WO2020253064A1 - 语音的识别方法及装置、计算机设备、存储介质 - Google Patents

语音的识别方法及装置、计算机设备、存储介质 Download PDF

Info

Publication number
WO2020253064A1
WO2020253064A1 PCT/CN2019/118415 CN2019118415W WO2020253064A1 WO 2020253064 A1 WO2020253064 A1 WO 2020253064A1 CN 2019118415 W CN2019118415 W CN 2019118415W WO 2020253064 A1 WO2020253064 A1 WO 2020253064A1
Authority
WO
WIPO (PCT)
Prior art keywords
natural language
dialogue
corpus
dialogue context
context
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2019/118415
Other languages
English (en)
French (fr)
Inventor
张师琲
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ping An Technology Shenzhen Co Ltd
Original Assignee
Ping An Technology Shenzhen Co Ltd
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 Ping An Technology Shenzhen Co Ltd filed Critical Ping An Technology Shenzhen Co Ltd
Publication of WO2020253064A1 publication Critical patent/WO2020253064A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/60Information retrieval; Database structures therefor; File system structures therefor of audio data
    • G06F16/63Querying
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/08Speech classification or search
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/08Speech classification or search
    • G10L15/18Speech classification or search using natural language modelling
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/22Procedures used during a speech recognition process, e.g. man-machine dialogue
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/26Speech to text systems
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208Noise filtering
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/08Speech classification or search
    • G10L2015/088Word spotting
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00Speech recognition
    • G10L15/22Procedures used during a speech recognition process, e.g. man-machine dialogue
    • G10L2015/223Execution procedure of a spoken command
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208Noise filtering
    • G10L2021/02087Noise filtering the noise being separate speech, e.g. cocktail party

