CN108663942A

CN108663942A - A kind of speech recognition apparatus control method, speech recognition apparatus and control server

Info

Publication number: CN108663942A
Application number: CN201710214492.8A
Authority: CN
Inventors: 兰荣鑫; 孙运杰; 王庆玉
Original assignee: Qingdao Science And Technology Co Ltd
Current assignee: Qingdao Science And Technology Co Ltd
Priority date: 2017-04-01
Filing date: 2017-04-01
Publication date: 2018-10-16
Anticipated expiration: 2037-04-01
Also published as: CN108663942B

Abstract

The invention discloses a kind of speech recognition apparatus control methods, belong to field of speech recognition.This method includes：Receive the equipment control data that control server is sent；The equipment control data carries the control command for starting speech recognition；It is powered for speech recognition circuit.In addition, the present invention also provides a kind of speech recognition apparatus and control servers.Energy consumption can more effectively be saved using scheme provided by the invention, reduce the equipment loss that the speech recognition circuit of speech recognition apparatus is generated because being chronically at sleep state.

Description

A kind of speech recognition apparatus control method, speech recognition apparatus and control server

Technical field

The present invention relates to technical field of voice recognition, more particularly to a kind of speech recognition apparatus control method, speech recognition Equipment and control server.

Background technology

With stepping up for living standard, the development and use of robot technology are increasingly popularized, to society It can economic development generation significant impact.Robot technology is the inevitable outcome of the demand of productivity dynamics, is mankind itself's development Inevitable outcome.Voice-enabled chat and human-computer interaction are to embody a big branch of intellectual technology, and these technologies are mainly based on language Sound identification, semantic understanding and voice wake up this three gate technique.Chinese invention patent application CN106486122A discloses one kind Intelligent sound interacts robot, and when in sleep state, the sound sent out in real time to external environment is monitored for robot And judgement, if containing preset wake-up word from the speech samples that microphone obtains, robot is waken up, and is directly entered Sound identification module.Chinese invention patent application CN104660792A discloses a kind of method and device waking up application, at this In method, speech recognition apparatus needs to monitor the equipment state of mobile terminal in real time in sleep procedure, when monitoring described set When standby state is with setting state consistency, starts recording module, that is, enter sound identification module.In the above prior art, voice From after the power is turned on, the speech recognition circuit where speech identifying function is directly entered sleep state for identification equipment or robot, from sleeping For dormancy state to during this section being waken up into listening state, equipment is there is still a need for handling a large amount of information, including continues to monitor outer The sound of portion's environment or, continuing to monitor the equipment state etc. of mobile terminal, and according to the sound of the external environment monitored or is set Standby state, judges whether sleep state being switched to listening state.In above process, there are speech recognition apparatus or robots After the power is turned on, speech recognition circuit is chronically at sleep state, causes to consume additional electricity, occupies the problem of processor resource.

Invention content

An embodiment of the present invention provides a kind of speech recognition apparatus control methods, are known with solving to exist in the prior art voice The speech recognition circuit of other equipment or robot is chronically at sleep state, causes to consume additional electricity, occupies processor resource The problem of.In order to the embodiment to disclosure some aspects there are one basic understanding, simple summary is shown below.This is general It is not extensive overview to include part, nor to determine key/critical component or describe the protection domain of these embodiments.Its Sole purpose is that some concepts are presented with simple form, in this, as the preamble of following detailed description.

According to a first aspect of the embodiments of the present invention, a kind of speech recognition apparatus control method is provided, voice is applied to Identification equipment side.

In some exemplary embodiments, a kind of speech recognition apparatus control method, including：

Receive the equipment control data that control server is sent；The equipment control data is carried for starting speech recognition Control command；

It powers for the speech recognition circuit.

In some illustrative embodiments, it before the equipment control data that the reception control server is sent, also wraps It includes：

Connection request is sent to the control server；The device type that the connection request carries sets for speech recognition It is standby.

In some illustrative embodiments, it after the equipment control data that the reception control server is sent, also wraps It includes：

To the control server sending device state report request；The equipment state report request carries the voice The speech recognition working condition of identification equipment.

In some illustrative embodiments, after the transmission connection request to the control server, further include：

Keep-alive message request is sent to the control server；The device type that the keep-alive message request carries is voice Identification equipment.

In the above-described embodiments, speech recognition apparatus is not that the speech recognition circuit is powered immediately after the power is turned on, institute's predicate Sound identification circuit is in inactive state；The control for starting speech recognition is carried when the speech recognition apparatus receives After system instruction, just power to the speech recognition circuit.Said program solves the voice of speech recognition apparatus in the prior art Identification circuit is chronically at sleep state, causes to consume additional electricity, occupies the problem of processor resource.To sum up use this programme The embodiment described can more effectively save energy consumption, reduce the equipment loss because being chronically at sleep state generation.

According to a second aspect of the embodiments of the present invention, a kind of speech recognition apparatus control method is provided, middle control is applied to Server side.

To speech recognition apparatus sending device control data, the equipment control data is used to indicate the speech recognition and sets Standby to power for speech recognition circuit, the equipment control data carries the control command for starting speech recognition.

In some illustrative embodiments, before the sending device control data to speech recognition apparatus, further include：

Receive the connection request that the speech recognition apparatus is sent；The device type that the connection request carries is known for voice Other equipment.

In some illustrative embodiments, after the connection request for receiving the speech recognition apparatus transmission, also Including：

The keep-alive message that the speech recognition apparatus period sends is received to ask；The equipment that the keep-alive message request carries Type is speech recognition apparatus；The period for sending the protection message request is preset fixed duration.

In some illustrative embodiments, after the sending device control data to speech recognition apparatus, further include：

Receive the equipment state report request that the speech recognition apparatus is sent；The equipment state report request carries institute State the speech recognition working condition of speech recognition apparatus.

In the above-described embodiments, the control for starting speech recognition is carried by being sent to speech recognition apparatus After the equipment control data of instruction, so that the speech recognition apparatus powers to the speech recognition circuit.Said program solves The speech recognition circuit of speech recognition apparatus is chronically at sleep state in the prior art, causes to consume additional electricity, occupancy The problem of processor resource.It to sum up uses this programme the embodiment described that can more effectively save energy consumption, reduces because of long-term place In the equipment loss that sleep state generates.

According to a third aspect of the embodiments of the present invention, a kind of speech recognition apparatus is provided.

In some exemplary embodiments, the speech recognition apparatus, including：Processor and power supply module

The processor, the equipment control data for receiving control server transmission；The equipment control data carries Control command for starting speech recognition；And it controls the power supply module and powers for the speech recognition circuit.

According to the fourth aspect of the present invention, a kind of control server is provided.

In some exemplary embodiments, the control server, including：

Signal projector is used for speech recognition apparatus sending device control data；The equipment control data is for referring to Show that the speech recognition apparatus powers to speech recognition circuit, the equipment control data carries the control for starting speech recognition System order.

In the above-described embodiments, the control for starting speech recognition is carried by being sent to speech recognition apparatus After the equipment control data of instruction, so that the speech recognition apparatus powers to the speech recognition circuit, and by the voice The state of identification circuit is set as sleep state.Said program solves the speech recognition electricity of speech recognition apparatus in the prior art Road is chronically at sleep state, causes to consume additional electricity, occupies the problem of processor resource.It to sum up uses described in this programme Embodiment can more effectively save energy consumption, reduce the equipment loss because being chronically at sleep state generation.

It should be understood that above general description and following detailed description is only exemplary and explanatory, not It can the limitation present invention.

Description of the drawings

The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the present invention Example, and be used to explain the principle of the present invention together with specification.

Fig. 1 is a kind of flow diagram of speech recognition apparatus control method shown according to an exemplary embodiment；

Fig. 2 is a kind of flow diagram of speech recognition apparatus control method shown according to an exemplary embodiment；

Fig. 3 is a kind of flow diagram of speech recognition apparatus control method shown according to an exemplary embodiment；

Fig. 4 is a kind of block diagram of speech recognition apparatus shown according to an exemplary embodiment；

Fig. 5 is a kind of block diagram of speech recognition apparatus shown according to an exemplary embodiment；

Fig. 6 is a kind of block diagram of control server shown according to an exemplary embodiment.

Specific implementation mode

The following description and drawings fully show specific embodiments of the present invention, to enable those skilled in the art to Put into practice them.Embodiment only represents possible variation.Unless explicitly requested, otherwise individual components and functionality is optional, and And the sequence of operation can change.The part of some embodiments and feature can be included in or replace other embodiments Part and feature.The range of embodiment of the present invention includes the entire scope of claims and the institute of claims There is obtainable equivalent.Herein, each embodiment can individually or generally be indicated that this is only with term " invention " It is merely for convenience, and if in fact disclosing the invention more than one, be not meant to automatically limit the range of the application For any single invention or inventive concept.Herein, relational terms such as first and second and the like are used only for one Entity, which either operates to distinguish with another entity or operation, to be existed without requiring or implying between these entities or operation Any actual relationship or sequence.Moreover, the terms "include", "comprise" or its any other variant be intended to it is non-exclusive Property include, so that process, method or equipment including a series of elements include not only those elements, but also to include Each embodiment is described other elements that are not explicitly listed by the way of progressive herein, what each embodiment stressed All it is difference from other examples, just to refer each other for identical similar portion between each embodiment.For embodiment For disclosed structure, product etc., since it is corresponding with part disclosed in embodiment, so description is fairly simple, it is related Place is referring to method part illustration.

