CN1612209A - Method and device for voice recording telephone number entry - Google Patents
Method and device for voice recording telephone number entry Download PDFInfo
- Publication number
- CN1612209A CN1612209A CNA2003101082388A CN200310108238A CN1612209A CN 1612209 A CN1612209 A CN 1612209A CN A2003101082388 A CNA2003101082388 A CN A2003101082388A CN 200310108238 A CN200310108238 A CN 200310108238A CN 1612209 A CN1612209 A CN 1612209A
- Authority
- CN
- China
- Prior art keywords
- voice
- user
- telephone
- typing
- telephone number
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Telephonic Communication Services (AREA)
Abstract
The invention discloses a method and device of writing in telephone number entry by voice, and the method includes: resolving the phrase to be written in into separate characters, marking tones after writing in them, prompting the user to select in many voice identification candidate items, and finally inputting the telephone number in number key or voice mode. A voice telephone directory device for implementing the method is provided with CPU, multifunctional interface card, and voice write /output device, and integrated with a telephone and connected with the number keys of the telephone; and the middle calculating course of the device is stored in an internal memory, and the voice identification program and address directory data to be called are stored in a flash memory. The invention improves the accuracy and error allowance for voice telephone directory identification, avoiding the fussy course of searching telephone directory as the user makes a telephone call and thus bringing about the convenient to the user.
Description
Technical field
The present invention relates to method and the device thereof of calling party in a kind of telephone system, particularly relate to the method and the device thereof of a kind of employing voice mode typing telephone book entries (name/place name and telephone number).
Background technology
Landline telephone does not have special telephone directory book usually, and the simple telephone number storage function that is had just is mapped several telephone numbers and certain several fixing numeral key.When the telephone number quantity of required storage was a lot, the user was difficult to remember the corresponding relation of telephone number and numeral key, and the number storage mode of therefore existing landline telephone does not meet people's use habit.
Only a few has the landline telephone of telephone directory book, and the input of its telephone directory is imported relevant name, information such as place name by the mode of button or hand-written phonetic usually.Do not have the landline telephone of display screen for great majority, adopt the button input quite loaded down with trivial details, adopt the handwriting input mode then to need the extra very worthy handwriting pad that adds, increased the cost of landline telephone.Even the landline telephone that has has LCD, in the bad situation of light or for the people who has defective vision, as blind person, the elderly, it also is difficult using the mode of button or hand-written phonetic to import information such as name/place name.
The device that adopts voice mode typing phone number entry is as Chinese utility model patent, application number 00206200.3, publication number 2415556, name is called " speech recognition searches the dial telephone book ", realize that by an outer belting speech recognition searches the telephone number of dialing landline telephone, preferably resolves the problem of landline telephone stored telephone clauses and subclauses difficulty.This utility model is by central processing unit, dynamic storage, static memory, AD conversion, keyboard, microphone, Microspeaker, dial-up circuit is formed the speech recognition dial-up circuit, speech recognition dial-up circuit, display driving integrated circuit link to each other, can the voice name and the telephone number thereof of input be stored, discern, and control module shows and dialing automatically.But, the described telephone directory device of this utility model precision of identifying speech is not high, do not relate to yet and need the fault-tolerance considered in the speech recognition, if user's asophia, adhesion, when ambiguous or described telephone directory is preserved the akin label of a plurality of pronunciations, clauses and subclauses, this telephone directory very big mistake will occur to the speech recognition of name or place name.In addition, need the specific letter box that connects to be connected and to work between this telephone directory and the phone, also brought inconvenience to the user.
Summary of the invention
The objective of the invention is provides a kind of method and device thereof of improved voice typing phone number entry in order to overcome shortcomings such as existing landline telephone voice call book device is not high to the phone number entry accuracy of identification of voice typing, fault-tolerance is not strong.
