Embodiment
Below in conjunction with accompanying drawing, the present invention is described in further detail.
Fig. 1 is the method flow diagram for carrying out speech recognition of a preferred embodiment of the invention.Step S11, step S21, step S22, step S27, step S12, step S13 and step S14 is comprised according to the method for this preferred embodiment.Wherein, method of the present invention realizes mainly through subscriber equipment.Preferably, method of the present invention is realized by the network equipment and subscriber equipment.Wherein, described subscriber equipment includes but not limited to PC, smart mobile phone, PDA etc.; The described network equipment includes but not limited to the server group that single network server, multiple webserver form or the cloud be made up of a large amount of computing machine or the webserver based on cloud computing (CloudComputing), wherein, cloud computing is the one of Distributed Calculation, the super virtual machine be made up of a group loosely-coupled computing machine collection.Wherein, described subscriber equipment and the network residing for the network equipment include but not limited to internet, wide area network, Metropolitan Area Network (MAN), LAN (Local Area Network), VPN etc.
It should be noted that; described subscriber equipment and the network equipment are only citing; other subscriber equipment that is existing or that may occur from now on, the network equipment or networks, as being applicable to the present invention, within also should being included in scope, and are contained in this with way of reference.
In step s 11, subscriber equipment sends voice messaging to be identified to the network equipment.
Wherein, subscriber equipment can receive the voice messaging to be identified from user, and sends to the network equipment.
Such as, subscriber equipment obtains and " makes a phone call to cuckoo green grass or young crops " from the voice messaging to be identified of user, and sends to the network equipment.
Then, in the step s 21, the network equipment obtains the voice messaging to be identified from subscriber equipment.
Such as, the network equipment receives and " makes a phone call to cuckoo green grass or young crops " from the voice messaging of subscriber equipment.
Then, in step S22, the network equipment carries out speech recognition to this voice messaging to be identified, and obtain recognition result information, wherein, described recognition result information comprises the recognition result content information of at least one sound bite in this speech message to be identified.
Preferably, the recognition result content information of a sound bite includes but not limited to following at least one item:
1) Word message of this sound bite;
Such as, this sound bite is carried out to the Chinese character information, English word information etc. of speech recognition gained.
It should be noted that, when a sound bite can identify obtain multiple pronunciation same or analogous Word message time, the network equipment can by wherein selecting one or more Word message as recognition result content information or its part, or the Word message that all identification obtains by the network equipment is as recognition result content information or its part.
2) Pinyin information of this sound bite;
Such as, this sound bite " cuckoo is blue or green " is carried out to the Pinyin information " duyuqing " etc. of speech recognition gained.
3) the speech waveform information of this sound bite;
Such as, the network equipment extracts the speech waveform information of the sound bite " cuckoo is blue or green " of its None-identified in " being made a phone call to cuckoo green grass or young crops " by voice messaging, as recognition result content information or its part.
Particularly, the network equipment, based on the voice messaging storehouse of pre-determining, identifies voice messaging to be identified, and obtains the recognition result content information of at least one sound bite in this voice messaging to be identified.
Such as, the network equipment is based on the speech message storehouse of pre-determining, speech message to be identified " is made a phone call to cuckoo green grass or young crops " and identifies, obtain Word message that sound bite " gives " " to ", the Word message that the Word message " Du Yuqin " of sound bite " cuckoo green grass or young crops " and Pinyin information " duyuqing " and sound bite " are made a phone call " " is made a phone call ".
Preferably, the network equipment carries out the classified information also comprising at least one sound bite in the recognition result information of speech recognition gained to voice messaging.Wherein, this classified information is for identifying the type of sound bite, and such as, sound bite belongs to name, place name, dialing class, mail class etc.
Wherein, the network equipment obtains the classified information of sound bite by various ways.Such as, when inquiring the information matched with sound bite in voice messaging storehouse, directly obtain the classified information of this information, as the classified information of sound bite; Again such as, the network equipment by carrying out semantic analysis to the text message of speech recognition gained, thus determines the classified information etc. of the sound bite that text information is corresponding.
More preferably, the recognition result content information comprising the sound bite for identifying its correspondence in this classified information is the need of by the identification information carrying out subscriber equipment and carry out local matching inquiry, the network equipment according to the classification of determined sound bite, can determine described identification information.Such as, predtermined category " name ", " place name " needs to carry out local matching inquiry by carrying out subscriber equipment, then when network equipment determination sound bite be categorized as " name " or " place name " time, the recognition result content information needs adding the sound bite for identifying its correspondence in classified information carry out the identification information of local matching inquiry by carrying out subscriber equipment, when network equipment determination sound bite be categorized as other classification time, the recognition result content information adding the sound bite for identifying its correspondence in classified information does not need by the identification information carrying out subscriber equipment and carry out local matching inquiry, or, the network equipment does not add identification information in classified information.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any speech recognition is carried out to voice messaging to be identified, obtain the implementation comprising the recognition result information of the recognition result content information of wherein at least one sound bite, all should be within the scope of the present invention.
Then, in step s 27, recognition result information is sent to subscriber equipment by the network equipment.
Then, in step s 12, subscriber equipment receives recognition result information that the network equipment feeds back, described voice messaging to be identified.
Then, in step s 13, subscriber equipment, according to recognition result content information, carries out matching inquiry in local user's information bank, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described.
Wherein, described local user's information bank comprises in the external storage equipment being stored in described subscriber equipment or this subscriber equipment, and for the information bank of storing subscriber information; Preferably, this local user's information bank can be used as an entirety, store all user profile, or, this local user's information bank comprises multiple independently user information database, and as comprised subscriber phone associated person information storehouse, user MSN associated person information storehouse, user commonly use information of place names storehouse, user commonly uses name information storehouse, dining room etc.
Wherein, described user profile item comprises an information of user; Such as, name of contact person, contact person's mailbox, contact phone, user commonly use place name, user commonly uses dining room title etc.
Particularly, subscriber equipment is according to recognition result content information, in local user's information bank, carry out matching inquiry, include but not limited in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described:
1) subscriber equipment is respectively according to the recognition result content information of all sound bites, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite in all sound bites in local user's information bank.
Such as, in step s 12, subscriber equipment obtain Word message that sound bite " gives " " to ", the Word message that the Word message " Du Yuqin " of sound bite " cuckoo green grass or young crops " and Pinyin information " duyuqing " and sound bite " are made a phone call " " is made a phone call "; Then subscriber equipment is respectively according to the recognition result content information of above-mentioned three sound bites, in local user's information bank, carry out matching inquiry, and only obtain the user profile item " name of contact person: cuckoo is blue or green " matched with the Word message " Du Yuqin " of sound bite " cuckoo is blue or green " and Pinyin information " duyuqing ".