Speech recognition apparatus is the reception that can realize voice data, sends, knows to the voice messaging in voice data Not, and a kind of equipment of speech text can be converted voice data to.Voice messaging includes voice, semanteme, logic or even vocal print Information.Voiceprint includes one or more vocal print features.

Control server, is a kind of equipment with Central Control Function, the equipment can it is various to sound, light, electricity etc. by It controls equipment and carries out centralized control and unified management.Controlled plant can be projector, show stand, video disc player, video recorder, deck, Power amplifier, microphone, computer, notebook, electronic screen, electrically driven curtain, light etc..

Speech recognition circuit is in speech recognition apparatus, the voice data for the external environment to monitoring judged, The circuit of identification or analysis.For speech recognition circuit according to the difference of the state of speech identifying function module, also corresponding includes three kinds State is not actuated, sleep state and listening state respectively.

It is not actuated, refer to the state that speech recognition circuit is in power-off, does not power in this case to speech recognition circuit, Voice, which knows circuit, can not execute any operation.

Sleep state refers to a kind of low power consumpting state of the speech recognition circuit for opposite listening state, in this shape Under state, speech recognition circuit is carried out for receiving external voice data, and by external voice data and preset wake-up keyword Matching judges whether to enter listening state according to matching result.

Listening state refers to that speech recognition circuit is in normal working condition, and in this case, speech recognition circuit connects External voice data is identified and semantic analysis after receiving external voice data.

When the state of speech recognition circuit is not actuated, the then power supply module of speech recognition apparatus, i.e. power supply, not to voice Identification circuit is powered.When the state of speech recognition circuit is sleep state, then power supply module powers to speech recognition circuit, voice Identification circuit is in low power consumpting state, only receives the first voice data that microphone is monitored, and to the content of the first voice data Judged, it is determined whether set the state of speech recognition circuit to listening state.When the state of speech recognition circuit is to listen State, speech recognition circuit is listened to receive the second speech data that microphone is monitored, second speech data is converted to the second voice The speech text of data is sent to control server.When first voice data is that speech recognition circuit is in sleep state, Mike The voice data that wind is monitored.When second speech data is that speech recognition circuit is in listening state, the voice number of microphone monitoring According to.

Fig. 1 is a kind of flow diagram of speech recognition apparatus control method of the embodiment of the present invention.As shown in Figure 1：

Step S101 receives the equipment control data that control server is sent；Equipment control data is carried for starting language The control command of sound identification；

Step S102 powers for speech recognition circuit.

The method of above-described embodiment is applied to speech recognition apparatus, and speech recognition circuit is in power-off shape after device power State, i.e. speech recognition circuit are not actuated；After receiving the equipment control data that control server is sent, according to control command, to language Sound identification circuit is powered.

Optionally, in the above-described embodiments, after powering for speech recognition apparatus, the state of speech recognition circuit is arranged For sleep state.Speech recognition apparatus obtains the first voice data in sleep state, if the first voice data is crucial with wake-up Word matches, then sets the state of speech recognition circuit to listening state.

Further, equipment control data be expansible markup language (eXtensible Markup Language, Xml) message, the information which need to specifically carry are as follows：The entitled asrDevCtrlReq of message, the parameter of carrying include：Session Mark (Session ID), data type and the corresponding value explanation of control command (CtrlCmd) and each parameter.

Wherein, xml message with double back vehicle ending " r n r n ", convenient for dividing multiple messages in ICP/IP protocol.If The reference of standby control data is arranged, as shown in table 1：

Table 1

According to table 1 it is found that the value of parameter CtrlCmd includes 1 and 2；When the value of parameter CtrlCmd is 1, representative is opened Dynamic speech recognition；When the value of parameter CtrlCmd is 2, represents and close speech recognition.

Speech recognition apparatus parses the equipment control data received, according to message name<asrDevCtrlReq >Field, can know that the message is equipment control data, the value according to the CtrlCmd carried in message and the parameter is 1, it would know that and need to start speech recognition, and then power to speech recognition circuit, speech recognition circuit enters sleep state.

Wherein, for speech recognition circuit in sleep state, the first voice data for receiving microphone monitoring works as voice When the state of identification circuit is sleep state, the first voice data monitored according to microphone determines whether speech recognition circuit State be set as listening state；

When the state of speech recognition circuit is listening state, the second speech data that microphone is monitored is converted into voice Text, and it is sent to control server.

In the above-described embodiments, speech recognition apparatus after the power is turned on, does not power to speech recognition circuit, at speech recognition circuit In inactive state；When speech recognition apparatus receives the control carried for starting speech recognition sent from control server It after system instruction, powers to speech recognition circuit, then speech recognition circuit is from original state, i.e. sleep state is advanced into, or, language The processor setting speech recognition circuit of sound identification equipment enters sleep state.Said program compared with prior art, due to voice Identification equipment after the power is turned on, does not power to speech recognition circuit, and speech recognition circuit still needs in monitoring of environmental in a sleep state Audio data and to audio data carry out analysis determine whether enter listening state.To sum up use the embodiment of this programme can be with It more effectively saves energy consumption, reduces speech recognition circuit and be chronically at sleep state and the equipment loss that generates.

In some illustrative embodiments, after the equipment control data that control server is sent in the reception, further include：

To control server sending device control response；

Equipment control response is xml message, and the information which need to specifically carry is as follows：Message is entitled The parameter of asrDevCtrlRsp, carrying includes：The number of session identification (SessionID), result (Result) and each parameter According to type and corresponding value explanation.

Wherein, xml message with double back vehicle ending " r n r n ", convenient for dividing multiple messages in ICP/IP protocol.

The parameter setting of equipment control response, as shown in table 2：

Table 2

According to table 2 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0, represent Speech recognition apparatus is properly received equipment control data；When the value of parameter Result is other, equipment control data is represented Reception failure.

In some optional embodiments, in step slo, it can be one or more to wake up keyword.

Further, it can be number or word to wake up keyword.

Further, if it is word to wake up keyword, category of language can be one or more.

Optionally, message setting can be sent with other equipment by waking up keyword, can also be speech recognition apparatus manufacture It is preceding preset, can also be to be obtained according to the input signal of user on a display screen.

In some optional embodiments, it is that request is arranged by the device parameter that control server is sent to wake up keyword It carries, by being parsed to device parameter setting request, obtains and wake up keyword.

Further, device parameter setting request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevParamSetReq, carrying includes：Session identification (SessionID) starts keyword (asrStartKeyword) and the data type of each parameter and corresponding value explanation；Wherein, asrStartKeyword Data type is text formatting (text) text, and one or more wake-up keywords are carried in the text texts；

The parameter setting of device parameter setting request, as shown in table 3：

Table 3

Speech recognition apparatus parses the message received, according to message name<asrDevParamSetReq>Word Section, can know that the message is equipment control data, according to the text text carried in message, can obtain and be taken in text texts The wake-up keyword of band.

Wherein, text can carry one or more wake-up keywords；

As text carries a wake-up keyword：" Europe "；

As text carries multiple wake-up keywords：" Europe；Hello；It opens；Speech recognition "

Such as table 4, text can also carry multiple wake-up key combinations；It includes more than two wake up to wake up key combination Keyword；When text, which contains, wakes up key combination, during the first voice data carries out matched with wake-up keyword, It needs to match the first voice data with key combination is waken up, i.e., need to match simultaneously all in wake-up key combination Wake up keyword.

In table 4, text may include a wake-up Keyword List, respectively between English branch between wake-up keyword Every such as table 4；

Wake up key combination 1	Hello；Europe
		Wake up key combination 2	Europe；It opens
Wake up key combination 3	Europe；Speech recognition

Table 4

By table 4 it is found that in the wake-up Keyword List carried in text, multiple wake-up key combinations can be contained, And it is defined to waking up keyword in different wake-up key combinations；

If the text such as tables 4 carried in the device parameter setting request that speech recognition apparatus receives, speech recognition circuit In a sleep state, it receives the first voice data to microphone to parse, if the first voice data includes any one of table 4 Key combination is waken up, then state is adjusted to listening state from sleep state, that is, completes the process of wake-up.

In the present embodiment, define that it is that control server obtains by sending device parameter setting is asked to wake up keyword It takes.

Optionally, it is multiple as waken up keyword, if text is " Europe；Hello；It opens；Speech recognition ", then can root According to the security level of setting, the matching way to waking up keyword selects.Wherein, the number of security level is multiple, can Choosing, security level is three-level, respectively high, medium and low.

Specifically, if security level is low, when speech recognition apparatus be in sleep state, by the first voice data and more A wake-up keyword is matched, and calculates the first matching degree.The corresponding matching of different security levels is stored in the database Threshold value.Corresponding matching threshold is compared when by the first matching degree and security level being low, if being less than the threshold value, is matched Failure；Conversely, then successful match.

Specifically, the matching threshold corresponding to security level, with the raising of security level, matching threshold is higher.Its In, when security level is high, matching threshold 80%；When security level is middle, matching threshold 60%；Work as security level When being low, matching threshold 40%.

In the above-described embodiments, the method for waking up keyword is arranged by control server, not only set than in the prior art The mode for setting wake-up keyword is flexible, and when it is multiple to wake up keyword, can be according to the demand of security level, differentiation Setting wakes up keyword match degree.By the setting of setting and matching threshold to waking up keyword, to a large amount of in environment Voice data is screened, and achievees the purpose that voice data denoising, is cut from sleep state to prevent frequently triggering speech recognition circuit Listening state is changed to, the energy consumption due to listening to state is more than sleep state, and the present embodiment has further reached saving energy consumption, The loss of speech recognition apparatus is reduced, equipment life is improved.