Definitions

  • This application relates to the computer field, in particular to a method and device for speech recognition, computer equipment, and storage media.
  • Chatbots are an important field in artificial intelligence. Through users sending natural language to the chatbots, the robots can complete a series of operations.
  • chat bots can only do some simple chatting conversations, and few can distinguish the true intentions of people, and cannot continuously understand the conversation intentions for a certain voice with limited material or combined with context.
  • the recognition rate is low.
  • the recognition rate can only be improved by acquiring a large amount of natural language, resulting in a low application range of speech recognition.
  • embodiments of the present application provide a voice recognition method and device, computer equipment, and storage medium.
  • an embodiment of the present application provides a voice recognition method, the method includes: collecting natural language; determining a dialogue context of the natural language, wherein the dialogue context is used to describe the current field of voice interaction And an interactive object; using a corpus database corresponding to the dialogue context to identify the natural language.
  • an embodiment of the present application provides a speech recognition device, the device includes: a collection module for collecting natural language; a determining module for determining the dialogue context of the natural language, wherein the The conversation context is used to describe the field of current voice interaction and the interaction object; the recognition module is used to recognize the natural language using a corpus database corresponding to the conversation context.
  • a storage medium in which a computer program is stored, wherein the computer program is configured to execute the steps in any one of the foregoing method embodiments when running.
  • an electronic device including a memory and a processor, the memory is stored with a computer program, and the processor is configured to run the computer program to execute any of the above Steps in the method embodiment.
  • FIG. 1 is a hardware structure block diagram of a mobile terminal for voice recognition according to an embodiment of the present application
  • Fig. 2 is a flowchart of a voice recognition method according to an embodiment of the present application
  • Figure 3 is a flowchart of identifying natural language in an embodiment of the present application.
  • Figure 4 is a schematic diagram of interaction in an embodiment of the present application.
  • Fig. 5 is a structural block diagram of a speech recognition device according to an embodiment of the present application.
  • FIG. 1 is a hardware structure block diagram of a voice recognition mobile terminal according to an embodiment of the present application.
  • the mobile terminal 10 may include one or more (only one is shown in FIG. 1) processor 102 (the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA. ) And a memory 104 for storing data.
  • the above mobile terminal may also include a transmission device 106 and an input/output device 108 for communication functions.
  • FIG. 1 is only for illustration, and does not limit the structure of the above-mentioned mobile terminal.
  • the mobile terminal 10 may also include more or fewer components than those shown in FIG. 1, or have a different configuration from that shown in FIG.
  • the memory 104 may be used to store computer programs, for example, software programs and modules of application software, such as computer programs corresponding to the voice recognition method in the embodiment of the present application.
  • the processor 102 executes the computer programs stored in the memory 104 by running Various functional applications and data processing, namely to achieve the above methods.
  • the memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory.
  • the memory 104 may further include a memory remotely provided with respect to the processor 102, and these remote memories may be connected to the mobile terminal 10 via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
  • the transmission device 106 is used to receive or send data via a network.
  • the above-mentioned specific example of the network may include a wireless network provided by the communication provider of the mobile terminal 10.
  • the transmission device 106 includes a network adapter (Network Interface Controller, NIC for short), which can be connected to other network devices through a base station to communicate with the Internet.
  • the transmission device 106 may be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet in a wireless manner.
  • RF Radio Frequency
  • FIG. 2 is a flowchart of a voice recognition method according to an embodiment of the present application. As shown in FIG. 2, the process includes the following steps:
  • Step S202 collecting natural language
  • Step S204 Determine the dialogue context of the natural language, where the dialogue context is used to describe the current voice interaction field and interaction object;
  • the field of contextual interaction is an interactive scene, such as weather query, route navigation, dialogue and chat, etc.
  • the interactive object is a target object controlled by voice, for example, the target object is the query engine of the system.
  • the recognition result corresponding to the natural voice, or according to the instruction information in the natural voice also includes the call engine, which is used to call the software or hardware of the device to assist the recognition, such as calling the built-in GPS module, etc.
  • Step S206 Recognizing the natural language using a corpus database corresponding to the dialogue context.
  • the execution subject of this embodiment may be a chat robot, a voice console, a voice central control device, an electronic device with a voice control function, etc., but it is not limited to this.
  • the dialogue context in this embodiment includes a dialogue scene, objects and equipment that the machine interacts with, etc., and it is determined that the dialogue context includes multiple solutions.
  • determining the dialogue context of the natural language includes:
  • the natural language in this embodiment is the voice uttered by a natural person or device.
  • the voice is first converted into text, and then a preset keyword list is used to find out whether any of the keyword lists is included in the text. Keywords, if included, extract matching keywords.
  • S12 Obtain a dialogue context through the keyword matching, wherein the dialogue context includes one of the following: knowledge search, Internet search, weather query, route navigation, dialogue chat, and command control.
  • the keywords and the dialogue context in this embodiment are matched by a preset mapping table.
  • the keyword set ⁇ a,b,c ⁇ corresponds to the dialogue context of knowledge search
  • the keyword set ⁇ b,c,d ⁇ corresponds to Internet search.
  • third-party software (such as maps, browsers, etc.) needs to be started, in some contexts, local knowledge graphs are needed, and in some contexts, corpus databases are used.
  • the map software needs to be started.
  • the collected natural language is "home” and the context is determined as route navigation.
  • the map software needs to be called to search for the driving navigation route from the current location to home; or in the knowledge
  • the natural language collected is "Zhang San's contact information”
  • the context is determined to be a knowledge search
  • Zhang San's contact information is searched in the local address book.
  • determining the dialogue context of the natural language includes:
  • the historical dialogue record in this embodiment is the dialogue content between the voice recognition device and the voice recognition object before the current natural language.
  • the dialogue record is stored in the local memory of the voice device.
  • time can be used as a parameter. Search for conversation records within a preset time range. For example, if a user conducts a human-machine conversation with a voice robot, three conversation records are generated within 1 minute, namely: A, B, C, the voice robot currently receives to be recognized If the natural voice of D is D, the dialogue scene of D can be determined according to the contents of A, B, and C. If the contents of D are related to A, B, and C, the dialogue environment is also the same. This is because A normal man-machine conversation will include multiple conversation records.
  • the dialogue scene is determined by the current natural language context (mainly the above), matching the corpus of the scene field, and then identifying: under the AIML mechanism, the context information in the dialogue is recorded, based on the context information, Use the configured topic library to match and recognize the corpus conversation of a certain domain, and then use the corpus of the domain for recognition.
  • the conversation scene is determined to be "TV control” through the previous chat records, and then "the background is too dark” is recognized in the corpus of the TV field, and then the light is brightened Background brightness, or determine the dialogue scene as "light control”, and then identify "lighting equipment is too dark” in the corpus of the lighting field, and then turn on the lighting equipment.
  • the method before using the corpus database corresponding to the dialogue context to recognize the natural language, the method further includes: setting the corpus database, where the corpus database includes: multiple rounds of dialogue data, multi-scene corpus , Internet corpus obtained through the solr (Search On Lucene Replication) search engine.
  • the above three corpora in this embodiment are three-dimensional databases, which can be used alone or in conjunction with each other.
  • multi-round dialogue data refers to a corpus that includes multiple sets of dialogues, which can be used for chatting, local query, etc., such as Food, TV, movies, music, work and other topics (each category includes multiple sets of logical dialogues).
  • Multi-scene corpus is a corpus under multiple specific scenarios, which can be used for chat, local query, machine control, etc., such as in For food-related fields and music-related fields, for the same question: "Ask the most popular places", the responses will be different.
  • the feedback is the restaurant and the concert hall, and the online corpus is the result of the online query As corpus.
  • an AIML model is used to configure the corpus database, AIML artificial intelligence markup language, AIML describes a class of data objects called AIML objects, and partially describes the behavior of computer programs that process them.
  • AIML objects are composed of units called topics and categories, which contain any parsed or unparsed data.
  • the parsed data is character data (character data) obtained by character analysis and AIML elements (AIML elements).
  • AIML elements are encapsulated in the stimulus-response knowledge contained in the document.
  • the character data in these elements can be parsed by the AIML interpreter or by the responder (Responder) Resolve.
  • the Solr search engine in this embodiment inputs natural language, and outputs corresponding answers in natural language.
  • the corpus database is configured by configuring the corresponding question template.
  • the template format is written in the form of AIML template.
  • the answer sentences for multiple sets of preset questions are obtained by crawling online, such as Baidu Baike, etc., and according to the type of question Link to the corresponding URL.
