar Documents

Publication Publication Date Title
US20200402515A1 (en) Dialog management with multiple modalities
KR102523982B1 (ko) 자동화된 어시스턴트를 호출하기 위한 다이내믹 및/또는 컨텍스트-특정 핫 워드
EP3360313B1 (en) Feedback controller for data transmissions
CN110050303B (zh) 基于第三方代理内容的语音到文本转换
CN106504743B (zh) 一种用于智能机器人的语音交互输出方法及机器人
KR102393876B1 (ko) 클라이언트-컴퓨팅된 콘텐츠 메타데이터에 기반한 음성 질의 QoS
CN111261151B (zh) 一种语音处理方法、装置、电子设备及存储介质
JP7230806B2 (ja) 情報処理装置、及び情報処理方法
WO2015054352A1 (en) Using voice commands to execute contingent instructions
US20220392450A1 (en) Systems and methods for addressing possible interruption during interaction with digital assistant
WO2021135652A1 (zh) 信息处理方法、信息控制中心设备及计算机可读存储介质
US20220066731A1 (en) Automatic adjustment of muted response setting
KR20230117239A (ko) 자동화된 어시스턴트 상호작용에서 레이턴시를 줄이기위한 방법 및 시스템
US11347379B1 (en) Captions for audio content
CN110958348B (zh) 语音处理方法、装置、用户设备及智能音箱
CN116016779A (zh) 语音通话翻译辅助方法、系统、计算机设备和存储介质
CN112188253A (zh) 语音控制方法、装置、智能电视和可读存储介质
US11735186B2 (en) Hybrid live captioning systems and methods
US11990116B1 (en) Dynamically rendered notifications and announcements
US11463507B1 (en) Systems for generating captions for audio content
JP2000081891A (ja) 認識対象音声の入力状態報知方法及び音声認識装置並びに認識対象音声の入力状態報知処理プログラムを記録した記録媒体
JP2006091912A (ja) 音声認識方法及び音声認識装置並びに音声認識処理プログラムを記録した記録媒体
CN117238284A (zh) 语音处理方法、装置及相关设备
CN114041283A (zh) 利用事件前和事件后输入流来接洽自动化助理
CN112699670A (zh) 离线虚拟助理的自动同步

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20909397

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022540600

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2020909397

Country of ref document: EP

Effective date: 20220801