In some illustrative embodiments, when the state of speech recognition circuit is listening state, microphone is monitored Second speech data be converted into speech text, and during being sent to control server, there is following interchangeable embodiment party Formula：

Mode one after the second speech data that microphone is monitored is converted into speech text, is transmitted directly to middle control service Device；

Mode two filters out after the second speech data that microphone is monitored is converted into speech text in speech text Then control instruction after screening is sent to control server by control instruction；Control instruction is the control instruction of speech text Keyword；Such as second speech data is converted into speech text and is：" hello, Europe.My closing television please be help, headlight is opened.It thanks Thank you！", the control instruction after screening is：" closing television " and " opening headlight ".

Above-described embodiment gives speech recognition circuit in the state of listening, the second speech data two monitored microphone Kind processing method, first way are only to carry out data type conversion to second speech data, and be sent to control server, by Speech text is identified in control server；The second way is not only to carry out data type conversion to second speech data, But also control instruction is extracted in speech text, the control instruction after extraction is sent to control server.

Optionally, because when speech recognition apparatus executes the operation of the control instruction in identification speech text, it is difficult to accomplish Complete identification to speech text calculates resolution so during identification, if resolution is more than recognition threshold, adopts With mode two；Conversely, then employing mode one, is all sent to control server by speech text, is known by control server Not.Wherein, resolution is the percentage that the data volume of the successfully text of identification accounts for the data volume of speech text.Optional identification threshold Value is 70%.Aforesaid way is avoided to be referred to because speech recognition apparatus recognition failures cause control server that can not obtain correct control The problem of enabling.

In some optional embodiments, before controlling the device parameter setting request that server is sent in the reception, further include：

Connection request is sent to control server；

Receive the connection response that control server is sent.

Further, connection request is xml message, and the information which need to specifically carry is as follows：

The entitled asrConnectReq of message, the parameter of carrying include：Session identification (SessionID), device identification (DeviceID), device type (DeviceType)；

The parameter setting of connection request, as shown in table 5：

Table 5

According to table 5 it is found that the value of parameter DeviceTypeID includes 0,1,2~10,11；Wherein, work as parameter When the value of DeviceTypeID is 0, all devices are represented；When the value of parameter DeviceTypeID is 1, equipment class is represented Type is control server；When the value of parameter DeviceTypeID is 2~10, it is reserved value, can be according to demand arranged not With the equipment of the distinct device type corresponding to value；When the value of parameter DeviceTypeID is 11, device type is represented For speech recognition apparatus.

Above-described embodiment gives the meaning corresponding to the parameter specifically carried in connection request and parameter value；Even It connects in request, preset device type includes：All devices, control server and speech recognition apparatus, in parameter In the value of DeviceTypeID, 2~10 is reserved, can be used for the connection with other types equipment.Further, parameter DeviceTypeID is 2~10, keeps for the controlled plant controlled by control server.Connection request in above-described embodiment, Pass through Portable device type parameter so that the equipment for receiving the connection request can be quickly through identification parameter DeviceTypeID Value, know the device type of requesting party, and in turn determine whether establish communication connection.Such mode is convenient, improves system Safety.

Further, connection response is xml message, and the information which need to specifically carry is as follows：

The entitled asrConnectRsp of message, the parameter of carrying include：Session identification (SessionID), result (Result) and the data type of each parameter and corresponding value explanation；

Connection response, as shown in table 6：

Table 6

According to table 6 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0, represent Speech recognition apparatus is properly received connection request；When the value of parameter Result is other, connection request reception failure is represented.

Above-described embodiment gives the connection response corresponding to connection request, is responded by feedback link, informs that voice is known Whether other equipment is properly received connection request.

In some optional embodiments, after controlling the connection response that server is sent in the reception, further include：

From receiving connection response, keep-alive message request is sent to control server periodically；Keep-alive message request is taken The device type of band is speech recognition apparatus；The period for sending protection message request is preset fixed duration.Further, Gu Periodically a length of 30s.

Further, keep-alive message request is xml message, and the information which need to specifically carry is as follows：Message is entitled The parameter of asrKeepAliveReq, carrying includes：Session identification (SessionID), device identification (DeviceID), equipment class Type (DeviceType)；The parameter setting of keep-alive message request, as shown in table 7：

Table 7

According to table 7 it is found that the value of parameter DeviceTypeID includes 0,1,2~10,11；Wherein, work as parameter When the value of DeviceTypeID is 0, all devices are represented；When the value of parameter DeviceTypeID is 1, equipment class is represented Type is control server；When the value of parameter DeviceTypeID is 2~10, it is reserved value, can be according to demand arranged not With the equipment of the distinct device type corresponding to value；When the value of parameter DeviceTypeID is 11, device type is represented For speech recognition apparatus.

Further, after sending keep-alive message request, further include：

Receive the keep-alive message response that control server is sent；

Wherein, keep-alive message response is xml message, and the information which need to specifically carry is as follows：Message is entitled The parameter of asrKeepAliveRsp, carrying includes：Session identification (SessionID), result (Result).

The parameter setting of keep-alive message request, as shown in table 8：

Table 8

According to table 8 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0, represent Control server is properly received keep-alive message request；When the value of parameter Result is other, represents control server and do not connect Receive keep-alive message request.By keep-alive message request and keep-alive message response, to maintain speech recognition apparatus and middle control to service Communication connection between device.

In some optional embodiments, after setting sound identification module enters listening state, further include：

It is sent to control server and obtains scene list request；

Receive the acquisition scene list response that control server is sent；Scene list response carries scenario key list, Scenario key list includes multiple scenario keys, with English branch interval between each scenario key.

Further, it is xml message to obtain scene list request, and the information which need to specifically carry is as follows：Message name Parameter for asrGetSceneListReq, carrying includes：Session identification (SessionID).Ginseng is carried in scene list request Number, can be such as table 9：

Table 9

Further, it is xml message to obtain scene list response, and the information which need to specifically carry is as follows：Message name Parameter for asrGetSceneListRsp, carrying includes：Session identification (SessionID), result (Result), scene list (SceneList).The setting of parameter in scene list response is obtained, it is specific such as table 10：

Table 10

Wherein, the scenario key list that SceneList is carried, can be such as table 11：

1	Parlor scene one	Open TV, light turns down one grade
			2	Parlor scene two	It opens virtual dreamland VR, shut curtain
3	Bedroom scene one	Closing television, shut curtain, light turns down one grade
			4	Bedroom scene two	It shuts curtain, open music, fragrant atmosphere in opening chamber

Table 11

After obtaining scenario key list, the scenario key in speech text can be known in the state of listening Not.Scenario key can only include scene, such as parlor scene one, parlor scene two, bedroom scene one and the bedroom in table 11 Scene two；Scenario key, can also include table 11 in the relevant control keyword of each scene.

In some optional embodiments, after sound identification module enters listening state, by the second language of microphone monitoring Sound data conversion is sent to the process of control server at speech text, specific as follows：

Receive the second speech data that recording module is monitored；

Second speech data is converted, the speech text of second speech data is obtained；

Primary voice data transmission request is sent to control server；Primary voice data transmission request carries the second voice The speech text of data.

Wherein, primary voice data transmission request is xml message, and the information which need to specifically carry is as follows：

The entitled asrOrigVoiceDataTransferReq of message, the parameter of carrying include：Session identification (SessionID), coding mode (Encoding), primary voice data (OrigVoiceData).

The setting of parameter, specific such as table 12 in primary voice data transmission request：

Table 12

From table 12 and above-mentioned code it is found that carrying original sound data in primary voice data transmission request, and Coding mode.Wherein, original sound data is text, is a kind of text message, and text needs to use Unicode (8-bit Unicode Transformation Format, UTF-8) it is encoded.

In some optional embodiments, after speech recognition circuit enters listening state, further include：

Triggering timer starts countdown；

If in the timing of timer, the second speech data of recording module monitoring is not received, then voice knowledge is set Other circuit enters sleep state.

In the above-described embodiments, after defining that speech recognition circuit enters listening state, and non-continuous maintenance listening state, When second speech data is not received in the timing in timer, then the listening from high power consumption by the state of speech recognition circuit It listens state to be switched to the sleep state of low-power consumption, avoids high energy consumption, the loss of equipment height caused by persistently maintaining listening state The problem of.

In some optional embodiments, further include：

To control server sending device state report request；Equipment state report request carries speech identifying function shape State.

Receive the equipment state report response that control server is sent；

Further, equipment state report request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevStatusReportReq, carrying includes：Session identification (SessionID), speech recognition working condition (asrWorkStatus)。

The setting of parameter, specific such as table 13 in equipment state report response：

Table 13

As shown in Table 13, when the value of parameter asrWorkStatus is 0, speech identifying function state is represented not open It is dynamic, when the value of parameter asrWorkStatus is 1, speech identifying function state is represented as sleep state, works as parameter When the value of asrWorkStatus is 2, speech identifying function state is represented as listening state.

By equipment state report request, its state can be sent to control server by speech recognition apparatus.

Further, equipment state report response is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevStatusReportRsp, carrying includes：Session identification (SessionID), result (Result)；

The setting of parameter in equipment state report response, such as table 14：

Table 14

According to table 14 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0, represent Control server is properly received equipment state report request；When the value of parameter Result is other, control server is represented Equipment state report request is not received.