  • Fig. 3 is a flowchart of identifying natural language in an embodiment of the present application. As shown in Fig. 3, recognizing the natural language using a corpus database corresponding to the dialogue context includes:
  • S302 Query a local corpus database corresponding to the dialogue context, where the local corpus database includes: multiple rounds of dialogue data and multi-scene corpus;
  • the dialogue context in this embodiment includes knowledge search, Internet search, weather query, route navigation, dialogue chat, command control, etc., among which the dialogue chat is multi-round dialogue materials, knowledge search, Internet search, weather query, route navigation, etc.
  • the command control is a multi-scenario corpus.
  • each scene of the dialogue context corresponds to a dedicated corpus database, but when the recognition of the multi-scenario corpus fails, it will be converted to a multi-round dialogue corpus for further inquiry
  • the collected natural language is: "Navigate to the front office building", which is determined as the dialogue context for route navigation.
  • the query for "front office building" in the local corpus database is unsuccessful, and then it turns to multiple rounds of dialogue materials to trigger Question: "Which office building is it?", the user answers: "A office building".
  • the method further includes: outputting response content corresponding to the recognition result, or generating a control instruction corresponding to the recognition result.
  • the natural language is converted into machine language that can be recognized by the corpus database
  • the corresponding semantics is searched in the local corpus database.
  • the solr search engine is used to search.
  • the response content under the semantics is further searched. Or it is a control instruction for the target object. For example, if the natural language collected is "what's the weather today", the weather condition can be searched out, and the response content "it is sunny today, 24°C, good air index", or the collected language To "the TV sound is too loud”, match the control command to lower the TV volume and send it to the TV console.
  • the natural language collected is "what's the weather today"
  • the weather condition can be searched out
  • the response content "it is sunny today, 24°C, good air index”
  • match the control command to lower the TV volume and send it to the TV console.
  • it is recognized in the corpus database it is searched in the anticipation database, and if it is recognized by a search engine, it is searched through the
  • FIG. 4 is a schematic diagram of the interaction of an embodiment of the present application.
  • Three contextual corpus databases of Internet search, weather query, and route navigation are built-in locally.
  • the solr search engine When the solr search engine also fails to recognize, when all the corpus databases are used to identify the user’s natural language, the error is reported.
  • the user will assist in completing the dialogue or operation, and further improve the corpus database, such as feeding the results to the corpus database, Modify or add the recognition content of the current question and answer in the local corpus, and learn the natural language that fails to be recognized.
  • collecting natural language includes: collecting background noise voice signals through a microphone; analyzing the background noise and the sound ray characteristics of the voice signal; removing the background noise according to the sound ray characteristics of the historical voice , Converting the voice signal into text to obtain the natural language, wherein the historical voice is a sample voice collected under a noise-free condition.
  • the solution of this embodiment is applied to a chat robot, it is optimized based on the AIML model, the universal language of international robots. Because the label configuration of AIML is very flexible, it can support the configuration of custom scenes, multiple rounds of dialogue materials, and multiple applications.
  • This scenario allows the robot to have more intelligent and professional dialogue functions; the addition of the solr search engine in the background can dock richer corpus resources and adapt to semantic-based quick search and dialogue functions; the robot has a learning function and can In conversations with people, learn sentences and answers that were previously unintelligible, and have the function of self-growth.
  • the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, of course, it can also be implemented by hardware, but in many cases the former is Better implementation.
  • the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as ROM/RAM, magnetic disk, The optical disc) includes several instructions to enable a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the method described in each embodiment of the present application.
  • a speech recognition device is also provided, which is used to implement the above-mentioned embodiments and preferred implementations, and those that have been described will not be repeated.
  • the term "module” can implement a combination of software and/or hardware with predetermined functions.
  • the devices described in the following embodiments are preferably implemented by software, hardware or a combination of software and hardware is also possible and conceived.
  • Fig. 5 is a structural block diagram of a speech recognition device according to an embodiment of the present application. As shown in Fig. 5, the device includes: a collection module 50, a determination module 52, and a recognition module 54, wherein,
  • the collection module 50 is used to collect natural language
  • the determining module 52 is configured to determine the dialogue context of the natural language, where the dialogue context is used to describe the current voice interaction field and interaction object;
  • the recognition module 54 is configured to recognize the natural language using a corpus database corresponding to the dialogue context.
  • the determining module includes: an extraction unit, configured to extract keywords in the natural language; and a matching unit, configured to obtain a dialogue context through the keyword matching, wherein the dialogue context includes the following One: Knowledge search, Internet search, weather query, route navigation, dialogue chat, command control.
  • the determining module includes: a searching unit, configured to find a historical conversation record before the natural language; and a determining unit, configured to determine the dialogue context based on the historical conversation record.
  • the device further includes: a setting module for setting the corpus database before the recognition module uses the corpus database corresponding to the dialogue context to recognize the natural language, wherein the corpus database Including: multi-round dialogue data, multi-scene corpus, Internet corpus obtained through solr search engine.
  • the recognition module includes: a query unit for querying a local corpus database corresponding to the dialogue context, wherein the local corpus database includes: multiple rounds of dialogue data, multiple scene corpus; a first recognition unit , Configured to use the local corpus database to recognize the natural language locally; a second recognition unit, configured to call the solr search engine to recognize the natural language when the local corpus database fails to recognize the natural language locally Natural language, and adding a recognition result corresponding to the natural language to the local corpus database.
  • the collection module includes: a collection unit for collecting background noise voice signals through a microphone; an analysis unit for analyzing the background noise and sound ray characteristics of the voice signal; and a noise reduction unit for The background noise is removed according to the sound ray characteristics of the historical speech, and the speech signal is converted into text to obtain the natural language, wherein the historical speech is a sample speech collected under a noise-free condition.
  • the device further includes: a processing unit, configured to output response content corresponding to the recognition result after the recognition module uses the corpus database corresponding to the dialogue context to recognize the natural language, or generate The control command corresponding to the recognition result.
  • a processing unit configured to output response content corresponding to the recognition result after the recognition module uses the corpus database corresponding to the dialogue context to recognize the natural language, or generate The control command corresponding to the recognition result.
  • each of the above modules can be implemented by software or hardware.
  • it can be implemented in the following manner, but not limited to this: the above modules are all located in the same processor; or, the above modules are combined in any combination The forms are located in different processors.
  • the disclosed system, device, and method may be implemented in other ways.
  • the device embodiments described above are merely illustrative, for example, the division of the units is only a logical function division, and there may be other divisions in actual implementation, for example, multiple units or components may be combined Or it can be integrated into another system, or some features can be ignored or not implemented.
  • the displayed or discussed mutual coupling or direct coupling or communication connection may be indirect coupling or communication connection through some interfaces, devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the objectives of the solutions of the embodiments.
  • each unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units may be integrated into one unit.
  • the above-mentioned integrated unit may be implemented in the form of hardware, or may be implemented in the form of hardware plus software functional units.
  • the above-mentioned integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium.
  • the above-mentioned software functional unit is stored in a storage medium and includes several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (Processor) execute the method described in each embodiment of the present application Part of the steps.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disk and other media that can store program code .
  • the embodiment of the present application also provides a storage medium in which a computer program is stored, wherein the computer program is configured to execute the steps in any of the foregoing method embodiments when running.
  • the foregoing storage medium may be configured to store a computer program for executing the following steps:
  • the foregoing storage medium may include, but is not limited to: U disk, Read-Only Memory (Read-Only Memory, ROM for short), Random Access Memory (Random Access Memory, RAM for short), Various media that can store computer programs, such as mobile hard disks, magnetic disks, or optical disks.
  • the embodiment of the present application also provides an electronic device, including a memory and a processor, the memory is stored with a computer program, and the processor is configured to run the computer program to execute the steps in any of the foregoing method embodiments.
  • the aforementioned electronic device may further include a transmission device and an input-output device, wherein the transmission device is connected to the aforementioned processor, and the input-output device is connected to the aforementioned processor.
  • the foregoing processor may be configured to execute the following steps through a computer program:

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Acoustics & Sound (AREA)
  • Human Computer Interaction (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Theoretical Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Signal Processing (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Machine Translation (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • User Interface Of Digital Computer (AREA)

Abstract

本发明实施例提供了一种语音的识别方法及装置、计算机设备、存储介质。一方面,该方法包括:采集自然语言;确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;使用与所述对话语境对应的语料数据库识别所述自然语言。通过本发明,解决了现有技术中语音效率低的技术问题,语音识别更智能,提高了语音识别率。

Description

语音的识别方法及装置、计算机设备、存储介质 【技术领域】
本申请涉及计算机领域,尤其涉及一种语音的识别方法及装置、计算机设备、存储介质。
【背景技术】
聊天机器人是人工智能中的重要领域,通过用户向聊天机器人发出自然语言,机器人可以完成一系列的操作。
相关技术中,聊天机器人只能做一些简单的闲聊对话,很少有能够分辨出人的真实意图,并且不能连续的针对某一个素材有限的语音或者结合上下文来理解对话意图,识别率低,需要获取大量的自然语言才能提高识别率,导致语音识别的应用范围一直不高。
针对相关技术中存在的上述问题,目前尚未发现有效的解决方案。
【发明内容】
有鉴于此,本申请实施例提供了一种语音的识别方法及装置、计算机设备、存储介质。
一方面,本申请实施例提供了一种语音的识别方法,所述方法包括:采集自然语言;确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;使用与所述对话语境对应的语料数据库识别所述自然语言。
另一方面,本申请实施例提供了一种语音的识别装置,所述装置包括:采集模块,用于采集自然语言;确定模块,用于确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;识别模 块,用于使用与所述对话语境对应的语料数据库识别所述自然语言。
根据本申请的又一个实施例,还提供了一种存储介质,所述存储介质中存储有计算机程序,其中,所述计算机程序被设置为运行时执行上述任一项方法实施例中的步骤。
根据本申请的又一个实施例,还提供了一种电子装置,包括存储器和处理器,所述存储器中存储有计算机程序,所述处理器被设置为运行所述计算机程序以执行上述任一项方法实施例中的步骤。
通过本申请,通过先确定对话语境和场景,进而在对应的语料数据库中识别自然语言,解决了现有技术中语音效率低的技术问题,语音识别更智能,提高了语音识别率。
【附图说明】
为了更清楚地说明本申请实施例的技术方案,下面将对实施例中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其它的附图。
图1是本申请实施例的一种语音的识别移动终端的硬件结构框图;
图2是根据本申请实施例的语音的识别方法的流程图;
图3是本申请实施例中识别自然语言的流程图;
图4是本申请实施例交互示意图;
图5是根据本申请实施例的语音的识别装置的结构框图。
【具体实施方式】
下文中将参考附图并结合实施例来详细说明本申请。需要说明的是,在不冲突的情况下,本申请中的实施例及实施例中的特征可以相互组合。
需要说明的是,本申请的说明书和权利要求书及上述附图中的术语“第 一”、“第二”等是用于区别类似的对象,而不必用于描述特定的顺序或先后次序。
实施例1
本申请实施例一所提供的方法实施例可以在移动终端、中控设备、计算机终端或者类似的运算装置中执行。以运行在移动终端上为例,图1是本申请实施例的一种语音的识别移动终端的硬件结构框图。如图1所示,移动终端10可以包括一个或多个(图1中仅示出一个)处理器102(处理器102可以包括但不限于微处理器MCU或可编程逻辑器件FPGA等的处理装置)和用于存储数据的存储器104,可选地,上述移动终端还可以包括用于通信功能的传输设备106以及输入输出设备108。本领域普通技术人员可以理解,图1所示的结构仅为示意,其并不对上述移动终端的结构造成限定。例如,移动终端10还可包括比图1中所示更多或者更少的组件,或者具有与图1所示不同的配置。
存储器104可用于存储计算机程序,例如,应用软件的软件程序以及模块,如本申请实施例中的语音的识别方法对应的计算机程序,处理器102通过运行存储在存储器104内的计算机程序,从而执行各种功能应用以及数据处理,即实现上述的方法。存储器104可包括高速随机存储器,还可包括非易失性存储器,如一个或者多个磁性存储装置、闪存、或者其他非易失性固态存储器。在一些实例中,存储器104可进一步包括相对于处理器102远程设置的存储器,这些远程存储器可以通过网络连接至移动终端10。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
传输装置106用于经由一个网络接收或者发送数据。上述的网络具体实例可包括移动终端10的通信供应商提供的无线网络。在一个实例中,传输装置106包括一个网络适配器(Network Interface Controller,简称为NIC),其可通过基站与其他网络设备相连从而可与互联网进行通讯。在一个实例中, 传输装置106可以为射频(Radio Frequency,简称为RF)模块,其用于通过无线方式与互联网进行通讯。
在本实施例中提供了一种语音的识别方法,图2是根据本申请实施例的语音的识别方法的流程图,如图2所示,该流程包括如下步骤:
步骤S202,采集自然语言;
步骤S204,确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;
在本实施例中,语境交互的领域为交互场景,如天气查询,路线导航,对话聊天等,交互对象为通过语音控制的目标对象,如目标对象为系统的查询引擎,通过查询引擎查询与自然语音对应的识别结果,或者是根据自然语音中的指令信息去,还包括调用引擎,通过调用引擎来调用设备的软件或硬件,以辅助识别,如调用内置的GPS模块等,还包括控制对象,如通过中控台控制电视机等与中控台网络连接的设备时,电视机为控制对象。
步骤S206,使用与所述对话语境对应的语料数据库识别所述自然语言。
通过本实施例的方案,通过先确定对话语境和场景,进而在对应的语料数据库中识别自然语言,解决了现有技术中语音效率低的技术问题,语音识别更智能,提高了语音识别率。