Again such as, in step s 12, subscriber equipment obtain Word message that sound bite " gives " " to ", the Word message that the speech waveform information of sound bite " cuckoo green grass or young crops " and sound bite " are made a phone call " " is made a phone call "; Then subscriber equipment " is given " according to sound bite and the recognition result content information of " making a phone call " respectively, local user's information bank, carry out matching inquiry in the user information database that stores text message, do not obtain the user profile item matched, and, subscriber equipment according to sound bite " cuckoo blue or green " local user's information bank, carry out matching inquiry in the user information database that stores voice messaging, and determine with the user profile item of the speech waveform information match of sound bite " cuckoo is blue or green " to be " name of contact person: cuckoo is blue or green ".Wherein, the speech waveform information in local user's information bank can from user, and, the text message that this speech waveform information is corresponding, or can be arranged by user with the user profile item of this speech waveform information match.
2) subscriber equipment selects part sound bite by all sound bites, and according to the recognition result content information of selected sound bite, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite in selected sound bite in local user's information bank.
Such as, in step s 12, subscriber equipment obtain Word message that sound bite " gives " " to ", the Word message that the Word message " Du Yuqin " of sound bite " cuckoo green grass or young crops " and Pinyin information " duyuqing " and sound bite " are made a phone call " " is made a phone call ", then subscriber equipment is based on universal word storehouse, judge Word message " to " and " making a phone call " as universal word, without the need to performing the operation of local matching inquiry to it, then, subscriber equipment is not judged as Word message " Du Yuqin " and the Pinyin information " duyuqing " of the sound bite " cuckoo is blue or green " of universal word according to its Word message, matching inquiry is carried out in local user's information bank, and only obtain the user profile item " name of contact person: cuckoo is blue or green " matched with the Word message " Du Yuqin " of sound bite " cuckoo is blue or green " and Pinyin information " duyuqing ".
3) subscriber equipment is according to recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the local user's information bank be associated with application pre-determining, to be launched.This implementation will be described in detail in follow-up preferred version, not repeat them here.
Preferably, recognition result content information comprises Word message and Pinyin information, step S13 comprises subscriber equipment further according to described Word message, the user profile item matched is inquired about in described local user's information bank, and when fail the user profile item inquiring described coupling time, according to described Pinyin information, matching inquiry is carried out, to obtain the step of the user profile item matched with the Pinyin information of all or part of sound bite at least one sound bite described in described local user's information bank.
Such as, for Word message " Du Yuqin " and the Pinyin information " duyuqing " of the sound bite obtained in step s 12 " cuckoo is blue or green ", subscriber equipment is inquired about in local user's information bank according to Word message " Du Yuqin ", and fails to inquire the user profile item matched; Then, subscriber equipment carries out matching inquiry according to Pinyin information " duyuqing " in local user's information bank, obtains the user profile item " name of contact person: cuckoo is blue or green " matched with Pinyin information " duyuqing ".
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any according to recognition result content information, matching inquiry is carried out in local user's information bank, to obtain the implementation of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, such as, respectively according to the Word message of a sound bite, Pinyin information and speech waveform information are inquired about in local user's information bank, and when inquiring multiple user profile item respectively with above-mentioned three information matches, by wherein selecting the user profile item that user's frequency of utilization is the highest, or the plurality of user profile item matched is presented to user for its selection etc., all should be within the scope of the present invention.
Then, in step S14, subscriber equipment, according to the application of described recognition result information to be used, performs corresponding operating to the user profile item matched with recognition result content information.
Wherein, the application of described described recognition result information to be used comprises and anyly may be defined as by subscriber equipment the application that needs to use described recognition result information.Preferably, the application of this recognition result information to be used includes but not limited to:
1) the current application being in active state in subscriber equipment;
Such as, in subscriber equipment current started and be in active state mailbox application; Again such as, current to the talk application etc. that another subscriber equipment dials in subscriber equipment.
2) application to be launched determined according to recognition result content information of subscriber equipment;
Such as, the order lexicon of pre-determining is stored in subscriber equipment, wherein, this order lexicon stores commonly used command vocabulary and application corresponding with each commonly used command vocabulary respectively, subscriber equipment obtain in step s 12 sound bite " to " Word message " to ", the Word message that the Word message " Du Yuqin " of sound bite " cuckoo blue or green " and Pinyin information " duyuqing " and sound bite " are made a phone call " " is made a phone call "; Then subscriber equipment is according to the text message of above-mentioned three sound bites, inquire about in the order lexicon of pre-determining, and determine corresponding with text message " phone " to be applied as talk application, then subscriber equipment is using the application of talk application as described recognition result information to be used.
3) also comprise the classified information of at least one sound bite in recognition result information, subscriber equipment, according to this classified information, determines application to be launched, as the application of described recognition result information to be used.This implementation will be described in detail in follow-up preferred version, not repeat them here.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, anyly may be defined as by subscriber equipment the application that needs to use described recognition result information, all should be included in the scope of the application of recognition result information to be used of the present invention.
Particularly, subscriber equipment can in several ways, carrys out the application according to recognition result information to be used, performs corresponding operating to the user profile item matched with recognition result content information.
Such as, recognition result information to be used be applied as talk application, the user profile item obtained in step S13 is " name of contact person: cuckoo blue or green "; Subscriber equipment, according to being applied as talk application, is determined to need name of contact person to present to user, then name of contact person " cuckoo is blue or green " is presented to user by subscriber equipment.
Again such as, recognition result information to be used be applied as outlook mailbox, the user profile item obtained in step S13 is " name of contact person: cuckoo blue or green "; Subscriber equipment is according to being applied as outlook mailbox, determine that it needs to obtain mailbox message corresponding to contact person, and be supplied to outlook, then subscriber equipment obtains the mailbox message " duyuqingxiaoi.com " of contact person " cuckoo is blue or green ", and this mailbox message is supplied to outlook mailbox, perform its on-unit for outlook mailbox.
Preferably, the corresponding operating that the user profile item that described subscriber equipment pair and recognition result content information match performs includes but not limited to following at least one item:
1) mode that the application of recognition result information to be used described in the described user profile Xiang Yiyu matched is associated is presented to user.
Such as, the user profile item matched comprises contact name, recognition result information to be used be applied as talk application; Then when talk application is when dialling, the contact name comprised in user profile item is presented to user by subscriber equipment.
Again such as, the user profile item matched comprises dining room title, and the map inquiry that is applied as of recognition result information to be used is applied; Then when map inquiry application query obtains dining room particular location, the dining room title comprised in user profile item is presented to user by subscriber equipment.
2) subscriber equipment obtains other user profile items be associated with the described user profile item matched, in order to described application on-unit.
Preferably, the storage incidence relation etc. that subscriber equipment can be existed in a user device by the type of user profile item, user profile item, determines and other user profile items that the described user profile item matched is associated.
Such as, the user profile item matched is " name of contact person: cuckoo blue or green ", recognition result information to be used be applied as outlook mailbox; Then subscriber equipment is according to being applied as outlook mailbox, acquisition is stored in the contact person's mailbox " duyuqingxiaoi.com " in same contact data volume with name of contact person " cuckoo is blue or green ", and outlook mailbox will be supplied to contact person's mailbox, operate in order to its pending mail sending.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any application according to described recognition result information to be used, the user profile item matched with recognition result content information is performed to the implementation of corresponding operating, all should be within the scope of the present invention.