In some optional embodiments, after the request of receiving device parameter setting, further include：

It is responded to control server sending device parameter setting.

Wherein, device parameter setting response is xml message, and the information which need to specifically carry is as follows：Message is entitled The parameter of asrDevParamSetRsp, carrying includes：Session identification (SessionID), result (Result)；

The setting of parameter in device parameter setting response, such as table 15：

Table 15

According to table 15 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0, represent Control server is properly received device parameter setting request；When the value of parameter Result is other, control server is represented Device parameter setting request is not received.

Receive the device parameter inquiry request that control server is sent；

It is responded to control server sending device parameter query；Device parameter inquiry response, which carries, wakes up keyword.

Further, device parameter inquiry request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevParamGetReq, carrying includes：Session identification (SessionID)；

The setting of parameter in device parameter setting response, such as table 16：

Table 16

Further, device parameter inquiry response is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevParamGetRsq, carrying includes：Session identification (SessionID), wakes up key at result (Result) Word (asrStartKeyword).The setting of parameter in device parameter inquiry response, with reference to table 17：

Table 17

In some optional embodiments, further include：

Receive the device status inquiries request that control server is sent；

It is responded to control server sending device status inquiry；Device status inquiries response carries speech identifying function shape State.

Further, device status inquiries request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevStatusGetReq, carrying includes：Session identification (SessionID).

The setting of parameter in device parameter setting response, such as table 18：

Table 18

Further, device status inquiries response is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevStatusGetRsq, carrying includes：Session identification (SessionID), result (Result), speech recognition Functional status (asrWorkStatus).The setting of parameter in device parameter inquiry response, with reference to table 19：

Table 19

As shown in Table 19, when the value of parameter asrWorkStatus is 0, speech identifying function state is represented not open It is dynamic, when the value of parameter asrWorkStatus is 1, speech identifying function state is represented as sleep state, works as parameter When the value of asrWorkStatus is 2, speech identifying function state is represented as listening state.

In the embodiment of the present invention in a first aspect, a kind of speech recognition apparatus for speech recognition apparatus of offer wakes up Method can more effectively save energy consumption using scheme provided by the invention, reduce speech recognition circuit and be chronically at sleep shape State and the equipment loss generated.

Fig. 2 is a kind of flow diagram of speech recognition apparatus control method of the embodiment of the present invention.As shown in Figure 2：

Step S201 receives the connection request that speech recognition apparatus is sent；The device type that connection request carries is voice Identification equipment；

Step S202, to speech recognition apparatus sending device control data；Equipment control data is carried for starting voice The control command of identification.

Wherein, equipment control data is used to indicate speech recognition apparatus and powers for speech recognition circuit, and by speech recognition The state of circuit is set as sleep state.

Further include the equipment control response for receiving speech recognition apparatus and sending after step S202.

Equipment control data is used to indicate speech recognition apparatus and powers to speech recognition circuit, and by speech recognition circuit State is set as sleep state.

Further, in step S202, equipment control data is xml message, and the information which need to specifically carry is such as Under：

The entitled asrDevCtrlReq of message, the parameter of carrying include：Session identification (SessionID), control command (CtrlCmd) and the data type of each parameter and corresponding value explanation；The value of parameter CtrlCmd includes 1 and 2；Work as ginseng When the value of number CtrlCmd is 1, represents and start speech recognition；When the value of parameter CtrlCmd is 2, represents and close voice knowledge Not.

The code of equipment control data is referred to the description in above-described embodiment, design parameter in equipment control data Setting refers to table 1.

Further, equipment control response is xml message, and the information which need to specifically carry is as follows：

The entitled asrDevCtrlRsp of message, the parameter of carrying include：Session identification (SessionID), result (Result) and the data type of each parameter and corresponding value explanation；The value of parameter Result includes 0 and other；When When the value of parameter Result is 0, represents speech recognition apparatus and be properly received equipment control data；When the value of parameter Result For other when, represent equipment control data reception failure.

The code of equipment control response is referred to the description in above-described embodiment, design parameter in equipment control response Setting refers to table 2.

In the above-described embodiments, the control command for starting speech recognition is used to indicate speech recognition apparatus to voice Identification circuit is powered, and speech recognition circuit enters sleep state.

In the prior art, after the power is turned on, speech recognition circuit is directly entered sleep state, microphone to speech recognition apparatus Start that the voice data in environment is monitored and is judged whether to wake up to enter listening state.The technical program, speech recognition After device power, speech recognition circuit is in off-position, by control server to speech recognition apparatus sending device control System request, and facilities air request carries the control command for starting speech recognition, so that speech recognition apparatus passes through knowledge The not control command, to be powered to speech recognition circuit, speech recognition circuit enters sleep state.Pass through above-mentioned side Formula, it is possible to prevente effectively from speech recognition apparatus is because being chronically at sleep state, and a large amount of power consumptions and equipment loss generated, reach Effect that is energy saving and improving service life of equipment.

In some illustrative embodiments, to before speech recognition apparatus sending device control data, further include：

It is asked to speech recognition circuit sending device parameter setting；Device parameter setting request, which carries, wakes up keyword.

Further, device parameter setting request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevParamSetReq, carrying includes：Session identification (SessionID) starts keyword (asrStartKeyword) and the data type of each parameter and corresponding value explanation；

The code of device parameter setting request, please refers to related description in previous embodiment；Device parameter setting request The setting of parameter, with reference to table 3.

Wherein, the data type of asrStartKeyword is text texts, is carried in the text texts one or more Wake up keyword；

Wherein, text can carry one or more wake-up keywords；

As text carries a wake-up keyword：" Europe "；

As text carries multiple wake-up keywords：" Europe；Hello；It opens；Speech recognition "；

Table 4 provides a wake-up Keyword List, includes more wake-up key combinations in wake-up Keyword List, With reference to table 4, text can also carry multiple wake-up key combinations；It includes that more than two wake-ups are crucial to wake up key combination Word.

In the above-described embodiments, the method that control server setting wakes up keyword, not only setting is called out than in the prior art The mode of awake keyword is flexible, can be arranged with differentiation and wake up keyword.By the setting to waking up keyword, and then control language Sound identification equipment screens a large amount of voice data in environment, achievees the purpose that voice data denoising, to prevent frequently triggering Speech recognition circuit is switched to listening state from sleep state, and the energy consumption due to listening to state is more than sleep state, this reality It applies example and has reached saving energy consumption, reduce the loss of speech recognition apparatus, improve the purpose of equipment life.

In some illustrative embodiments, before being asked to speech recognition apparatus sending device parameter setting, further include：

Receive the connection request that speech recognition apparatus is sent；The device type that connection request carries is speech recognition apparatus；

Connection response is sent to speech recognition apparatus；Connection response carries the parameter for being successfully established connection；

The entitled asrConnectReq of message, the parameter of carrying include：Session identification (SessionID), device identification (DeviceID), device type (DeviceType)；The parameter setting of connection request, as shown in table 5；The code of connection request is asked With reference to the embodiment of first aspect present invention.

The parameter setting of connection request, as shown in table 6；The code of connection request please refers to the implementation of first aspect present invention Example.According to table 6 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0, voice knowledge is represented Other equipment is properly received connection request；When the value of parameter Result is other, connection request reception failure is represented.

The parameter setting of connection request, as shown in table 5；The code of connection request please refers to the reality of the first aspect of the present invention Apply example.

In some optional embodiments, after sending connection response to speech recognition apparatus, further include：

The keep-alive message that the speech recognition apparatus period sends is received to ask；The device type that keep-alive message request carries is language Sound identification equipment；The period for sending protection message request is preset fixed duration.

Further, a length of 30s when fixed.

Further, keep-alive message request is xml message, and the information which need to specifically carry is as follows：

The entitled asrKeepAliveReq of message, the parameter of carrying include：Session identification (SessionID), device identification (DeviceID), device type (DeviceType)；

The code of keep-alive message request, please refers to the respective description in the embodiment of the first aspect of the present invention.Keep-alive disappears The parameter setting of request is ceased, as shown in table 7.According to table 7 it is found that the value of parameter DeviceTypeID include 0,1,2~10, 11；Wherein, when the value of parameter DeviceTypeID is 0, all devices are represented；When the value of parameter DeviceTypeID is When 1, device type is represented as control server；It is reserved value when the value of parameter DeviceTypeID is 2~10, it can be with The equipment of distinct device type corresponding to different values is set according to demand；When the value of parameter DeviceTypeID is 11 When, device type is represented as speech recognition apparatus.

Further, after receiving keep-alive message request, further include：

Keep-alive message response is sent to speech recognition apparatus；

Wherein, keep-alive message response is xml message, and the information which need to specifically carry is as follows：

The entitled asrKeepAliveRsp of message, the parameter of carrying include：Session identification (SessionID), result (Result)；The embodiment of the first aspect of the present invention gives the code of keep-alive message response, and keep-alive message is listed in table 8 The parameter setting of response, according to table 8 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0 When, it represents control server and is properly received keep-alive message request；When the value of parameter Result is other, service is controlled in representative Device does not receive keep-alive message request.By keep-alive message request and keep-alive message response, to maintain speech recognition apparatus in Control the communication connection between server.

In some optional embodiments, further include：

Receive the equipment state report request that speech recognition apparatus is sent；Equipment state report request carries speech recognition and sets Standby speech recognition working condition, speech recognition working condition include：Not actuated, sleep state and listening state.