本实施例的执行主体可以是聊天机器人,语音控制台,语音中控设备,带有语音控制功能的电子设备等,但不限于此。
本实施例中的对话语境包括对话场景,机器交互参与的对象和设备等,确定对话语境包括多种方案。
在一个示例中,确定所述自然语言的对话语境包括:
S11,提取所述自然语言中的关键字;
本实施例中的自然语言为自然人或者设备发出的语音,在提取关键字时,先将语音转换为文字,然后使用预设的关键字列表,在文字中查找是否包括 关键字列表中的任一关键字,如果包括则提取匹配的关键字。
S12,通过所述关键字匹配得到对话语境,其中,所述对话语境包括以下之一:知识搜索,互联网搜索,天气查询,路线导航,对话聊天,指令控制。
本实施例的关键字和对话语境通过预设映射表来匹配,如关键字集合{a,b,c}对应知识搜索的对话语境,关键字集合{b,c,d}对应互联网搜索的对话语境等。
在一些语境下,需要启动第三方软件(如地图,浏览器等),在一些语境下,需要用到本地知识图谱,在一些语境下,用到语料数据库。如路线导航的语境中,需要启动地图软件,如采集到的自然语言为“回家”,确定语境为路线导航,需要调用地图软件,搜索从当前位置到家的行车导航路线;或者在知识搜索的语境中,如采集到的自然语言为“张三的联系方式”,确定语境为知识搜索,在本地的通讯录中搜索出张三的联系方式等。
在另一个示例中,确定所述自然语言的对话语境包括:
S21,查找所述自然语言之前的历史对话记录;
本实施例的历史对话记录为当前自然语言之前的语音识别设备与语音识别对象之间的对话内容,该对话记录保存在语音设备的本地存储器,在查找历史对话记录时,可以根据时间作为参数,查找预设时间范围之内的对话记录,例如,用户与语音机器人进行人机对话,在1分钟之内产生了三条对话记录,分别为:A,B,C,语音机器人当前接收到的待识别的自然语音为D,则可以根据A,B,C的内容来确定D的对话场景,如果D中的内容与A,B,C有关联性,则对话语境也是相同的,这是由于一次正常的人机对话都会包括多条对话记录。
S22,根据所述历史对话记录确定所述对话语境。
在本示例的方案中,通过当前自然语言的上下文(主要是上文)来确定对话场景,匹配该场景领域的语料,进而识别:在AIML机制下,记录对话 中的上下文信息,基于上下文信息,通过配置好的话题库来匹配识别某一领域类的语料对话,进而使用该领域的语料进行识别。如:接收到用户的自然语言“太暗了”,通过之前的聊天记录,确定对话场景为“电视机控制”,进而在电视机领域的语料中识别出“背景太暗了”,进而调亮背景亮度,或者确定对话场景为“灯光控制”,进而在灯光照明领域的语料中识别出“照明设备太暗了”,进而打开照明设备。
在本实施例中,在使用与所述对话语境对应的语料数据库识别所述自然语言之前,还包括:设置所述语料数据库,其中,所述语料数据库包括:多轮对话语料,多场景语料,通过solr(Search On Lucene Replication)搜索引擎得到的网络语料。本实施例中的上述三种语料库是三个维度的数据库,可以单独使用,也可以相互配合使用,其中,多轮对话语料是指包括多组对话的语料库,可用于聊天,本地查询等,如饮食、电视、电影、音乐、工作等多个主题(每个类别包括多组逻辑对话),多场景语料为在多个特定场景下的语料库,可用于聊天,本地查询,机器控制等,如在饮食相关的领域和在音乐相关的领域,对于同样的问题:“请问人气高的地方”,则应答会有所差异,反馈的是分别是餐厅和音乐厅,网络语料则是以网络查询的结果作为语料。
可选的,采用AIML模型进行配置语料数据库,AIML人工智能标记语言,AIML描述一种称为AIML对象的数据对象的类(class),并部分描述了处理他们的计算机程序的行为(behavior)。AIML对象是由所谓的主题(topics)和类型(categories)的单元组成,其中包含任何解析或未解析的数据。其中,解析数据(Parsed data)是字符解析得到的字符数据(character data)、和AIML元素(AIML elements)。AIML元素封装在包含在文档中的刺激-反应知识(stimulus-response knowledge),在这些元素中的字符数据(Character data)可以由AIML编译器(AIML interpreter)解析,也可以由响应器(Responder) 解析。本实施例的Solr搜索引擎输入的是自然语言,输出对应的回答也是自然语言。在一个实施方式中,通过配置相应的问题模板来配置语料数据库,模板形式采用AIML模板形式进行编写,多组预设问题的回答语句通过网上爬取获得,如百度百科等,并根据问题的类型与相应的网址相链接。
图3是本申请实施例中识别自然语言的流程图,如图3所示,使用与所述对话语境对应的语料数据库识别所述自然语言包括:
S302,查询与所述对话语境对应的本地语料数据库,其中,所述本地语料数据库包括:多轮对话语料,多场景语料;
本实施例中的对话语境包括知识搜索,互联网搜索,天气查询,路线导航,对话聊天,指令控制等,其中,对话聊天为多轮对话语料,知识搜索,互联网搜索,天气查询,路线导航,指令控制为多场景语料,在多场景对话语料中,每一个对话语境的场景都对应一个专用的语料数据库,但是在多场景语料识别失败时,会转为多轮对话语料的语料库,进一步询问自然语言的实际含义。在一个示例中,采集到的自然语言为:“导航到前面写字楼”,确定为路线导航的对话语境,在本地语料数据库中查询“前面写字楼”没有成功,进而转到多轮对话语料,触发问题:“请问是哪一栋写字楼?”,用户回答:“A写字楼”。
S304,使用所述本地语料数据库在本地识别所述自然语言;
可选的,在使用与所述对话语境对应的语料数据库识别所述自然语言之后,还包括:输出与识别结果对应的应答内容,或,生成与识别结果对应的控制指令。
将自然语言转换为语料数据库可以识别的机器语言后,在本地语料数据库查找对应的语义,在查找失败时,使用solr搜索引擎进行搜索,识别出语义后,进一步搜索出该语义下的应答内容,或者是针对目标对象的控制指令,如采集到自然语言为“今天的天气怎样”,可以搜索出天气状况,并输出应 答内容“今天天气晴,24℃,空气指数良好”,或者采集到的语言为“电视声音太大了”,匹配出用于调低电视音量的控制指令,并发送给电视的控制台。如果是在语料数据库识别出来的,则在预料数据库查找,如果是通过搜索引擎识别出来的,则通过搜索引擎查找。当然,也可以使用独立的数据库(该数据库保存了与每个语义对应的应答内容或控制指令)。图4是本申请实施例交互示意图,本地内置了互联网搜索,天气查询,路线导航三个语境的语料数据库。
S306,在使用所述本地语料数据库在本地识别所述自然语言失败时,调用所述solr搜索引擎识别所述自然语言,并在所述本地语料数据库中添加与所述自然语言对应的识别结果。
在使用solr搜索引擎也识别失败时,在采用所有语料数据库识别用户的自然语言都失败时,反馈错误,由用户来协助完成对话或操作,并进一步完善语料数据库,如将结果反馈到语料数据库,在本地语料库中修改或添加当前问答的识别内容,学习识别失败的自然语言。