As one of the preferred version of the present embodiment, recognition result information also comprises the classified information of at least one sound bite, the step S15 that the present invention performs before being also included in step S13, and step S13 comprises step S13 ' further.
In step S15, subscriber equipment, according to the classified information in recognition result information, determines application to be launched, as the application of described recognition result information to be used.
Particularly, subscriber equipment, according to classified information, determines application to be launched, and the mode as the application of described recognition result information to be used includes but not limited to:
1) subscriber equipment is directly according to classified information, determines application to be launched.
Such as, subscriber equipment obtain in step s 12 sound bite " to ", " cuckoo blue or green " and the classified information of " making a phone call " be respectively " common wordss ", " name " and " dialing ", then subscriber equipment is according to above-mentioned three classified informations, inquire about in the classified information of pre-determining and the mapping table of application, obtain the application " talk application " corresponding with classified information " dialing ", as application to be launched.
2) subscriber equipment is according to classified information, determines the application type of application to be launched, and according to application type, determines application to be launched.
Such as, subscriber equipment obtain in step s 12 sound bite " to ", the classified information of " cuckoo blue or green " and " mail " is respectively " common wordss ", " name " and " mail "; Then subscriber equipment is according to above-mentioned three classified informations, inquires about in the classified information of pre-determining and the mapping table of application type, and obtaining the application type corresponding with classified information " mail " is mailbox; Then, the outlook mailbox given tacit consent in this application type selected by subscriber equipment, as application to be launched.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any according to the classified information in recognition result information, determine application to be launched, as the implementation of the application of described recognition result information to be used, all should be within the scope of the present invention.
In step S13 ', subscriber equipment is according to recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the local user's information bank be associated with described application to be launched.
Wherein, that the local user's information bank be associated with described application to be launched comprises pre-determining, there is incidence relation with application to be launched local user's information bank.
Such as, the local user's information bank be associated with talk application comprises local address book, apply with outlook mailbox the local user's information bank be associated and comprise local mailbox associated person information, apply with map inquiry the local user's information bank be associated and comprise user and commonly use dining room name information and user commonly uses information of place names etc.
Wherein, in this step, subscriber equipment is according to recognition result content information, matching inquiry is carried out in the local user's information bank be associated with described application to be launched, to obtain the mode of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, with subscriber equipment in step S13 according to recognition result content information, matching inquiry is carried out in local user's information bank, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
Preferably, the present invention also comprised step S16 before step S13, and abovementioned steps S13 ' comprises step S13 ".
In step s 16, subscriber equipment, according to classified information, determines the recognition result content information needing to carry out described matching inquiry in the recognition result content information of at least one sound bite described.
Particularly, subscriber equipment, according to classified information, determines that the mode needing to carry out the recognition result content information of described matching inquiry in the recognition result content information of at least one sound bite described includes but not limited to:
1) when the recognition result content information comprising the sound bite for identifying its correspondence in classified information is the need of when carrying out the identification information of described matching inquiry, subscriber equipment, directly according to this identification information, determines the recognition result content information needing to carry out matching inquiry at least one sound bite.
Such as, identification information " 1 " needs to carry out matching inquiry for mark, and identification information " 0 " is for identifying without the need to carrying out matching inquiry.Subscriber equipment obtain in step s 12 sound bite " to ", " cuckoo blue or green " and the classified information of " making a phone call " be respectively " common wordss; 0 ", " name; 1 " and " dialing; 0 ", then subscriber equipment directly comprises identification information " 1 " according in the classified information of sound bite " cuckoo is blue or green ", determines that the recognition result content information of this sound bite needs to carry out described matching inquiry.
Again such as, sound bite corresponding to the predetermined classified information comprising identification information " 1 " needs to carry out matching inquiry, does not comprise sound bite corresponding to the classified information of identification information " 1 " without the need to carrying out matching inquiry.Subscriber equipment obtain in step s 12 sound bite " to ", " cuckoo blue or green " and the classified information of " making a phone call " be respectively " common wordss ", " name; 1 " and " dialing ", then subscriber equipment directly comprises identification information " 1 " according in the classified information of sound bite " cuckoo is blue or green ", determines that this sound bite needs to carry out described matching inquiry.
2) subscriber equipment is according to classified information, pre-determining, inquire about, to determine the recognition result content information needing to carry out matching inquiry at least one sound bite in the classified information storehouse that needs to carry out matching inquiry.
Such as, subscriber equipment obtain in step s 12 sound bite " to ", " cuckoo blue or green " and the classified information of " making a phone call " be respectively " common wordss ", " name " and " dialing ", then subscriber equipment is according to above-mentioned three classified informations, pre-determining, inquire about in the classified information storehouse that needs to carry out matching inquiry, and inquiry obtains classified information " name ", then the recognition result content information of the sound bite " cuckoo is blue or green " that subscriber equipment determination classified information " name " is corresponding needs to carry out matching inquiry.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any according to classified information, determine the implementation needing to carry out the recognition result content information of described matching inquiry in the recognition result content information of at least one sound bite described, all should be within the scope of the present invention.
In step S13 " in; subscriber equipment carries out the recognition result content information of matching inquiry as required; in the local user's information bank be associated with described application to be launched, carry out matching inquiry, to obtain and the described user profile item needing the recognition result content information carrying out matching inquiry to match.
Wherein, the recognition result content information of matching inquiry is carried out as required in this step, matching inquiry is carried out in the local user's information bank be associated with described application to be launched, to obtain the mode with the described user profile item needing the recognition result content information carrying out matching inquiry to match, with subscriber equipment in step S13 according to recognition result content information, matching inquiry is carried out in local user's information bank, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
The invention enables subscriber equipment according to local user's information bank, local error correction can be carried out to the information of network equipment speech recognition gained, improve the accuracy rate of speech recognition; In addition, because error-correction operation is performed by subscriber equipment, therefore reduce the burden of the network equipment, and because user directly upgrades its user information database usually on a user device, therefore, carry out speech recognition based on local user's information bank, can ensure error-correction operation based on user profile be up-to-date.
Fig. 2 is the method flow diagram for carrying out speech recognition of another preferred embodiment of the present invention.The present embodiment comprises step S11, step S21, step S22, step S23, step S24, step S17 and step S18.
In step s 11, subscriber equipment sends speech message to be identified to the network equipment.
Then, in the step s 21, the network equipment obtains the voice messaging to be identified from described subscriber equipment.
Then, in step S22, the network equipment carries out speech recognition to described voice messaging to be identified, and obtain recognition result information, wherein, described recognition result information comprises the recognition result content information of at least one sound bite in described voice messaging to be identified.
Wherein, this step is described in detail with reference to the embodiment shown in FIG. 1, does not repeat them here.
Then, in step S23, the network equipment is according to described recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the network subscriber information storehouse of user using described subscriber equipment.