It is responded to speech recognition apparatus sending device state report.

Further, equipment state report request is xml message, and the information which need to specifically carry is as follows：

The entitled asrDevStatusReportReq of message, the parameter of carrying include：Session identification (SessionID), voice Identify working condition (asrWorkStatus).

The code of specific equipment state report request, please refers to and is said accordingly in the embodiment of the first aspect of the present invention It is bright.Table 13 gives the design parameter setting of equipment state report request.As shown in Table 13, when parameter asrWorkStatus's When value is 0, speech identifying function state is represented to be not actuated, when the value of parameter asrWorkStatus is 1, represents language Sound identification function state is sleep state, when the value of parameter asrWorkStatus is 2, represents speech identifying function state To listen to state.By equipment state report request, control server knows the speech identifying function state of speech recognition apparatus.

The explanation of equipment state report response, please refers to and illustrates accordingly in the embodiment of the first aspect of the present invention.Table 14 give the design parameter setting of equipment state report request.The setting of parameter in equipment state report response, such as table 14：Ginseng The value of number Result includes 0 and other；When the value of parameter Result is 0, represents control server and be properly received equipment State report request；When the value of parameter Result be other when, represent control server do not receive equipment state report ask It asks.

In some optional embodiments, further include：

The primary voice data transmission request that speech recognition apparatus is sent is received, primary voice data transmission request carries the The speech text of two voice data；

The speech text that the second speech data that request carries is transmitted according to primary voice data, to one or more controlled Equipment is controlled.

Further, primary voice data transmission request is xml message, and the information which need to specifically carry is as follows：Disappear Entitled asrOrigVoiceDataTransferReq is ceased, the parameter of carrying includes：Session identification (SessionID), coding mode (Encoding), primary voice data (OrigVoiceData).The code of primary voice data transmission request, please refers to this hair Illustrate accordingly in the embodiment of bright first aspect.The design parameter that table 12 gives primary voice data transmission request is set It sets, it is specific such as table 12：Original sound data and coding mode are carried in primary voice data transmission request.Wherein, original Voice data is text, is a kind of text message, and text needs to use Unicode (8-bit Unicode Transformation Format, UTF-8) it is encoded.

Further, the speech text of second speech data can be that second speech data is converted into the complete of speech text Text, or, the control instruction for being converted into filtering out in speech text from second speech data.If the text of second speech data is Control instruction after screening then reads institute's control instruction, and is controlled to one or more controlled plants according to the control instruction System.It is converted into the full text sheet of speech text if the text of second speech data is second speech data, to the full text in this Appearance is identified, and filters out control instruction included in full text sheet, and set according to the control instruction is controlled to one or more It is standby to be controlled.

In some optional embodiments, method further includes：

Receive the device parameter setting response that speech recognition apparatus is sent.

Wherein, device parameter setting response is xml message, and the information which need to specifically carry is as follows：Message is entitled The parameter of asrDevParamSetRsp, carrying includes：Session identification (SessionID), result (Result).It can according to table 15 Know, the value of parameter Result includes 0 and other；When the value of parameter Result is 0, control server successful reception is represented Device parameter setting request；When the value of parameter Result is other, represents control server and do not receive device parameter and set Set request.

In some optional embodiments, method further includes：

It is asked to speech recognition apparatus sending device parameter query；

Receive the device parameter inquiry response that speech recognition apparatus is sent；Device parameter inquiry response, which carries, wakes up key Word.

Further, device parameter inquiry request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevParamGetReq, carrying includes：Session identification (SessionID).Ginseng in device parameter inquiry request Several settings is with reference to table 16.

Further, device parameter inquiry response is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevParamGetRsq, carrying includes：Session identification (SessionID), wakes up key at result (Result) Word (asrStartKeyword).The setting of parameter is with reference to table 17 in device parameter inquiry response.

In some optional embodiments, further include：

To speech recognition apparatus sending device status query request；

Receive the device status inquiries response that speech recognition apparatus is sent；Device status inquiries response carries speech recognition work( It can state.

Further, device status inquiries request is xml message, and with reference to table 18, the information which need to specifically carry is such as Under：The entitled asrDevStatusGetReq of message, the parameter of carrying include：Session identification (SessionID).

Further, device status inquiries response is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevStatusGetRsq, carrying includes：Session identification (SessionID), result (Result), speech recognition Functional status (asrWorkStatus).The setting of parameter in device parameter inquiry response, with reference to table 19.Work as parameter When the value of asrWorkStatus is 0, speech identifying function state is represented to be not actuated, when taking for parameter asrWorkStatus When value is 1, speech identifying function state is represented as sleep state, when the value of parameter asrWorkStatus is 2, represents language Sound identification function state is listening state.

In some optional embodiments, method further includes：

Receive the acquisition scene list request that speech recognition apparatus is sent；

It is sent to speech recognition apparatus and obtains scene list response；Scene list response carries scenario key list, field Scape Keyword List includes multiple scenario keys, with English branch interval between each scenario key.

Further, it is xml message to obtain scene list request, and the information which need to specifically carry is as follows：Message name Parameter for asrGetSceneListReq, carrying includes：Session identification (SessionID).Ginseng is carried in scene list request Number, is referred to table 9.

Further, it is xml message to obtain scene list response, and the information which need to specifically carry is as follows：Message name Parameter for asrGetSceneListRsp, carrying includes：Session identification (SessionID), result (Result), scene list (SceneList)；The setting of parameter in scene list response is obtained, it is specific such as table 10.The scenario key that SceneList is carried List, can be as shown in table 11.Scenario key can only include scene, as in table 11 parlor scene one, parlor scene two, Bedroom scene one and bedroom scene two；Scenario key, can also include table 11 in the relevant control keyword of each scene. In the present embodiment, control server can control each scene in scene list, therefore obtain scene row receiving After table request, by obtaining scene list response, scene list SceneList is sent to speech recognition apparatus so that voice After identification equipment knows the scenery control service that control server provides, and speech recognition apparatus obtains the list, Ke Yigen According to the scenario key in SceneList, fining identification is carried out to second speech data.To sum up use this programme method can be with Energy consumption is more effectively saved, the equipment that the speech recognition circuit of speech recognition apparatus is generated because being chronically at sleep state is reduced Loss.

In order to preferably describe the mistake of interaction and control between speech recognition apparatus and control server in above-described embodiment Journey illustrates the interaction of equipment room referring to Fig. 3.Fig. 3 is a kind of stream of speech recognition apparatus control method of the present invention Journey schematic diagram.As shown in Figure 3：

Step S301, speech recognition apparatus send connection request asrConnectReq to control server；

Step S302, control server send connection response asrConnectRsp to speech recognition apparatus；

Step S303, speech recognition apparatus send keep-alive message request to control server periodically asrKeepAliveReq；Wherein, sending cycle 30s.

Step S304, control server send keep-alive message to speech recognition apparatus and respond asrKeepAliveRsp；

Step S305, control server are asked to speech recognition apparatus sending device parameter setting asrDevParamSetReq；Device parameter setting request, which carries, wakes up keyword；

Step S306, speech recognition apparatus are responded to control server sending device parameter setting asrDevParamSetRsp；

Step S307, control server is to speech recognition apparatus sending device control data asrDevCtrlReq；Equipment control System request carries the control instruction for starting speech recognition circuit；

Step S308, speech recognition apparatus is to control server sending device control response asrDevCtrlRsp；

Speech recognition circuit is switched to sleep state by step S309, speech recognition apparatus from not actuated；

Step S310, speech recognition apparatus is to control server sending device state report request AsrDevStatusReportReq, report state are sleep state；

Step S311, control server are responded to speech recognition apparatus sending device state report asrDevStatusReportRsp；

Step S312, speech recognition apparatus are received containing keyword is waken up, and listening state is switched to from sleep state；

Step S313, speech recognition apparatus is to control server sending device state report request AsrDevStatusReportReq, report state are listening state；

Step S314, control server are responded to speech recognition apparatus sending device state report asrDevStatusReportRsp；

Step S315, speech recognition apparatus monitor voice data, identify the phonetic control command in voice data；

Step S316, speech recognition apparatus send primary voice data transmission request to control server AsrOrigVoiceDataTransferReq, the request carry phonetic control command text；

Step S317, control server send primary voice data transmission response to speech recognition apparatus asrOrigVoiceDataTransferRsp；

Step S318, control server control controlled plant or scene according to phonetic control command text；Scene Including multiple controlled plants；

Step S319, speech recognition apparatus is in the state of listening, meeting opening timing device, if be more than the setting of timer Between, unlisten receives voice data, then speech recognition circuit is switched to sleep state from listening state；

Step S320, speech recognition apparatus is to control server sending device state report request AsrDevStatusReportReq, report state are sleep state；

Step S321, control server are responded to speech recognition apparatus sending device state report AsrDevStatusReportRsp is informed and is had been received by equipment state report request.