To achieve these goals, the technical scheme that the present invention takes: a kind of method of voice typing phone number entry, in the result who receives the execution phonetic entry of user speech input back, described method comprises the following steps:
(a) name that the prompting user will be to be imported or ground noun phrase are decomposed into isolated word and read in a voice telephone number book device;
(b) the prompting user is this word filling tone after reading in the voice of described isolated word;
(c) a plurality of speech recognition candidates of providing according to certain rule of monitoring speech recognition device and feed back to the user, the prompting user selects one as a result of in above-mentioned a plurality of speech recognition candidates;
(d) after complete typing of phone number entry, control described voice telephone number book device and will form the name/place name of these clauses and subclauses and feed back to the user once more by phonetic synthesis output, the user confirms errorless back input telephone number.
A kind of voice telephone number book device of realizing said method, it is characterized in that being provided with in the described device central processing unit that speech data is handled, described central processing unit is by versatile interface card connection voice typing/output unit, connect telephone number key dish and telephone set by interface card, the intermediate operations process of described central processing unit is carried out in internal storage, and voice recognition processing program and the address-book data that need call are kept in the flash memory.
The invention has the beneficial effects as follows, describedly a kind ofly be used to improve the method for voice telephone number book accuracy of identification by rationally disconnected word, the tone of after phonetic entry, annotating, and step such as system ambiguous identification artificial screening net result, improved the precision and the fault-tolerance of speech recognition phone clauses and subclauses greatly.The voice telephone number book device of realizing the method for the invention is a kind of embedded system, can directly put into telephone set and not change original profile of phone and size.Described method and device thereof have changed the traditional approach that the user calls, make the user remove the complicated processes of making a phone call to search telephone directory from, the user only needs that phone is said the name or the place name that need to dial and corresponding number can be transferred to, and it is very convenient that the user is called.The voice input device that the present invention utilizes phone itself just to have passes through central processing unit of a versatile interface card connection, this central processing unit can be according to the name or the geographical name data of plug-in recognizing voice typing, and the mode by button or speech recognition is input to corresponding number in the flash memory of device and goes, thereby produces telephone book entries.The method of the invention and device thereof all adopt SI (Speech Independent) speech recognition technology, based on the irrelevant pattern of speaker's voice, can realize people's typing, many people use.The common noise of most landline telephone working environments is little than mobile phone, speech recognition technology can obtain very high accuracy of identification, a plurality of users of the landline telephone of family or mass organizations use have identical conversation object usually in addition, therefore, the present invention is specially adapted to the voice telephone number book application in the landline telephone.
Description of drawings
Fig. 1 is a speech recognition phone numbering directory apparatus structure synoptic diagram;
Fig. 2 is a speech recognition phone numbering directory device voice typing subsystem process flow diagram;
Fig. 3 is a speech recognition phone numbering directory device phonetic dialing subsystem process flow diagram;
Fig. 4 is that speech recognition phone numbering directory device is revised the subsystem process flow diagram.
Embodiment
Below in conjunction with accompanying drawing the present invention is specifically described: a kind of method of voice typing phone number entry, carry out the result of phonetic entry receiving user speech input back, described method comprises the following steps: that name that (a) prompting user will be to be imported or ground noun phrase are decomposed into isolated word and read in a voice telephone number book device; (b) the prompting user is this word filling tone after reading in the voice of described isolated word; (c) a plurality of speech recognition candidates of providing according to certain rule of monitoring speech recognition device and feed back to the user, the prompting user selects one as a result of in above-mentioned a plurality of speech recognition candidates; (d) after complete typing of phone number entry, control described voice telephone number book device and will form the name/place name of these clauses and subclauses and feed back to the user once more by phonetic synthesis output, the user confirms errorless back input telephone number.In the step (b), the user is by the telephone key input tone of several definition.Feedback described in the step (c) comprises that the literal of phone set display shows or the speech play of voice operation demonstrator.Certain rule refers to the recognition confidence score descending sort from high to low that a plurality of speech recognition candidates provide according to speech recognition device described in the step (c).The quantity of the speech recognition candidates that the user can be self-defined be provided by speech recognition device in the step (c).The user can import telephone number by digital keys or two kinds of modes of voice typing in the step (c).The method of described voice typing phone number entry comprises that further the user revises and deletion the result items purpose.The method of described voice typing phone number entry further comprise the telephone set caller ID for certain bar as a result during the telephone number of project correspondence voice operation demonstrator with this project as a result of speech play.