Wherein, the network equipment according to the identity information from subscriber equipment, described user, can determine the network subscriber information storehouse of the user using described subscriber equipment.Preferably, following at least one the information that the identity information of described user can provide according to subscriber equipment is determined:
1) identification information of subscriber equipment;
Such as, the chip serial number of subscriber equipment; User device system sequence number; The mobile identification number of subscriber equipment, as cell-phone number etc.
2) log-on message of user;
Such as, user ID and password etc.
Wherein, the network subscriber information storehouse of user can be synchronized in the network equipment by subscriber equipment, and such as, when subscriber equipment access network, local user's information bank is synchronized in the network equipment by subscriber equipment automatically; Such as, or the network subscriber information storehouse of user can directly be set up in the network device by user or upgrade, and, subscriber equipment is created or the amendment page by the network subscriber information storehouse that the network equipment provides, and sets up or upgrades its network subscriber information storehouse etc.
Particularly, the network equipment is according to the user identity determined, obtain its network subscriber information storehouse, and according to described recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in network subscriber information storehouse.
Wherein, in this step, the network equipment is according to recognition result content information, matching inquiry is carried out in network subscriber information storehouse, to obtain the mode of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, with subscriber equipment in step S13 according to recognition result content information, matching inquiry is carried out in local user's information bank, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
Preferably, in this step, the network equipment, according to the Word message in recognition result content information, inquires about the user profile item matched in network subscriber information storehouse; And when fail the user profile item inquiring described coupling time, according to described Pinyin information, in the network subscriber information storehouse of user using subscriber equipment, matching inquiry is carried out, to obtain the user profile item matched with the Pinyin information of all or part of sound bite at least one sound bite described described.
Then, in step s 24 which, recognition result content information and user profile item are supplied to described subscriber equipment by the network equipment.
Then, in step S17, subscriber equipment receives recognition result information that the network equipment feeds back, voice messaging to be identified and user profile item.
Then, in step S18, subscriber equipment, according to the application of recognition result content information to be used, performs corresponding operating to user profile item.
Wherein, this step is same or similar with step S14, does not repeat them here.
Wherein, when corresponding operating comprises other user profile items obtaining and be associated with user profile item, subscriber equipment is by its local user's information bank, obtain other user profile items that this is associated, or, subscriber equipment can send request to the network equipment, other user profile items obtaining this to ask the network equipment in network subscriber information storehouse and be associated, and is supplied to subscriber equipment.
As one of the preferred version of the present embodiment, recognition result content information also comprises the classified information of at least one sound bite, the method of the present embodiment also comprises step S25, step S23 comprises step S23 ', step S24 comprises step S24 ', step S17 comprises step S17 ', and step S18 comprises step S18 '.
In step s 25, the network equipment, according to classified information, determines the application of described recognition result information to be used in described subscriber equipment.
Wherein, in this step, the network equipment is according to classified information, determine the mode of the application of described recognition result information to be used in described subscriber equipment, with subscriber equipment in step S15 according to classified information, determine that the mode of the application of described recognition result information to be used in subscriber equipment is same or similar, do not repeat them here.
In step S23 ', the network equipment is according to described recognition result content information, use the user of described subscriber equipment, carry out matching inquiry, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the network subscriber information storehouse that is associated with described application.
Wherein, in this step, the network equipment is according to described recognition result content information, using the user of described subscriber equipment, matching inquiry is carried out in the network subscriber information storehouse be associated with described application, to obtain the mode of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, with the network equipment in step S23 according to described recognition result content information, matching inquiry is carried out in the network subscriber information storehouse of user using described subscriber equipment, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
Preferably, the network equipment also can according to determined application, other user profile items be associated with the described user profile item matched by acquisition in network subscriber information storehouse.
In step S24 ', the identification information of described recognition result content information, described user profile item and described application is supplied to described subscriber equipment by the network equipment.
Preferably, the identification information of described recognition result content information, described user profile item, other user profile items be associated with this user profile item and described application is supplied to described subscriber equipment by the network equipment.
Then, in step S17 ', subscriber equipment receive network equipment feedback, the identification information of the application of recognition result content information, user profile item and described recognition result information to be used.
Preferably, subscriber equipment receive described recognition result content information, described user profile item, other user profile items be associated with this user profile item and described application identification information be supplied to described subscriber equipment.
Then, in step S18 ', subscriber equipment, according to identification information, determines the application of recognition result content information to be used, and according to this application, performs corresponding operating to user profile item.
Preferably, the present embodiment also comprised step S26 before step S23, and step S23 ' comprises step S23 ".
In step S26, the network equipment, according to classified information, determines the recognition result content information needing to carry out described matching inquiry in the recognition result content information of at least one sound bite described.
Wherein, in this step, the network equipment is according to classified information, determine the mode needing to carry out the recognition result content information of described matching inquiry in the recognition result content information of at least one sound bite described, with subscriber equipment in step S16 according to classified information, determine that the mode needing to carry out the recognition result content information of described matching inquiry in the recognition result content information of at least one sound bite described is same or similar, do not repeat them here.
In step S23 " in; the network equipment carries out the recognition result content information of matching inquiry as required; use the user of described subscriber equipment, carry out matching inquiry, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the network subscriber information storehouse that is associated with described application.
Wherein, in this step, the network equipment carries out the recognition result content information of matching inquiry as required, using the user of described subscriber equipment, matching inquiry is carried out in the network subscriber information storehouse be associated with described application, to obtain the mode of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, with step S13 " in subscriber equipment carry out the recognition result content information of matching inquiry as required, using the user of described subscriber equipment, matching inquiry is carried out in the local user's information bank be associated with described application, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
The present embodiment makes the network equipment according to network subscriber information storehouse, can identify that the recognition result content information obtained carries out error correction to it, can provide personalized recognition result to user.
Fig. 3 is the system architecture schematic diagram for carrying out speech recognition of a preferred embodiment of the invention.The system of the present embodiment comprises subscriber equipment and the network equipment; Wherein, subscriber equipment comprises the recognition device for performing operation of the present invention, and this recognition device comprises dispensing device 11, receiving trap 12, first matching inquiry device 13 and actuating unit 14.
Dispensing device 11 in subscriber equipment sends voice messaging to be identified to the network equipment.
Wherein, dispensing device 11 can receive the voice messaging to be identified from user, and sends to the network equipment.
Such as, the voice messaging to be identified that dispensing device 11 obtains from user " is made a phone call to cuckoo green grass or young crops ", and sends to the network equipment.
Then, the network equipment obtains the voice messaging to be identified from subscriber equipment.
Such as, the network equipment receives and " makes a phone call to cuckoo green grass or young crops " from the voice messaging of subscriber equipment.
Then, the network equipment carries out speech recognition to this voice messaging to be identified, and obtain recognition result information, wherein, described recognition result information comprises the recognition result content information of at least one sound bite in this speech message to be identified.
Preferably, the recognition result content information of a sound bite includes but not limited to following at least one item:
1) Word message of this sound bite;
Such as, this sound bite is carried out to the Chinese character information, English word information etc. of speech recognition gained.