In the above-described embodiments, speech recognition apparatus and control server, which are established, communicates to connect, and speech recognition apparatus can be It establishes after communication connection and sends keep-alive message to control server every 30s, to maintain the Continued communication of both sides to connect；Middle control clothes Device be engaged in the request of speech recognition apparatus sending device parameter setting, speech recognition apparatus record wakes up keyword；Speech recognition is set The standby equipment control data for receiving control server and sending, powers to the speech recognition circuit where speech identifying function, voice Identification circuit is switched to sleep state from not actuated；In a sleep state, speech recognition apparatus monitors voice data, if listening to Keyword is waken up, then speech recognition circuit is switched to listening state from sleep state；It, can be to the language of monitoring under listening state Sound data are identified, and the text message containing control instruction that will identify that is sent to control server.From above-described embodiment In, it can be seen that after the power is turned on, speech recognition circuit is in inactive state to speech recognition apparatus, does not give speech recognition at this time Circuit is powered, but after receiving equipment control data, it just powers to speech recognition circuit, and it is set and enters sleep state. In this way, speech recognition apparatus is avoided to be chronically at high energy consumption caused by sleep state, occupy hardware resource Problem.In addition, when the speech recognition circuit of speech recognition apparatus is in the listening state of high power consumption, meeting opening timing device, if Voice data is not heard in setting time, then speech recognition circuit is switched to the sleep shape of low-power consumption from listening state State has further reached energy saving, has reduced the effect of equipment loss.

Fig. 4 is a kind of block diagram of speech recognition apparatus of the embodiment of the present invention.As shown in figure 4,

In some exemplary embodiments, speech recognition apparatus, including：Processor 401, power supply module 402 and voice Identification circuit 403；

Processor 401, the equipment control data for receiving control server transmission；Equipment control data is carried for opening The control command of dynamic speech recognition；And control power supply module 402 is powered for speech recognition circuit 403；

Power supply module 402, for powering for speech recognition circuit；

Speech recognition circuit 403, for identification voice data.

After the power is turned on, speech recognition circuit is in inactive state to speech recognition apparatus, i.e., is not supplied to speech recognition circuit Electricity, but after receiving equipment control data, it just powers to speech recognition circuit, and it is set and enters sleep state.Pass through this The mode of sample avoids speech recognition apparatus and is chronically at high energy consumption caused by sleep state, occupies the problem of hardware resource.

Fig. 5 is an a kind of specific embodiment of speech recognition apparatus of the embodiment of the present invention, as shown in figure 5,

In some exemplary embodiments, speech recognition apparatus, including：

Signal receiver 501, power supply module 502, microphone 503, speech recognition circuit 504 and processor 505；Wherein,

Signal receiver 501, the equipment control data for receiving control server transmission；Wherein, equipment control data Carry the control command for starting speech recognition；

Processor 505, the control command for starting speech recognition that equipment control data carries for identification, to power supply Module 502 sends power supply signal；And after power supply module 502 is the power supply of speech recognition circuit 504, to speech recognition circuit 504 send first state switching signal；And after the successful match signal for receiving the transmission of speech recognition circuit 504, to language Sound identification circuit 504 sends the second state switching signal；

Power supply module 502, the power supply signal for receiving the transmission of processor 505 are powered to speech recognition circuit；

Microphone 503, for monitoring voice data, and, send voice data to speech recognition circuit 504；

Speech recognition circuit 504, the first state switching signal for receiving the transmission of processor 505, sets state to Sleep state；When in sleep state, the first voice data that microphone 503 is monitored is received, is sentenced according to the first voice data It is disconnected whether to processor 505 to send successful match signal；And

The second state switching signal that processor 505 is sent is received, sets state to listening state；When in listening shape When state, the second speech data that microphone 503 is monitored is received, second speech data is converted to the voice text of second speech data Originally it is sent to control server.

Wherein, equipment control data is xml message, and the information which need to specifically carry is as follows：

The entitled asrDevCtrlReq of message, the parameter of carrying include：Session identification (SessionID), control command (CtrlCmd) and the data type of each parameter and corresponding value explanation；The value of parameter CtrlCmd includes 1 and 2；Work as ginseng When the value of number CtrlCmd is 1, represents and start speech recognition；When the value of parameter CtrlCmd is 2, represents and close voice knowledge Not.The code and parameter setting of specific equipment control data, please refer to correlation in the embodiment of the first aspect of the present invention and retouch It states.

Processor 505 to receive equipment control come in row parse, according to message name<asrDevCtrlReq>Word Section, can know that the message is equipment control data, be 1 according to the value of the CtrlCmd carried in message and the parameter, can Know and need to start speech recognition, and then first state switching signal is sent to power supply module 502, is receiving speech recognition electricity After the successful match signal that road 504 is sent, the second state switching signal, speech recognition circuit are sent to speech recognition circuit 504 504 enter sleep state.

In the above-described embodiments, speech recognition apparatus after the power is turned on, do not power to speech recognition circuit 504 by power supply module 502, Speech recognition circuit 504 is in inactive state；It is carried from what control server was sent when speech recognition apparatus 504 is received It after control instruction for starting speech recognition, powers to speech recognition circuit 504, then speech recognition circuit 504 is advanced certainly Enter original state, i.e. sleep state, or, speech recognition circuit 504, which is arranged, in the processor 505 of speech recognition apparatus enters sleep shape State.Said program compared with prior art, after the power is turned on due to speech recognition apparatus, does not power to speech recognition circuit 504, and voice Identification circuit 504 still need in a sleep state the audio data in monitoring of environmental and to audio data carry out analysis determine whether into Enter listening state.It to sum up uses the embodiment of this programme that can more effectively save energy consumption, reduces the voice of speech recognition apparatus The equipment loss that identification circuit is generated because being chronically at sleep state.

In some illustrative embodiments,

Signal receiver 501 is additionally operable to receive the device parameter setting request that control server is sent；Device parameter is arranged Request, which carries, wakes up keyword.

Processor 505 is according to message name<asrDevParamSetReq>Field can know that the message is equipment control Request, according to the text text carried in message, can obtain the wake-up keyword carried in text texts.

Wherein, text can carry one or more wake-up keywords；

As text carries a wake-up keyword：" Europe "；

The code of device parameter setting request, please refers to the embodiment of the first aspect of the present invention.The setting of device parameter Such as table 4, text can also carry multiple wake-up key combinations；It includes more than two wake-up keywords to wake up key combination； When text contain wake up key combination when, the first voice data with wake up keyword carry out it is matched during, need by First voice data is matched with key combination is waken up, i.e., need to match simultaneously and wake up wake-up pass all in key combination Key word.

If the text such as tables 4 carried in the device parameter setting request that processor 505 receives, speech recognition circuit 504 In a sleep state, it receives the first voice data to microphone 503 to parse, if the first voice data includes to appoint in table 4 State is then adjusted to listening state from sleep state, that is, completes what speech recognition apparatus woke up by one wake-up key combination Process.By the restriction to waking up keyword, listening state is switched to from sleep state to speech recognition circuit 504 and carries out difference Change control.

Speech recognition circuit 504 is additionally operable to when in sleep state, receives the first voice data that microphone is monitored, First voice data is matched with keyword is waken up, if successful match, successful match signal is sent to processor.

In the above-described embodiments, keyword match degree can be waken up according to the demand of security level, differentiation setting.Pass through Setting to the setting and the matching threshold that wake up keyword, screens a large amount of voice data in environment, reaches voice The purpose of data de-noising is switched to listening state to prevent frequently triggering speech recognition circuit from sleep state, due to listening to state Energy consumption be more than sleep state, therefore the present embodiment has further reached saving energy consumption, reduces the loss of speech recognition apparatus, carries High equipment life.

In some optional embodiments, speech recognition apparatus further includes：

Signal projector 506, for after signal receiver 501 receives equipment control data, being sent out to control server Send equipment control response；

Wherein, equipment control response is xml message, and the information which need to specifically carry is as follows：

The entitled asrDevCtrlRsp of message, the parameter of carrying include：Session identification (SessionID), result (Result) and the data type of each parameter and corresponding value explanation；The parameter setting of equipment control response, such as 2 institute of table Show；The code of equipment control response please refers to embodiment in the first aspect of the present invention.According to table 2 it is found that parameter Result Value includes 0 and other；When the value of parameter Result is 0, represents speech recognition apparatus and be properly received equipment control data； When the value of parameter Result is other, equipment control data reception failure is represented.

In some optional embodiments,

Signal projector 506 is additionally operable to send connection request to control server；

Signal receiver 501 is additionally operable to receive the connection response that control server is sent.

The entitled asrConnectReq of message, the parameter of carrying include：Session identification (SessionID), device identification (DeviceID), device type (DeviceType)；The parameter setting of connection request, as shown in table 5；The code of connection request, Please refer to embodiment in the first aspect of the present invention.According to table 5 it is found that number DeviceTypeID value include 0,1,2~10, 11；Wherein, when the value of parameter DeviceTypeID is 0, all devices are represented；When the value of parameter DeviceTypeID is When 1, device type is represented as control server；It is reserved value when the value of parameter DeviceTypeID is 2~10, it can be with The equipment of distinct device type corresponding to different values is set according to demand；When the value of parameter DeviceTypeID is 11 When, device type is represented as speech recognition apparatus.Above-described embodiment gives the parameter specifically carried in connection request, Yi Jican Meaning corresponding to numerical value；In a connection request, preset device type includes：All devices, control server and voice are known Other equipment, in the value of parameter DeviceTypeID, 2~10 is reserved, can be used for the connection with other types equipment.Into one Step, parameter DeviceTypeID is 2~10, keeps for the controlled plant controlled by control server.In above-described embodiment Connection request passes through Portable device type parameter so that the equipment for receiving the connection request can be quickly through identification parameter The value of DeviceTypeID knows the device type of requesting party, and determines whether to establish communication connection in turn.Such mode is just Victory improves security of system.