在本实施例的实施方式中,采集自然语言包括:通过麦克风采集背景噪声的语音信号;解析所述背景噪声和所述语音信号的声线特性;根据历史语音的声线特性去掉所述背景噪声,将所述语音信号转换为文字,得到所述自然语言,其中,所述历史语音为在无噪音条件下采集的样本语音。
在本实施例的方案在应用在聊天机器人上时,基于国际机器人的通用语言AIML的模型进行优化,由于AIML的标签配置非常灵活,可以支持自定义场景的配置,多轮对话语料,应用于多种场景,可以让机器人拥有更智能化,更专业化的对话功能;后台增加solr搜索引擎可以对接更为丰富的语料资源,并且适应基于语义的快速搜索和对话功能;机器人具备学习功能,可以在与人的对话中学习之前不能理解的语句与对答,拥有自我成长功能。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到根据 上述实施例的方法可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。基于这样的理解,本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中,包括若干指令用以使得一台终端设备(可以是手机,计算机,服务器,或者网络设备等)执行本申请各个实施例所述的方法。
实施例2
在本实施例中还提供了一种语音的识别装置,该装置用于实现上述实施例及优选实施方式,已经进行过说明的不再赘述。如以下所使用的,术语“模块”可以实现预定功能的软件和/或硬件的组合。尽管以下实施例所描述的装置较佳地以软件来实现,但是硬件,或者软件和硬件的组合的实现也是可能并被构想的。
图5是根据本申请实施例的语音的识别装置的结构框图,如图5所示,该装置包括:采集模块50,确定模块52,识别模块54,其中,
采集模块50,用于采集自然语言;
确定模块52,用于确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;
识别模块54,用于使用与所述对话语境对应的语料数据库识别所述自然语言。
可选的,所述确定模块包括:提取单元,用于提取所述自然语言中的关键字;匹配单元,用于通过所述关键字匹配得到对话语境,其中,所述对话语境包括以下之一:知识搜索,互联网搜索,天气查询,路线导航,对话聊天,指令控制。
可选的,所述确定模块包括:查找单元,用于查找所述自然语言之前的历史对话记录;确定单元,用于根据所述历史对话记录确定所述对话语境。
可选的,所述装置还包括:设置模块,用于在所述识别模块使用与所述对话语境对应的语料数据库识别所述自然语言之前,设置所述语料数据库,其中,所述语料数据库包括:多轮对话语料,多场景语料,通过solr搜索引擎得到的网络语料。
可选的,所述识别模块包括:查询单元,用于查询与所述对话语境对应的本地语料数据库,其中,所述本地语料数据库包括:多轮对话语料,多场景语料;第一识别单元,用于使用所述本地语料数据库在本地识别所述自然语言;第二识别单元,用于在使用所述本地语料数据库在本地识别所述自然语言失败时,调用所述solr搜索引擎识别所述自然语言,并在所述本地语料数据库中添加与所述自然语言对应的识别结果。
可选的,所述采集模块包括:采集单元,用于通过麦克风采集背景噪声的语音信号;解析单元,用于解析所述背景噪声和所述语音信号的声线特性;降噪单元,用于根据历史语音的声线特性去掉所述背景噪声,将所述语音信号转换为文字,得到所述自然语言,其中,所述历史语音为在无噪音条件下采集的样本语音。
可选的,所述装置还包括:处理单元,用于在所述识别模块使用与所述对话语境对应的语料数据库识别所述自然语言之后,输出与识别结果对应的应答内容,或,生成与识别结果对应的控制指令。
需要说明的是,上述各个模块是可以通过软件或硬件来实现的,对于后者,可以通过以下方式实现,但不限于此:上述模块均位于同一处理器中;或者,上述各个模块以任意组合的形式分别位于不同的处理器中。
实施例3
在本申请所提供的几个实施例中,应该理解到,所揭露的系统,装置和方法,可以通过其它的方式实现。例如,以上所描述的装置实施例仅仅是示意性的,例如,所述单元的划分,仅仅为一种逻辑功能划分,实际实现时可 以有另外的划分方式,例如,多个单元或组件可以结合或者可以集成到另一个系统,或一些特征可以忽略,或不执行。另一点,所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口,装置或单元的间接耦合或通信连接,可以是电性,机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外,在本申请各个实施例中的各功能单元可以集成在一个处理单元中,也可以是各个单元单独物理存在,也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现,也可以采用硬件加软件功能单元的形式实现。
上述以软件功能单元的形式实现的集成的单元,可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中,包括若干指令用以使得一台计算机装置(可以是个人计算机,服务器,或者网络装置等)或处理器(Processor)执行本申请各个实施例所述方法的部分步骤。而前述的存储介质包括:U盘、移动硬盘、只读存储器(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory,RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
本申请的实施例还提供了一种存储介质,该存储介质中存储有计算机程序,其中,该计算机程序被设置为运行时执行上述任一项方法实施例中的步骤。
可选地,在本实施例中,上述存储介质可以被设置为存储用于执行以下步骤的计算机程序:
S1,采集自然语言;
S2,确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;
S3,使用与所述对话语境对应的语料数据库识别所述自然语言。
可选地,在本实施例中,上述存储介质可以包括但不限于:U盘、只读存储器(Read-Only Memory,简称为ROM)、随机存取存储器(Random Access Memory,简称为RAM)、移动硬盘、磁碟或者光盘等各种可以存储计算机程序的介质。
本申请的实施例还提供了一种电子装置,包括存储器和处理器,该存储器中存储有计算机程序,该处理器被设置为运行计算机程序以执行上述任一项方法实施例中的步骤。
可选地,上述电子装置还可以包括传输设备以及输入输出设备,其中,该传输设备和上述处理器连接,该输入输出设备和上述处理器连接。
可选地,在本实施例中,上述处理器可以被设置为通过计算机程序执行以下步骤:
S1,采集自然语言;
S2,确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;
S3,使用与所述对话语境对应的语料数据库识别所述自然语言。
以上所述仅为本申请的较佳实施例而已,并不用以限制本申请,凡在本申请的精神和原则之内,所做的任何修改、等同替换、改进等,均应包含在本申请保护的范围之内。