It should be noted that, when a sound bite can identify obtain multiple pronunciation same or analogous Word message time, the network equipment can by wherein selecting one or more Word message as recognition result content information or its part, or the Word message that all identification obtains by the network equipment is as recognition result content information or its part.
2) Pinyin information of this sound bite;
Such as, this sound bite " cuckoo is blue or green " is carried out to the Pinyin information " duyuqing " etc. of speech recognition gained.
3) the speech waveform information of this sound bite;
Such as, the network equipment extracts the speech waveform information of the sound bite " cuckoo is blue or green " of its None-identified in " being made a phone call to cuckoo green grass or young crops " by voice messaging, as recognition result content information or its part.
Particularly, the network equipment, based on the voice messaging storehouse of pre-determining, identifies voice messaging to be identified, and obtains the recognition result content information of at least one sound bite in this voice messaging to be identified.
Such as, the network equipment is based on the speech message storehouse of pre-determining, speech message to be identified " is made a phone call to cuckoo green grass or young crops " and identifies, obtain Word message that sound bite " gives " " to ", the Word message that the Word message " Du Yuqin " of sound bite " cuckoo green grass or young crops " and Pinyin information " duyuqing " and sound bite " are made a phone call " " is made a phone call ".
Preferably, the network equipment carries out the classified information also comprising at least one sound bite in the recognition result information of speech recognition gained to voice messaging.Wherein, this classified information is for identifying the type of sound bite, and such as, sound bite belongs to name, place name, dialing class, mail class etc.
Wherein, the network equipment obtains the classified information of sound bite by various ways.Such as, when inquiring the information matched with sound bite in voice messaging storehouse, directly obtain the classified information of this information, as the classified information of sound bite; Again such as, the network equipment by carrying out semantic analysis to the text message of speech recognition gained, thus determines the classified information etc. of the sound bite that text information is corresponding.
More preferably, the recognition result content information comprising the sound bite for identifying its correspondence in this classified information is the need of by the identification information carrying out subscriber equipment and carry out local matching inquiry, the network equipment according to the classification of determined sound bite, can determine described identification information.Such as, predtermined category " name ", " place name " needs to carry out local matching inquiry by carrying out subscriber equipment, then when network equipment determination sound bite be categorized as " name " or " place name " time, the recognition result content information needs adding the sound bite for identifying its correspondence in classified information carry out the identification information of local matching inquiry by carrying out subscriber equipment, when network equipment determination sound bite be categorized as other classification time, the recognition result content information adding the sound bite for identifying its correspondence in classified information does not need by the identification information carrying out subscriber equipment and carry out local matching inquiry, or, the network equipment does not add identification information in classified information.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any speech recognition is carried out to voice messaging to be identified, obtain the implementation comprising the recognition result information of the recognition result content information of wherein at least one sound bite, all should be within the scope of the present invention.
Then, recognition result information is sent to subscriber equipment by the network equipment.
Then, the receiving trap 12 in subscriber equipment receives recognition result information that the network equipment feeds back, described voice messaging to be identified.
Then, the first matching inquiry device 13 in subscriber equipment is according to recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in local user's information bank.
Wherein, described local user's information bank comprises in the external storage equipment being stored in described subscriber equipment or this subscriber equipment, and for the information bank of storing subscriber information; Preferably, this local user's information bank can be used as an entirety, store all user profile, or, this local user's information bank comprises multiple independently user information database, and as comprised subscriber phone associated person information storehouse, user MSN associated person information storehouse, user commonly use information of place names storehouse, user commonly uses name information storehouse, dining room etc.
Wherein, described user profile item comprises an information of user; Such as, name of contact person, contact person's mailbox, contact phone, user commonly use place name, user commonly uses dining room title etc.
Particularly, first matching inquiry device 13 is according to recognition result content information, in local user's information bank, carry out matching inquiry, include but not limited in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described:
1) the first matching inquiry device 13 is respectively according to the recognition result content information of all sound bites, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite in all sound bites in local user's information bank.
Such as, receiving trap 12 obtain sound bite " to " Word message " to ", the Word message that the Word message " Du Yuqin " of sound bite " cuckoo blue or green " and Pinyin information " duyuqing " and sound bite " are made a phone call " " is made a phone call "; Then the first matching inquiry device 13 is respectively according to the recognition result content information of above-mentioned three sound bites, in local user's information bank, carry out matching inquiry, and only obtain the user profile item " name of contact person: cuckoo is blue or green " matched with the Word message " Du Yuqin " of sound bite " cuckoo is blue or green " and Pinyin information " duyuqing ".
Again such as, receiving trap 12 obtain sound bite " to " Word message " to ", the Word message that the speech waveform information of sound bite " cuckoo blue or green " and sound bite " are made a phone call " " is made a phone call ", then the first matching inquiry device 13 " is given " according to sound bite and the recognition result content information of " making a phone call " respectively, at local user's information bank, store in the user information database of text message and carry out matching inquiry, do not obtain the user profile item matched, and, first matching inquiry device 13 is according to sound bite " cuckoo blue or green " at local user's information bank, store in the user information database of voice messaging and carry out matching inquiry, and determine with the user profile item of the speech waveform information match of sound bite " cuckoo is blue or green " to be " name of contact person: cuckoo is blue or green ".Wherein, the speech waveform information in local user's information bank can from user, and, the text message that this speech waveform information is corresponding, or can be arranged by user with the user profile item of this speech waveform information match.
2) the first matching inquiry device 13 selects part sound bite by all sound bites, and according to the recognition result content information of selected sound bite, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite in selected sound bite in local user's information bank.
Such as, receiving trap 12 obtain sound bite " to " Word message " to ", the Word message that the Word message " Du Yuqin " of sound bite " cuckoo blue or green " and Pinyin information " duyuqing " and sound bite " are made a phone call " " is made a phone call ", then the first matching inquiry device 13 is based on universal word storehouse, judge Word message " to " and " making a phone call " as universal word, without the need to performing the operation of local matching inquiry to it, then, first matching inquiry device 13 is not judged as Word message " Du Yuqin " and the Pinyin information " duyuqing " of the sound bite " cuckoo is blue or green " of universal word according to its Word message, matching inquiry is carried out in local user's information bank, and only obtain the user profile item " name of contact person: cuckoo is blue or green " matched with the Word message " Du Yuqin " of sound bite " cuckoo is blue or green " and Pinyin information " duyuqing ".
3) the first matching inquiry device 13 is according to recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the local user's information bank be associated with application pre-determining, to be launched.This implementation will be described in detail in follow-up preferred version, not repeat them here.
Preferably, recognition result content information comprises Word message and Pinyin information, and the first matching inquiry device 13 comprises the first characters matching inquiry unit (not shown) and the first phonetic matching inquiry device (not shown) further.First characters matching inquiry unit is according to described Word message, the user profile item matched is inquired about in described local user's information bank, and when the first characters matching inquiry unit fails the user profile item inquiring described coupling, first phonetic matching inquiry device is according to described Pinyin information, matching inquiry is carried out, to obtain the user profile item matched with the Pinyin information of all or part of sound bite at least one sound bite described in described local user's information bank.