The entitled asrConnectRsp of message, the parameter of carrying include：Session identification (SessionID), result (Result) and the data type of each parameter and corresponding value explanation；The parameter setting of connection response, as shown in table 6；Even The code for connecing response please refers to embodiment in the first aspect of the present invention.According to table 6 it is found that the value of parameter Result includes 0 And other；When the value of parameter Result is 0, represents speech recognition apparatus and be properly received connection request；As parameter Result Value be other when, represent connection request reception failure.Above-described embodiment gives the connection response corresponding to connection request, It is responded by feedback link, informs whether speech recognition apparatus is properly received connection request.

In some optional embodiments,

Signal projector 506 is additionally operable to send keep-alive message request to control server periodically；Keep-alive message request is taken The device type of band is speech recognition apparatus；The period for sending protection message request is preset fixed duration.Further, Gu Periodically a length of 30s.

Signal receiver 501 is additionally operable to receive the keep-alive message response that control server is sent.

Further, keep-alive message request is xml message, and the information which need to specifically carry is as follows：Message is entitled The parameter of asrKeepAliveReq, carrying includes：Session identification (SessionID), device identification (DeviceID), equipment class Type (DeviceType)；The parameter setting of keep-alive message request, as shown in table 7；The code of keep-alive message request, please refers to this hair Embodiment in bright first aspect.According to table 7 it is found that the value of parameter DeviceTypeID includes 0,1,2~10,11；Wherein, When the value of parameter DeviceTypeID is 0, all devices are represented；When the value of parameter DeviceTypeID is 1, represent Device type is control server；It is reserved value when the value of parameter DeviceTypeID is 2~10, it can be according to demand The equipment of distinct device type corresponding to different values is set；When the value of parameter DeviceTypeID is 11, representative is set Standby type is speech recognition apparatus.

Further, keep-alive message response is xml message, and the information which need to specifically carry is as follows：Message is entitled The parameter of asrKeepAliveRsp, carrying includes：Session identification (SessionID), result (Result)；Keep-alive message responds Parameter setting, as shown in table 8；The code of keep-alive message response, please refers to embodiment in the first aspect of the present invention.According to table 8 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0, control server success is represented Receive keep-alive message request；When the value of parameter Result is other, represents control server and do not receive keep-alive message and ask It asks.By keep-alive message request and keep-alive message response, to maintain the communication link between speech recognition apparatus and control server It connects.

In some optional embodiments,

Signal projector 506 is additionally operable to send acquisition scene list request to control server；

Signal receiver 501 is additionally operable to receive the acquisition scene list response that control server is sent；Scene list response Carry scenario key list.

Further, it is xml message to obtain scene list request, and the information which need to specifically carry is as follows：Message name Parameter for asrGetSceneListReq, carrying includes：Session identification (SessionID).Obtain the ginseng of scene list request Number setting, as shown in table 9.

Further, it is xml message to obtain scene list response, and the information which need to specifically carry is as follows：

The entitled asrGetSceneListRsp of message, the parameter of carrying include：Session identification (SessionID), result (Result), scene list (SceneList).As shown in table 10, scenario key list includes multiple scenario keys, each field With English branch interval between scape keyword.Wherein, the scenario key list that SceneList is carried, can be as shown in table 11. Scenario key can only include scene, such as parlor scene one, parlor scene two, bedroom scene one and the bedroom scene in table 11 Two；Scenario key, can also include table 11 in the relevant control keyword of each scene.

In some optional embodiments,

Speech recognition circuit 504 is additionally operable to send primary voice data transmission request to control server；Raw tone number The speech text of carrying second speech data is asked according to transmission.

Wherein, primary voice data transmission request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrOrigVoiceDataTransferReq, carrying includes：Session identification (SessionID), coding mode (Encoding), primary voice data (OrigVoiceData).The setting of parameter, reference in primary voice data transmission request The code of 12. primary voice data of table transmission request, please refers to the embodiment of the first aspect of the present invention.From table 12 and code portions Divide it is found that carrying original sound data and coding mode in primary voice data transmission request.Wherein, original sound Data are text, are a kind of text messages, and text needs to be encoded with UTF-8.

In some optional embodiments,

Shown signal projector 506 is additionally operable to control server sending device state report request；Equipment state is reported Request carries speech recognition working condition.

Shown signal receiver 501 is additionally operable to receive the equipment state report response that control server is sent；

Further, equipment state report request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevStatusReportReq, carrying includes：Session identification (SessionID), speech recognition working condition (asrWorkStatus).The parameter setting of equipment state report request refers to table 13, as shown in Table 13, works as parameter When the value of asrWorkStatus is 0, speech identifying function state is represented to be not actuated, when taking for parameter asrWorkStatus When value is 1, speech identifying function state is represented as sleep state, when the value of parameter asrWorkStatus is 2, represents language Sound identification function state is listening state.The code of equipment state report request please refers to the implementation of the first aspect of the present invention Example.

Further, equipment report response is xml message, and the information which need to specifically carry is as follows：

The entitled asrDevStatusReportRsp of message, the parameter of carrying include：Session identification (SessionID), result (Result).With reference to table 14, it is known that the value of parameter Result includes 0 and other；When the value of parameter Result is 0, generation Table control server is properly received equipment state report request；When the value of parameter Result is other, service is controlled in representative Device does not receive equipment state report request.In the third aspect of the embodiment of the present invention, a kind of speech recognition apparatus provided is adopted With scheme provided by the invention more effectively to save energy consumption, the speech recognition circuit of speech recognition apparatus is reduced because being chronically at Sleep state and the equipment loss generated.

A kind of above-mentioned control server is illustrated with reference to Fig. 6.As shown in Figure 6：

In some exemplary embodiments, control server, including：

Signal projector 601 is used for speech recognition apparatus sending device control data；The carrying of equipment control data is used for Start the control command of speech recognition.

Signal receiver 602, the connection request for receiving speech recognition apparatus transmission；The equipment class that connection request carries Type is speech recognition apparatus.

Further, signal receiver 602 are additionally operable to receive the equipment control response that speech recognition apparatus is sent；

Further, equipment control data is xml message, and the information which need to specifically carry is as follows：

The entitled asrDevCtrlReq of message, the parameter of carrying include：Session identification (SessionID), control command (CtrlCmd) and the data type of each parameter and corresponding value explanation；The value of parameter CtrlCmd includes 1 and 2；Work as ginseng When the value of number CtrlCmd is 1, represents and start speech recognition；When the value of parameter CtrlCmd is 2, represents and close voice knowledge Not.The code of equipment control data is referred to the description in the embodiment of the first aspect of the present invention, in equipment control data The setting of design parameter refers to table 1.

Further, equipment control response is xml message, and the information which need to specifically carry is as follows：Message is entitled The parameter of asrDevCtrlRsp, carrying includes：The number of session identification (SessionID), result (Result) and each parameter According to type and corresponding value explanation；The value of parameter Result includes 0 and other；When the value of parameter Result is 0, generation Table speech recognition apparatus is properly received equipment control data；When the value of parameter Result is other, represents equipment control and ask Ask reception failure.The code of equipment control response please refers to the description in the embodiment of the first aspect of the present invention, equipment control The setting of design parameter refers to table 2 in response.

In the prior art, after the power is turned on, speech recognition circuit is directly entered sleep state, microphone to speech recognition apparatus Start that the voice data in environment is monitored and is judged whether to wake up to enter listening state.The technical program, speech recognition After device power, speech recognition circuit is in off-position, by control server to speech recognition apparatus sending device control System request, and facilities air request carries the control command for starting speech recognition, so that speech recognition apparatus passes through knowledge Other control command, to be powered to speech recognition circuit, so that speech recognition circuit enters sleep state.Pass through above-mentioned side Formula, it is possible to prevente effectively from speech recognition apparatus is because being chronically at sleep state, and a large amount of power consumptions and equipment loss generated, reach Effect that is energy saving and improving service life of equipment.

In some illustrative embodiments,

Signal projector 601 is additionally operable to ask to speech recognition circuit sending device parameter setting；Device parameter setting is asked Carrying is asked to wake up keyword.

Further, device parameter setting request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevParamSetReq, carrying includes：Session identification (SessionID) starts keyword (asrStartKeyword) and the data type of each parameter and corresponding value explanation；The generation of device parameter setting request Code, please refers to related description in the embodiment of invention the first aspect；The setting of the parameter of device parameter setting request, with reference to table 3。

Wherein, text can carry one or more wake-up keywords；

As text carries a wake-up keyword：" Europe "；

Table 4 in such as embodiment of invention the first aspect, text can also carry multiple wake-up key combinations；It wakes up and closes The combination of key word wakes up keyword comprising more than two.

In some illustrative embodiments, control server further includes：

Signal receiver 602 is additionally operable to receive the connection request that speech recognition apparatus is sent；The equipment that connection request carries Type is speech recognition apparatus.