A kind of speech recognition phone numbering directory device that is used to realize described method, it is characterized in that being provided with in the device central processing unit 6 that speech data is handled, described central processing unit 6 connects voice typing/output unit 1 by versatile interface card 5, connect telephone number key dish 2 and telephone set 3 by interface card 4, the intermediate operations process of described central processing unit 6 is carried out in internal storage 7, voice recognition processing program and the address-book data that need call are kept in the flash memory 8, as shown in Figure 1.
Described voice telephone number book device can be divided into three subsystems, difference called after typing subsystem, the modification subsystem of dial-out subsystem and voice numbering directory.Wherein the typing subsystem has " speech recognition+key mode " and " full voice identification " two kinds of working methods.This dual mode all will be by the name/place name in the speech recognition technology typing address book.Because name/place name of being imported may be the combination of any literal, for avoiding reading in continuously between the word that name/place name causes that voice change and recognition system is done voice and cut apart the discrimination that causes and descend, during typing name/place name clauses and subclauses are resolved into isolated word, the method for phonetic entry then improves discrimination; In the typing subsystem, the typing of name/place name is disposable work, uses repeatedly after the correct typing, need not change for a long time, does not therefore influence the ease for use of system.For dwindling the hunting zone of speech recognition system,, improve the typing discrimination by the method that institute's typing speech plus is transferred; Add tone and refer to that the mode by 4 kinds of different keys of key mode input realizes after with each word of phonetic entry.In the typing subsystem, speech recognition device provides a plurality of speech recognition candidates and is not only a most probable speech recognition candidates, the order of a plurality of speech recognition candidates is to arrange to the minimum word descending mode of identification score according to the highest word of recognition confidence score, in a plurality of speech recognition candidate that refer to here a plurality of no more than 6 usually.Beginning in the typing subsystem is word for word typing, and is in time selected by the user by the pairing possible a plurality of words of the current typing of voice feedback; After whole word typings of name or place name are finished, read this phrase by the phonetic synthesis feedback again.In transfering to subsystem, when the user says certain name/place name, according to recognition result, system might provide several candidate entries (this may appearing at when storing the clauses and subclauses of the similar pronunciation of a plurality of pronunciations in the voice telephone number book device), the speech that the clauses and subclauses of this moment are made up of a plurality of words.All speech recognition candidate are broadcasted according to the recognition confidence score from high to low successively by after the phonetic synthesis.Have display screen as telephone set and also can provide Word message, finally from a plurality of speech recognition candidate item, select the project as a result of speech recognition by button or phonetic order by the user by display screen.Thisly the accuracy of voice typings and the fault-tolerance of total system have been improved by the method for user's artificial screening at last by system ambiguous identification.
So-called " speech recognition+key mode " emphasized by the name/information of place names in the voice typing phone number entry, by the numeral of telephone number key dish button typing telephone number correspondence.Its workflow is: the user imports the voice of the isolated word of name/place name one by one by voice typing/output unit 1, and import the tone of this word by the telephone number key dish, the voice data of 6 pairs of inputs of central processing unit carries out feature extraction, carry out speech recognition by plug-in, provide recognition sequence, select to determine final recognition result by button by the user.After whole name/the place name typing is finished, the user presses the acknowledgement key of telephone set regulation and imports the telephone number of this voice correspondence, central processing unit 6 deposits in above-mentioned voice and telephone number in the flash memory 8 of address book correspondence as a phone number entry, play the voice of whole piece phone number entry then and judge once more for the user whether the telephone number of this voice correspondence is correct, as correctly then finishing the phonetic entry of this telephone number, as the incorrect voice directory modification subroutine that then enters, as shown in Figure 2.
Corresponding user's operating process is as follows:
1. press " typing of the clauses and subclauses voice " function key in the voice telephone number book device,, read in the name that needs typing, the isolated word of place name one by one.
2. unclamp " typing of clauses and subclauses voice " function key, key in the tone of typing word by numeric keypad.After the key entry tone was finished, speech recognition engine began to discern name, place name.