Such as, for Word message " Du Yuqin " and the Pinyin information " duyuqing " of the sound bite " cuckoo is blue or green " of receiving trap 12 acquisition, first characters matching inquiry unit is inquired about in local user's information bank according to Word message " Du Yuqin ", and fails to inquire the user profile item with this word information match; Then, the first phonetic matching inquiry device carries out matching inquiry according to Pinyin information " duyuqing " in local user's information bank, obtains the user profile item " name of contact person: cuckoo is blue or green " matched with Pinyin information " duyuqing ".
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any according to recognition result content information, matching inquiry is carried out in local user's information bank, to obtain the implementation of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, such as, respectively according to the Word message of a sound bite, Pinyin information and speech waveform information are inquired about in local user's information bank, and when inquiring multiple user profile item respectively with above-mentioned three information matches, by wherein selecting the user profile item that user's frequency of utilization is the highest, or the plurality of user profile item matched is presented to user for its selection etc., all should be within the scope of the present invention.
Then, the actuating unit 14 in subscriber equipment, according to the application of described recognition result information to be used, performs corresponding operating to the user profile item matched with recognition result content information.
Wherein, the application of described described recognition result information to be used comprises and anyly may be defined as by subscriber equipment the application that needs to use described recognition result information.Preferably, the application of this recognition result information to be used includes but not limited to:
1) the current application being in active state in subscriber equipment;
Such as, in subscriber equipment current started and be in active state mailbox application; Again such as, current to the talk application etc. that another subscriber equipment dials in subscriber equipment.
2) application to be launched determined according to recognition result content information of subscriber equipment;
Such as, the order lexicon of pre-determining is stored in subscriber equipment, wherein, this order lexicon stores commonly used command vocabulary and application corresponding with each commonly used command vocabulary respectively, receiving trap 12 obtain sound bite " to " Word message " to ", the Word message that the Word message " Du Yuqin " of sound bite " cuckoo blue or green " and Pinyin information " duyuqing " and sound bite " are made a phone call " " is made a phone call "; Then subscriber equipment is according to the text message of above-mentioned three sound bites, inquire about in the order lexicon of pre-determining, and determine corresponding with text message " phone " to be applied as talk application, then subscriber equipment is using the application of talk application as described recognition result information to be used.
3) also comprise the classified information of at least one sound bite in recognition result information, subscriber equipment, according to this classified information, determines application to be launched, as the application of described recognition result information to be used.This implementation will be described in detail in follow-up preferred version, not repeat them here.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, anyly may be defined as by subscriber equipment the application that needs to use described recognition result information, all should be included in the scope of the application of recognition result information to be used of the present invention.
Particularly, actuating unit 14 can in several ways, carrys out the application according to recognition result information to be used, performs corresponding operating to the user profile item matched with recognition result content information.
Such as, recognition result information to be used be applied as talk application, the user profile item that the first matching inquiry device 13 obtains be " name of contact person: cuckoo green grass or young crops "; Actuating unit 14, according to being applied as talk application, is determined to need name of contact person to present to user, then name of contact person " cuckoo is blue or green " is presented to user by actuating unit 14.
Again such as, recognition result information to be used be applied as outlook mailbox, the user profile item that the first matching inquiry device 13 obtains be " name of contact person: cuckoo green grass or young crops "; Actuating unit 14 is according to being applied as outlook mailbox, determine that it needs to obtain mailbox message corresponding to contact person, and be supplied to outlook, then actuating unit 14 obtains the mailbox message " duyuqingxiaoi.com " of contact person " cuckoo is blue or green ", and this mailbox message is supplied to outlook mailbox, perform its on-unit for outlook mailbox.
Preferably, actuating unit 14 includes but not limited to following at least one item to the corresponding operating that the user profile item matched with recognition result content information performs:
1) mode that the application of recognition result information to be used described in the described user profile Xiang Yiyu matched is associated is presented to user by actuating unit 14.
Such as, the user profile item matched comprises contact name, recognition result information to be used be applied as talk application; Then when talk application is when dialling, the contact name comprised in user profile item is presented to user by actuating unit 14.
Again such as, the user profile item matched comprises dining room title, and the map inquiry that is applied as of recognition result information to be used is applied; Then when map inquiry application query obtains dining room particular location, the dining room title comprised in user profile item is presented to user by actuating unit 14.
2) actuating unit 14 obtains other user profile items be associated with the described user profile item matched, in order to described application on-unit.
Preferably, the storage incidence relation etc. that actuating unit 14 can be existed in a user device by the type of user profile item, user profile item, determines and other user profile items that the described user profile item matched is associated.
Such as, the user profile item matched is " name of contact person: cuckoo blue or green ", recognition result information to be used be applied as outlook mailbox; Then actuating unit 14 is according to being applied as outlook mailbox, acquisition is stored in the contact person's mailbox " duyuqingxiaoi.com " in same contact data volume with name of contact person " cuckoo is blue or green ", and outlook mailbox will be supplied to contact person's mailbox, operate in order to its pending mail sending.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any application according to described recognition result information to be used, the user profile item matched with recognition result content information is performed to the implementation of corresponding operating, all should be within the scope of the present invention.
As one of the preferred version of the present embodiment, recognition result information also comprises the classified information of at least one sound bite, recognition device also comprises the first application determining device (not shown), and the first matching inquiry device comprises the first sub-matching inquiry device (not shown) further.
First application determining device, according to the classified information in recognition result information, determines application to be launched, as the application of described recognition result information to be used.
Particularly, the first application determining device is according to classified information, and determine application to be launched, the mode as the application of described recognition result information to be used includes but not limited to:
1) the first application determining device is directly according to classified information, determines application to be launched.
Such as, receiving trap 12 obtain sound bite " to ", " cuckoo blue or green " and the classified information of " making a phone call " be respectively " common wordss ", " name " and " dialing ", then the first application determining device is according to above-mentioned three classified informations, inquire about in the classified information of pre-determining and the mapping table of application, obtain the application " talk application " corresponding with classified information " dialing ", as application to be launched.
2) the first application determining device is according to classified information, determines the application type of application to be launched, and according to application type, determines application to be launched.
Such as, receiving trap 12 obtain sound bite " to ", the classified information of " cuckoo blue or green " and " mail " is respectively " common wordss ", " name " and " mail "; Then the first application determining device is according to above-mentioned three classified informations, and inquire about in the classified information of pre-determining and the mapping table of application type, obtaining the application type corresponding with classified information " mail " is mailbox; Then, the first application determining device selects the outlook mailbox given tacit consent in this application type, as application to be launched.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any according to the classified information in recognition result information, determine application to be launched, as the implementation of the application of described recognition result information to be used, all should be within the scope of the present invention.
First sub-matching inquiry device is according to recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the local user's information bank be associated with described application to be launched.
Wherein, that the local user's information bank be associated with described application to be launched comprises pre-determining, there is incidence relation with application to be launched local user's information bank.