The entitled asrConnectReq of message, the parameter of carrying include：Session identification (SessionID), device identification (DeviceID), device type (DeviceType)；The parameter setting of connection request, as shown in table 5；The code of connection request, Please refer to related description in the embodiment of invention the first aspect.According to table 5 it is found that the value of parameter DeviceTypeID includes 0,1,2~10,11；Wherein, when the value of parameter DeviceTypeID is 0, all devices are represented；Work as parameter When the value of DeviceTypeID is 1, device type is represented as control server；When the value of parameter DeviceTypeID is 2 It is reserved value when~10, the equipment of the distinct device type corresponding to different values can be set according to demand；Work as parameter When the value of DeviceTypeID is 11, device type is represented as speech recognition apparatus.Above-described embodiment gives connection request In meaning corresponding to the parameter that specifically carries and parameter value；In a connection request, preset device type includes：It is all Equipment, control server and speech recognition apparatus, in the value of parameter DeviceTypeID, 2~10 is reserved, can be used for The connection of other types equipment.Further, parameter DeviceTypeID is 2~10, keeps for and is controlled by control server Controlled plant.Connection request in above-described embodiment passes through Portable device type parameter so that the equipment for receiving the connection request It can know the device type of requesting party quickly through the value of identification parameter DeviceTypeID, and determine whether to establish in turn Communication connection.Such mode is convenient, improves security of system.

Signal projector 601 is additionally operable to send connection response to speech recognition apparatus；Connection response carries the company of being successfully established The parameter connect.

The parameter setting of connection request, as shown in table 6；The code of connection request please refers to the reality of invention the first aspect Apply related description in example.According to table 6 it is found that the value of parameter Result includes 0 and other；When the value of parameter Result is 0 When, it represents speech recognition apparatus and is properly received connection request；When the value of parameter Result is other, represents connection request and connect Receive failure.

Above-described embodiment gives the connection response corresponding to connection request, is responded by feedback link, signal projector 601, by sending connection response, inform whether speech recognition apparatus is properly received connection request.

In some optional embodiments,

Signal receiver 602 is additionally operable to receive the keep-alive message request of speech recognition apparatus period transmission；Keep-alive message is asked It is speech recognition apparatus to seek the device type of carrying；The period for sending protection message request is preset fixed duration.Further , a length of 30s when fixed.

The entitled asrKeepAliveReq of message, the parameter of carrying include：Session identification (SessionID), device identification (DeviceID), device type (DeviceType)；The code of keep-alive message request, please refers to the reality of the first aspect of the present invention Apply the respective description in example.The parameter setting of keep-alive message request, as shown in table 7.According to table 7 it is found that parameter The value of DeviceTypeID includes 0,1,2~10,11；Wherein, when the value of parameter DeviceTypeID is 0, institute is represented There is equipment；When the value of parameter DeviceTypeID is 1, device type is represented as control server；Work as parameter When the value of DeviceTypeID is 2~10, it is reserved value, the distinct device corresponding to different values can be set according to demand The equipment of type；When the value of parameter DeviceTypeID is 11, device type is represented as speech recognition apparatus.

Signal projector 601 is additionally operable to send keep-alive message response to speech recognition apparatus；

In some optional embodiments,

Signal receiver 602 is additionally operable to receive the equipment state report request that speech recognition apparatus is sent；Equipment state report The speech recognition working condition that request carries speech recognition apparatus is accused, speech recognition working condition includes：Not actuated, sleep state With the state of listening.

Signal projector 601 is additionally operable to respond to speech recognition apparatus sending device state report.

Further, equipment state report request is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevStatusReportReq, carrying includes：Session identification (SessionID), speech recognition working condition (asrWorkStatus).The code of specific equipment state report request, in the embodiment for please referring to the first aspect of the present invention Corresponding explanation.Table 13 gives the design parameter setting of equipment state report request.As shown in Table 13, work as parameter When the value of asrWorkStatus is 0, speech identifying function state is represented to be not actuated, when taking for parameter asrWorkStatus When value is 1, speech identifying function state is represented as sleep state, when the value of parameter asrWorkStatus is 2, represents language Sound identification function state is listening state.By equipment state report request, control server knows the language of speech recognition apparatus Sound identification function state.

Further, equipment state report response is xml message, and the information which need to specifically carry is as follows：Message name Parameter for asrDevStatusReportRsp, carrying includes：Session identification (SessionID), result (Result)；Equipment The explanation of state report response, please refers to and illustrates accordingly in the embodiment of the first aspect of the present invention.Table 14 gives equipment The design parameter of state report request is arranged.The setting of parameter in equipment state report response, such as table 14：Parameter Result's takes Value includes 0 and other；When the value of parameter Result is 0, represents control server and be properly received equipment state report request； When the value of parameter Result is other, represents control server and do not receive equipment state report request.

In some optional embodiments,

Signal receiver 602 is additionally operable to receive the primary voice data transmission request that speech recognition apparatus is sent, original language Sound data transfer request carries the speech text of second speech data；

Control server further includes：

Processor 603, the speech text for transmitting the second speech data that request carries according to primary voice data are right One or more controlled plants are controlled.

Further, primary voice data transmission request is xml message, and the information which need to specifically carry is as follows：Disappear Entitled asrOrigVoiceDataTransferReq is ceased, the parameter of carrying includes：Session identification (SessionID), coding mode (Encoding), primary voice data (OrigVoiceData).

The code of primary voice data transmission request, please refers to and is said accordingly in the embodiment of the first aspect of the present invention It is bright.Table 12 gives the design parameter setting of primary voice data transmission request, specific such as table 12：Primary voice data transmission is asked Original sound data and coding mode are carried in asking.Wherein, original sound data is text, is a kind of text message, and And text needs to be encoded with Unicode (8-bit Unicode Transformation Format, UTF-8).

In some optional embodiments,

Signal receiver 602 is additionally operable to receive the acquisition scene list request that speech recognition apparatus is sent；

Signal projector 601 is additionally operable to send acquisition scene list response to speech recognition apparatus；Scene list response is taken Band scenario key list, scenario key list includes multiple scenario keys, with English branch between each scenario key Interval.

Further, it is xml message to obtain scene list response, and the information which need to specifically carry is as follows：Message name Parameter for asrGetSceneListRsp, carrying includes：Session identification (SessionID), result (Result), scene list (SceneList)；The setting of parameter in scene list response is obtained, it is specific such as table 10.The scenario key that SceneList is carried List, can be as shown in table 11.Scenario key can only include scene, as in table 11 parlor scene one, parlor scene two, Bedroom scene one and bedroom scene two；Scenario key, can also include table 11 in the relevant control keyword of each scene. In the present embodiment, control server can control each scene in scene list, therefore obtain scene row receiving After table request, by obtaining scene list response, scene list SceneList is sent to speech recognition apparatus so that voice After identification equipment knows the scenery control service that control server provides, and speech recognition apparatus obtains the list, Ke Yigen According to the scenario key in SceneList, fining identification is carried out to second speech data.The middle control of this programme is to sum up used to take Business device can more effectively save energy consumption, and the speech recognition power supply for reducing speech recognition apparatus is produced because being chronically at sleep state Raw equipment loss.

It should be understood that the invention is not limited in the flow and structure that are described above and are shown in the accompanying drawings, And various modifications and changes may be made without departing from the scope thereof.The scope of the present invention is only limited by the attached claims System.

Claims

1. a kind of speech recognition apparatus control method, which is characterized in that including：

Receive the equipment control data that control server is sent；The equipment control data carries the control for starting speech recognition System order；

It powers for the speech recognition circuit.

2. the method as described in claim 1, which is characterized in that in the equipment control data that the reception control server is sent Before, further include：

Connection request is sent to the control server；The device type that the connection request carries is speech recognition apparatus.

3. the method as described in claim 1, which is characterized in that in the equipment control data that the reception control server is sent Afterwards, further include：

To the control server sending device state report request；The equipment state report request carries the speech recognition The speech recognition working condition of equipment.

4. method as claimed in claim 2, which is characterized in that after the transmission connection request to the control server, Further include：

Keep-alive message request is sent to the control server；The device type that the keep-alive message request carries is speech recognition Equipment.

5. a kind of speech recognition apparatus control method, which is characterized in that including：

Receive the connection request that the speech recognition apparatus is sent；The device type that the connection request carries sets for speech recognition It is standby；

To speech recognition apparatus sending device control data, the equipment control data is used to indicate the speech recognition apparatus and is Speech recognition circuit is powered, and the equipment control data carries the control command for starting speech recognition.

6. method as claimed in claim 5, which is characterized in that described to speech recognition apparatus sending device control data Before, further include：

It is asked to the speech recognition circuit sending device parameter setting；The device parameter setting request carries the wake-up and closes Key word.

7. method as claimed in claim 6, which is characterized in that asked in the connection for receiving the speech recognition apparatus transmission After asking, further include：

The keep-alive message that the speech recognition apparatus period sends is received to ask；The device type that the keep-alive message request carries For speech recognition apparatus；The period for sending the protection message request is preset fixed duration.

8. method as claimed in claim 5, which is characterized in that described to speech recognition apparatus sending device control data Afterwards, further include：

Receive the equipment state report request that the speech recognition apparatus is sent；The equipment state report request carries institute's predicate The speech recognition working condition of sound identification equipment.

9. a kind of speech recognition apparatus, which is characterized in that including：Processor, power supply module and speech recognition circuit；

The processor, the equipment control data for receiving control server transmission；The equipment control data carrying is used for Start the control command of speech recognition；And it controls the power supply module and powers for the speech recognition circuit；

The power supply module, for powering for the speech recognition circuit；

The speech recognition circuit, for identification voice data.

10. a kind of control server, which is characterized in that including：

Signal receiver, the connection request sent for receiving the speech recognition apparatus；The equipment that the connection request carries Type is speech recognition apparatus；

Signal projector, for speech recognition apparatus sending device control data, the equipment control data to be used to indicate institute It states speech recognition apparatus to power for speech recognition circuit, the equipment control data carries the control for starting speech recognition and orders It enables.