Claims (20)

  1. 一种语音的识别方法,其中,所述方法包括:
    采集自然语言;
    确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;
    使用与所述对话语境对应的语料数据库识别所述自然语言;
    其中,在使用与所述对话语境对应的语料数据库识别所述自然语言之前,还包括:设置所述语料数据库,其中,所述语料数据库包括:多轮对话语料,多场景语料,通过solr搜索引擎得到的网络语料;
    其中,使用与所述对话语境对应的语料数据库识别所述自然语言包括:查询与所述对话语境对应的本地语料数据库,其中,所述本地语料数据库包括:多轮对话语料,多场景语料;使用所述本地语料数据库在本地识别所述自然语言;在使用所述本地语料数据库在本地识别所述自然语言失败时,调用所述solr搜索引擎识别所述自然语言,并在所述本地语料数据库中添加与所述自然语言对应的识别结果。
  2. 根据权利要求1所述的方法,确定所述自然语言的对话语境包括:
    提取所述自然语言中的关键字;
    通过所述关键字匹配得到对话语境,其中,所述对话语境包括以下之一:知识搜索,互联网搜索,天气查询,路线导航,对话聊天,指令控制。
  3. 根据权利要求1所述的方法,确定所述自然语言的对话语境包括:
    查找所述自然语言之前的历史对话记录;
    根据所述历史对话记录确定所述对话语境。
  4. 根据权利要求1所述的方法,采集自然语言包括:
    通过麦克风采集背景噪声的语音信号;
    解析所述背景噪声和所述语音信号的声线特性;
    根据历史语音的声线特性去掉所述背景噪声,将所述语音信号转换为文字,得到所述自然语言,其中,所述历史语音为在无噪音条件下采集的样本语音。
  5. 根据权利要求1所述的方法,在使用与所述对话语境对应的语料数据库识别所述自然语言之后,所述方法还包括:
    输出与识别结果对应的应答内容,或,生成与识别结果对应的控制指令。
  6. 根据权利要求5所述的方法,在生成与识别结果对应的控制指令之后,所述方法还包括:
    建立与控制对象的通信链路;
    通过所述通信链路向所述控制对象发送所述控制指令。
  7. 根据权利要求1所述的方法,在所述solr搜索引擎识别所述自然语言失败时,所述方法还包括:
    反馈识别错误的指示信息,并获取用户输入的协助识别结果;
    将所述协助识别结果反馈到语料数据库,在本地语料库中修改或添加当前问答的识别内容。
  8. 一种语音的识别装置,其中,所述装置包括:
    采集模块,用于采集自然语言;
    确定模块,用于确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;
    识别模块,用于使用与所述对话语境对应的语料数据库识别所述自然语言。
  9. 根据权利要求8所述的装置,所述确定模块包括:
    提取单元,用于提取所述自然语言中的关键字;
    匹配单元,用于通过所述关键字匹配得到对话语境,其中,所述对话语境包括以下之一:知识搜索,互联网搜索,天气查询,路线导航,对话聊天,指令控制;
    其中,所述装置还包括:设置模块,用于在所述识别模块使用与所述对 话语境对应的语料数据库识别所述自然语言之前,设置所述语料数据库,其中,所述语料数据库包括:多轮对话语料,多场景语料,通过solr搜索引擎得到的网络语料。
    其中,所述识别模块包括:查询单元,用于查询与所述对话语境对应的本地语料数据库,其中,所述本地语料数据库包括:多轮对话语料,多场景语料;第一识别单元,用于使用所述本地语料数据库在本地识别所述自然语言;第二识别单元,用于在使用所述本地语料数据库在本地识别所述自然语言失败时,调用所述solr搜索引擎识别所述自然语言,并在所述本地语料数据库中添加与所述自然语言对应的识别结果。
  10. 根据权利要求8所述的装置,所述确定模块包括:
    查找单元,用于查找所述自然语言之前的历史对话记录;
    确定单元,用于根据所述历史对话记录确定所述对话语境。
  11. 根据权利要求8所述的装置,所述装置还包括:
    建立模块,用于在所述识别模块生成与识别结果对应的控制指令之后,建立与控制对象的通信链路;
    发送模块,用于通过所述通信链路向所述控制对象发送所述控制指令。
  12. 根据权利要求8所述的装置,所述识别模块包括:
    处理模块,用于在所述识别模块通过所述solr搜索引擎识别所述自然语言失败时,反馈识别错误的指示信息,并获取用户输入的协助识别结果;
    学习模块,用于将所述协助识别结果反馈到语料数据库,在本地语料库中修改或添加当前问答的识别内容。
  13. 根据权利要求8所述的装置,所述采集模块包括:
    采集单元,用于通过麦克风采集背景噪声的语音信号;
    解析单元,用于解析所述背景噪声和所述语音信号的声线特性;
    降噪单元,用于根据历史语音的声线特性去掉所述背景噪声,将所述语 音信号转换为文字,得到所述自然语言,其中,所述历史语音为在无噪音条件下采集的样本语音。
  14. 根据权利要求8所述的装置,所述装置还包括:
    处理单元,用于在所述识别模块使用与所述对话语境对应的语料数据库识别所述自然语言之后,输出与识别结果对应的应答内容,或,生成与识别结果对应的控制指令。
  15. 一种计算机设备,包括存储器和处理器,所述存储器存储有计算机程序,所述处理器执行所述计算机程序时实现一种语音的识别方法的步骤,包括:
    采集自然语言;
    确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;
    使用与所述对话语境对应的语料数据库识别所述自然语言;
    其中,在使用与所述对话语境对应的语料数据库识别所述自然语言之前,还包括:设置所述语料数据库,其中,所述语料数据库包括:多轮对话语料,多场景语料,通过solr搜索引擎得到的网络语料;
    其中,使用与所述对话语境对应的语料数据库识别所述自然语言包括:查询与所述对话语境对应的本地语料数据库,其中,所述本地语料数据库包括:多轮对话语料,多场景语料;使用所述本地语料数据库在本地识别所述自然语言;在使用所述本地语料数据库在本地识别所述自然语言失败时,调用所述solr搜索引擎识别所述自然语言,并在所述本地语料数据库中添加与所述自然语言对应的识别结果。
  16. 根据权利要求15所述的计算机设备,确定所述自然语言的对话语境包括:
    提取所述自然语言中的关键字;
    通过所述关键字匹配得到对话语境,其中,所述对话语境包括以下之一:知识搜索,互联网搜索,天气查询,路线导航,对话聊天,指令控制。
  17. 根据权利要求15所述的计算机设备,确定所述自然语言的对话语境包括:
    查找所述自然语言之前的历史对话记录;
    根据所述历史对话记录确定所述对话语境。
  18. 一种计算机存储介质,其上存储有计算机程序,所述计算机程序被处理器执行时实现一种语音的识别方法的步骤,包括:
    采集自然语言;
    确定所述自然语言的对话语境,其中,所述对话语境用于描述当前语音交互的领域和交互对象;
    使用与所述对话语境对应的语料数据库识别所述自然语言;
    其中,在使用与所述对话语境对应的语料数据库识别所述自然语言之前,还包括:设置所述语料数据库,其中,所述语料数据库包括:多轮对话语料,多场景语料,通过solr搜索引擎得到的网络语料;
    其中,使用与所述对话语境对应的语料数据库识别所述自然语言包括:查询与所述对话语境对应的本地语料数据库,其中,所述本地语料数据库包括:多轮对话语料,多场景语料;使用所述本地语料数据库在本地识别所述自然语言;在使用所述本地语料数据库在本地识别所述自然语言失败时,调用所述solr搜索引擎识别所述自然语言,并在所述本地语料数据库中添加与所述自然语言对应的识别结果。
  19. 根据权利要求18所述的计算机存储介质,确定所述自然语言的对话语境包括:
    提取所述自然语言中的关键字;
    通过所述关键字匹配得到对话语境,其中,所述对话语境包括以下之一:知识搜索,互联网搜索,天气查询,路线导航,对话聊天,指令控制。
  20. 根据权利要求18所述的计算机存储介质,确定所述自然语言的对话语境包括:
    查找所述自然语言之前的历史对话记录;
    根据所述历史对话记录确定所述对话语境。
PCT/CN2019/118415 2019-06-20 2019-11-14 语音的识别方法及装置、计算机设备、存储介质 Ceased WO2020253064A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910534912.X 2019-06-20
CN201910534912.XA CN110265013A (zh) 2019-06-20 2019-06-20 语音的识别方法及装置、计算机设备、存储介质