Such as, the local user's information bank be associated with talk application comprises local address book, apply with outlook mailbox the local user's information bank be associated and comprise local mailbox associated person information, apply with map inquiry the local user's information bank be associated and comprise user and commonly use dining room name information and user commonly uses information of place names etc.
Wherein, first sub-matching inquiry device is according to recognition result content information, matching inquiry is carried out in the local user's information bank be associated with described application to be launched, to obtain the mode of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, with the first matching inquiry device according to recognition result content information, matching inquiry is carried out in local user's information bank, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
Preferably, recognition device also comprises the first match objects determining device (not shown), and aforementioned first word matching inquiry device comprises the second sub-matching inquiry device (not shown).
First match objects determining device, according to classified information, determines the recognition result content information needing to carry out described matching inquiry in the recognition result content information of at least one sound bite of speech message to be identified.
Particularly, the first match objects determining device, according to classified information, determines that the mode needing to carry out the recognition result content information of described matching inquiry in the recognition result content information of at least one sound bite described includes but not limited to:
1) when the recognition result content information comprising the sound bite for identifying its correspondence in classified information is the need of when carrying out the identification information of described matching inquiry, first match objects determining device, directly according to this identification information, determines the recognition result content information needing to carry out matching inquiry at least one sound bite.
Such as, identification information " 1 " needs to carry out matching inquiry for mark, and identification information " 0 " is for identifying without the need to carrying out matching inquiry.Receiving trap 12 obtain sound bite " to ", " cuckoo blue or green " and the classified information of " making a phone call " be respectively " common wordss; 0 ", " name; 1 " and " dialing; 0 ", then the first match objects determining device directly comprises identification information " 1 " according in the classified information of sound bite " cuckoo is blue or green ", determines that the recognition result content information of this sound bite needs to carry out described matching inquiry.
Again such as, sound bite corresponding to the predetermined classified information comprising identification information " 1 " needs to carry out matching inquiry, does not comprise sound bite corresponding to the classified information of identification information " 1 " without the need to carrying out matching inquiry.Receiving trap 12 obtain sound bite " to ", " cuckoo blue or green " and the classified information of " making a phone call " be respectively " common wordss ", " name; 1 " and " dialing ", then the first match objects determining device directly comprises identification information " 1 " according in the classified information of sound bite " cuckoo is blue or green ", determines that this sound bite needs to carry out described matching inquiry.
2) the first match objects determining device is according to classified information, pre-determining, inquire about, to determine the recognition result content information needing to carry out matching inquiry at least one sound bite in the classified information storehouse that needs to carry out matching inquiry.
Such as, receiving trap 12 obtain sound bite " to ", " cuckoo blue or green " and the classified information of " making a phone call " be respectively " common wordss ", " name " and " dialing ", then the first match objects determining device is according to above-mentioned three classified informations, pre-determining, inquire about in the classified information storehouse that needs to carry out matching inquiry, and inquiry obtains classified information " name ", then the recognition result content information of the sound bite " cuckoo is blue or green " that the first match objects determining device determination classified information " name " is corresponding needs to carry out matching inquiry.
It should be noted that, above-mentioned citing is only and technical scheme of the present invention is described better, but not limitation of the present invention, those skilled in the art should understand that, any according to classified information, determine the implementation needing to carry out the recognition result content information of described matching inquiry in the recognition result content information of at least one sound bite described, all should be within the scope of the present invention.
Second sub-matching inquiry device carries out the recognition result content information of matching inquiry as required, matching inquiry is carried out, to obtain and the described user profile item needing the recognition result content information carrying out matching inquiry to match in the local user's information bank be associated with described application to be launched.
Wherein, second sub-matching inquiry device carries out the recognition result content information of matching inquiry as required, matching inquiry is carried out in the local user's information bank be associated with described application to be launched, to obtain the mode with the described user profile item needing the recognition result content information carrying out matching inquiry to match, with the first matching inquiry device 13 according to recognition result content information, matching inquiry is carried out in local user's information bank, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
The invention enables subscriber equipment according to local user's information bank, local error correction can be carried out to the information of network equipment speech recognition gained, improve the accuracy rate of speech recognition; In addition, because error-correction operation is performed by subscriber equipment, therefore reduce the burden of the network equipment, and because user directly upgrades its user information database usually on a user device, therefore, carry out speech recognition based on local user's information bank, can ensure error-correction operation based on user profile be up-to-date.
Fig. 4 is the system architecture schematic diagram for carrying out speech recognition of another preferred embodiment of the present invention.The system of the present embodiment comprises subscriber equipment and the network equipment; Wherein, the network equipment comprises the servicing unit for performing the present invention's operation, and this servicing unit comprises voice acquisition device 21, speech recognition equipment 22, second matching inquiry device 23 and generator 24.
First, subscriber equipment sends speech message to be identified to the network equipment.
Then, the voice acquisition device 21 in the network equipment obtains the voice messaging to be identified from described subscriber equipment.
Then, speech recognition equipment 22 in the network equipment carries out speech recognition to described voice messaging to be identified, obtain recognition result information, wherein, described recognition result information comprises the recognition result content information of at least one sound bite in described voice messaging to be identified.
Wherein, speech recognition equipment 22 carries out speech recognition to described voice messaging to be identified, obtain the mode of recognition result information, with with reference to the middle network equipment embodiment illustrated in fig. 3, speech recognition is carried out to described voice messaging to be identified, the mode obtaining recognition result information is same or similar, does not repeat them here.
Then, the second matching inquiry device 23 in the network equipment is according to described recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the network subscriber information storehouse of user using described subscriber equipment.
Wherein, the second matching inquiry device 23 according to the identity information from subscriber equipment, described user, can determine the network subscriber information storehouse of the user using described subscriber equipment.Preferably, following at least one the information that the identity information of described user can provide according to subscriber equipment is determined:
1) identification information of subscriber equipment;
Such as, the chip serial number of subscriber equipment; User device system sequence number; The mobile identification number of subscriber equipment, as cell-phone number etc.
2) log-on message of user;
Such as, user ID and password etc.
Wherein, the network subscriber information storehouse of user can be synchronized in the network equipment by subscriber equipment, and such as, when subscriber equipment access network, local user's information bank is synchronized in the network equipment by subscriber equipment automatically; Such as, or the network subscriber information storehouse of user can directly be set up in the network device by user or upgrade, and, subscriber equipment is created or the amendment page by the network subscriber information storehouse that the network equipment provides, and sets up or upgrades its network subscriber information storehouse etc.
Particularly, second matching inquiry device 23 is according to the user identity determined, obtain its network subscriber information storehouse, and according to described recognition result content information, matching inquiry is carried out, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in network subscriber information storehouse.