Publications (1)

Publication Number Publication Date
WO2020253064A1 true WO2020253064A1 (zh) 2020-12-24

Family

ID=67919649

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/118415 Ceased WO2020253064A1 (zh) 2019-06-20 2019-11-14 语音的识别方法及装置、计算机设备、存储介质

Country Status (2)

Country Link
CN (1) CN110265013A (zh)
WO (1) WO2020253064A1 (zh)

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110265013A (zh) * 2019-06-20 2019-09-20 平安科技(深圳)有限公司 语音的识别方法及装置、计算机设备、存储介质
CN110809095A (zh) * 2019-10-25 2020-02-18 大唐网络有限公司 语音外呼的方法和装置
CN111210816B (zh) * 2019-12-16 2023-09-05 北京淇瑀信息科技有限公司 一种智能语音配置管理方法、装置及电子设备
CN113241066B (zh) * 2020-01-22 2022-04-22 广州汽车集团股份有限公司 语音交互方法及其系统、语音交互设备
CN113241067B (zh) * 2020-01-22 2022-04-22 广州汽车集团股份有限公司 一种语音交互方法及其系统、语音交互设备
CN113808593B (zh) * 2020-06-16 2025-06-03 阿里巴巴集团控股有限公司 语音交互系统、相关方法、装置及设备
CN112000787B (zh) * 2020-08-17 2021-05-14 上海小鹏汽车科技有限公司 语音交互方法、服务器和语音交互系统
CN114678012B (zh) * 2022-02-18 2024-10-22 青岛海尔科技有限公司 语音交互数据的处理方法和装置、存储介质及电子装置
CN114840671B (zh) * 2022-04-29 2025-02-18 北京百度网讯科技有限公司 对话生成方法、模型的训练方法、装置、设备及介质
CN115294988B (zh) * 2022-07-20 2025-01-24 北方民族大学 一种面向协作的语音交互系统和方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150019221A1 (en) * 2013-07-15 2015-01-15 Chunghwa Picture Tubes, Ltd. Speech recognition system and method
CN106201424A (zh) * 2016-07-08 2016-12-07 北京甘为乐博科技有限公司 一种信息交互方法、装置及电子设备
CN106297782A (zh) * 2016-07-28 2017-01-04 北京智能管家科技有限公司 一种人机交互方法及系统
CN109299320A (zh) * 2018-10-30 2019-02-01 上海智臻智能网络科技股份有限公司 一种信息交互方法、装置、计算机设备和存储介质
CN109902158A (zh) * 2019-01-24 2019-06-18 平安科技(深圳)有限公司 语音交互方法、装置、计算机设备及存储介质
CN110265013A (zh) * 2019-06-20 2019-09-20 平安科技(深圳)有限公司 语音的识别方法及装置、计算机设备、存储介质

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6499013B1 (en) * 1998-09-09 2002-12-24 One Voice Technologies, Inc. Interactive user interface using speech recognition and natural language processing
CN107146622B (zh) * 2017-06-16 2021-02-19 合肥美的智能科技有限公司 冰箱、语音交互系统、方法、计算机设备、可读存储介质
CN108304561B (zh) * 2018-02-08 2019-03-29 北京信息职业技术学院 一种基于有限数据的语义理解方法、设备及机器人

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150019221A1 (en) * 2013-07-15 2015-01-15 Chunghwa Picture Tubes, Ltd. Speech recognition system and method
CN106201424A (zh) * 2016-07-08 2016-12-07 北京甘为乐博科技有限公司 一种信息交互方法、装置及电子设备
CN106297782A (zh) * 2016-07-28 2017-01-04 北京智能管家科技有限公司 一种人机交互方法及系统
CN109299320A (zh) * 2018-10-30 2019-02-01 上海智臻智能网络科技股份有限公司 一种信息交互方法、装置、计算机设备和存储介质
CN109902158A (zh) * 2019-01-24 2019-06-18 平安科技(深圳)有限公司 语音交互方法、装置、计算机设备及存储介质
CN110265013A (zh) * 2019-06-20 2019-09-20 平安科技(深圳)有限公司 语音的识别方法及装置、计算机设备、存储介质

Also Published As

Publication number Publication date
CN110265013A (zh) 2019-09-20

Similar Documents

Publication Publication Date Title
WO2020253064A1 (zh) 语音的识别方法及装置、计算机设备、存储介质
CN110334347B (zh) 基于自然语言识别的信息处理方法、相关设备及存储介质
US10937413B2 (en) Techniques for model training for voice features
CN112084315B (zh) 问答互动方法、装置、存储介质及设备
CN116521841B (zh) 用于生成回复信息的方法、装置、设备及介质
JP6730994B2 (ja) 問答情報の処理方法、装置、記憶媒体及び装置
WO2021073298A1 (zh) 一种语音信息的处理方法、装置、智能终端以及存储介质
WO2018224034A1 (zh) 智能问答方法、服务器、终端及存储介质
CN109309751B (zh) 语音记录方法、电子设备及存储介质
CN112417158A (zh) 文本数据分类模型的训练方法、分类方法、装置和设备
CN111427444B (zh) 一种智能设备的控制方法和设备
CN111178081B (zh) 语义识别的方法、服务器、电子设备及计算机存储介质
CN110399470B (zh) 会话消息处理
CN112364622B (zh) 对话文本分析方法、装置、电子装置及存储介质
WO2021139243A1 (zh) 基于人机交互的数据处理方法、装置、设备及存储介质
US10854189B2 (en) Techniques for model training for voice features
CN117312521A (zh) 用于智能客服对话的处理方法及相关产品
CN110992937B (zh) 语言离线识别方法、终端及可读存储介质
CN109271503A (zh) 智能问答方法、装置、设备及存储介质
CN113763929A (zh) 一种语音评测方法、装置、电子设备和存储介质
CN118838998A (zh) 一种人机交互方法、装置及计算机可读存储介质
CN114461749B (zh) 对话内容的数据处理方法、装置、电子设备和介质
CN117194620A (zh) 信息处理方法、装置、设备及存储介质
CN117351952A (zh) 语音指令交互方法、装置、电子设备及存储介质
CN117332062A (zh) 一种数据处理方法和相关装置

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: 19933435

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19933435

Country of ref document: EP

Kind code of ref document: A1