Wherein, second matching inquiry device 23 is according to recognition result content information, matching inquiry is carried out in network subscriber information storehouse, to obtain the mode of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, with the first matching inquiry device 13 according to recognition result content information, matching inquiry is carried out in local user's information bank, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
Preferably, the second matching inquiry device 23 comprises the second characters matching inquiry unit (not shown) and the second phonetic matching inquiry device (not shown) further.Second characters matching inquiry unit, according to the Word message in recognition result content information, inquires about the user profile item matched in network subscriber information storehouse; When fail the user profile item inquiring described coupling time, second phonetic matching inquiry device is according to described Pinyin information, in the network subscriber information storehouse of user using subscriber equipment, matching inquiry is carried out, to obtain the user profile item matched with the Pinyin information of all or part of sound bite at least one sound bite described described.
Then, recognition result content information and user profile item are supplied to described subscriber equipment by the generator 24 in the network equipment.
Then, subscriber equipment receives recognition result information that the network equipment feeds back, voice messaging to be identified and user profile item.
Then, subscriber equipment, according to the application of recognition result content information to be used, performs corresponding operating to user profile item.
Wherein, subscriber equipment is according to the application of recognition result content information to be used, user profile item is performed to the mode of corresponding operating, with with reference to middle actuating unit 14 embodiment illustrated in fig. 3 according to the application of recognition result content information to be used, mode user profile item being performed to corresponding operating is same or similar, does not repeat them here.
Wherein, when corresponding operating comprises other user profile items obtaining and be associated with user profile item, subscriber equipment is by its local user's information bank, obtain other user profile items that this is associated, or, subscriber equipment can send request to the network equipment, other user profile items obtaining this to ask the network equipment in network subscriber information storehouse and be associated, and is supplied to subscriber equipment.
As one of the preferred version of the present embodiment, recognition result content information also comprises the classified information of at least one sound bite, the servicing unit of the present embodiment also comprises the second application determining device (not shown), second matching inquiry device 23 comprises the 3rd sub-matching inquiry device (not shown), and generator 24 comprises sub-generator (not shown).
The second application determining device in the network equipment, according to classified information, determines the application of described recognition result information to be used in described subscriber equipment.
Wherein, second application determining device is according to classified information, determine the mode of the application of described recognition result information to be used in described subscriber equipment, with with reference to first applying determining device subscriber equipment according to classified information in embodiment illustrated in fig. 3, determine that the mode of the application of described recognition result information to be used in subscriber equipment is same or similar, do not repeat them here.
In the network equipment the 3rd sub-matching inquiry device is according to described recognition result content information, use the user of described subscriber equipment, carry out matching inquiry, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the network subscriber information storehouse that is associated with described application.
Wherein, 3rd sub-matching inquiry device is according to described recognition result content information, using the user of described subscriber equipment, matching inquiry is carried out in the network subscriber information storehouse be associated with described application, to obtain the mode of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, with the second matching inquiry device 23 according to described recognition result content information, matching inquiry is carried out in the network subscriber information storehouse of user using described subscriber equipment, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
Preferably, the network equipment also can according to determined application, other user profile items be associated with the described user profile item matched by acquisition in network subscriber information storehouse.
The identification information of described recognition result content information, described user profile item and described application is supplied to described subscriber equipment by the sub-generator in the network equipment.
Preferably, the identification information of described recognition result content information, described user profile item, other user profile items be associated with this user profile item and described application is supplied to described subscriber equipment by sub-generator.
Then, subscriber equipment receive network equipment feedback, the identification information of the application of recognition result content information, user profile item and described recognition result information to be used.
Preferably, subscriber equipment receive described recognition result content information, described user profile item, other user profile items be associated with this user profile item and described application identification information be supplied to described subscriber equipment.
Then, subscriber equipment, according to identification information, determines the application of recognition result content information to be used, and according to this application, performs corresponding operating to user profile item.
Preferably, servicing unit also comprises the second match objects determining device (not shown), and the 3rd sub-matching inquiry device comprises the 4th sub-matching inquiry device (not shown).
The second match objects determining device in the network equipment, according to classified information, determines the recognition result content information needing to carry out described matching inquiry in the recognition result content information of at least one sound bite described.
Wherein, second match objects determining device is according to classified information, determine the mode needing to carry out the recognition result content information of described matching inquiry in the recognition result content information of at least one sound bite described, with with reference to the first match objects determining device in embodiment illustrated in fig. 3 according to classified information, determine that the mode needing to carry out the recognition result content information of described matching inquiry in the recognition result content information of at least one sound bite described is same or similar, do not repeat them here.
In the network equipment the 4th sub-matching inquiry device carries out the recognition result content information of matching inquiry as required, use the user of described subscriber equipment, carry out matching inquiry, to obtain the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described in the network subscriber information storehouse that is associated with described application.
Wherein, 4th sub-matching inquiry device carries out the recognition result content information of matching inquiry as required, using the user of described subscriber equipment, matching inquiry is carried out in the network subscriber information storehouse be associated with described application, to obtain the mode of the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, with the recognition result content information carrying out matching inquiry with reference to the second sub-matching inquiry device in embodiment illustrated in fig. 3 as required, using the user of described subscriber equipment, matching inquiry is carried out in the local user's information bank be associated with described application, same or similar in the mode obtaining the user profile item matched with the recognition result content information of all or part of sound bite at least one sound bite described, do not repeat them here.
The present embodiment makes the network equipment according to network subscriber information storehouse, can identify that the recognition result content information obtained carries out error correction to it, can provide personalized recognition result to user
It should be noted that, the present invention can be implemented in the assembly of software and/or software restraint, such as, special IC (ASIC) can be adopted, load on software program in general object computing machine or any other similar software and/or hardware device and realize.
Software program of the present invention can perform to realize step mentioned above or function by processor.Similarly, software program of the present invention (comprising relevant data structure) can be stored in computer readable recording medium storing program for performing, such as, and RAM storer, magnetic or CD-ROM driver or flexible plastic disc and similar devices.In addition, steps more of the present invention or function can adopt hardware to realize, such as, as coordinating with processor thus performing the circuit etc. of each function or step.
In addition, a part of the present invention can be applied to computer program, such as computer program instructions, when it is performed by computing machine, by the operation of this computing machine, can call or provide according to method of the present invention and/or technical scheme.And call the programmed instruction of method of the present invention, may be stored in fixing or moveable recording medium, and/or be transmitted by the data stream in broadcast or other signal bearing medias, and/or be stored in the working storage of the computer equipment run according to described programmed instruction.
To those skilled in the art, obviously the invention is not restricted to the details of above-mentioned one exemplary embodiment, and when not deviating from spirit of the present invention or essential characteristic, the present invention can be realized in other specific forms.Therefore, no matter from which point, all should embodiment be regarded as exemplary, and be nonrestrictive, scope of the present invention is limited by claims instead of above-mentioned explanation, and all changes be therefore intended in the implication of the equivalency by dropping on claim and scope are included in the present invention.Any Reference numeral in claim should be considered as the claim involved by limiting.In addition, obviously " comprising " one word do not get rid of other unit or step, odd number does not get rid of plural number.Multiple unit of stating in system claims or device also can be realized by software or hardware by a unit or device.First, second word such as grade is used for representing title, and does not represent any specific order.