After 3 identifications were finished, recognition system provided the recognition result sequence, broadcasted (broadcasting most probable recognition result earlier) successively by voice feedback, selected final recognition result by the user.The method of selecting can be after certain word broadcasts immediately by predefined acknowledgement key (as certain numerical key ' 1 ', back function key ' # ' etc.), also can behind whole candidate sequence close of transmissions, select sequence number to obtain.
After the input of 4 each word was finished, whether system prompt continued voice typing word, and the user by voice answering ' is ', then continued typing, and the typing of this name/place name is finished in answer ' denying ' or ' not being ' then system.
5. by pressing the original numerical key input of phone name, the pairing telephone number of place name.
6. after the end of input, voice telephone number book device can pass through function key, or predefined time delay judges whether end of input voice telephone number book device, and voice feedback confirms that subsystem asks the user to confirm whether whole isolated words of the name/place name of importing are discerned correctly once more.And with second the step same procedure obtain user feedback.Confirm that as the user whole telephone item is correct, this clauses and subclauses typing success, otherwise, prompting user typing again, or abandon these clauses and subclauses of typing.Again typing and abandon selecting all can be by obtaining by numerical key or audio recognition method.
So-called " full voice recognition method " be meant by the name/information of place names in the voice typing phone number entry, and the numeral of the mode typing telephone number correspondence of reading in by voice once more.Its workflow is identical with the workflow of above-mentioned " speech recognition+key mode " substantially, but the telephone number typing in the 3rd step is also finished by audio recognition method.Two kinds of methods are compared, and " speech recognition+key mode " accuracy is higher, and it is more reliable to work, and this mode does not need to hold susceptible to user acceptance to telephone set increase extras.
Transfer to the workflow of subsystem to be: the user is by voice typing/output unit 1 sound import, the voice data of 6 pairs of inputs of central processing unit carries out feature extraction, carry out speech recognition according to plug-in, and provide the highest preceding several candidates of degree of confidence according to the recognition confidence score and discern clauses and subclauses.Recognition system is judged the clauses and subclauses whether similar pronunciation is arranged in the address book according to the recognition confidence of the highest candidate item of several degree of confidence before providing.If the degree of confidence difference of preceding several candidate item not quite then show the clauses and subclauses that similar pronunciation is arranged in the address book; Apparently higher than other candidates, then there are not similar pronunciation clauses and subclauses as first candidate's degree of confidence in the address book.Exist if any similar pronunciation clauses and subclauses, then system discerns clauses and subclauses by speech synthesis technique with the candidate of given number and feeds back to user's (phonetic synthesis is by described voice typing/output unit 1 realization), selects net result by the user.The first candidate's recognition result score that obtains as recognition system is apparently higher than other candidate result, then system directly with first candidate as final recognition result.When the user can not get required name/place name in candidate sequence, require heavily to import or abandon phonetic dialing to system feedback.After recognition system provides the required clauses and subclauses of user, the user presses the affirmation of telephone set regulation and is good for, central processing unit 6 will be transferred address book address entry data from flash memory 8, to transfer to by telephone set telephone number that should voice by interface card 4 and corresponding digital-to-analog conversion, as shown in Figure 3.Corresponding user's operating process is as follows:
1. the user presses voice telephone number book device " phonetic dialing " function key, the call voice input pickup is said name, the place name that need dial.
2. press voice telephone directory book device " phonetic dialing " function key again, speech recognition system begins identification, obtains the result.
3. confirm subsystem feedback recognition result by voice feedback, confirm final recognition result by the user.The available confirmation method identical with the typing subsystem confirmed recognition result by the user.
4. the user confirms correctly, and the dial-out system transfers to telephone number corresponding with this name, place name in the numbering directory.Otherwise the prompting user reads name, place name again, or abandons dialing.Again typing and abandon selecting all can be by obtaining by numerical key or audio recognition method, as shown in Figure 3.The dial-out system of this patent does not influence traditional dialing phone.
When having the clauses and subclauses of a plurality of similar pronunciations in the numbering directory, transfer to subsystem can adopt speech recognition device to provide by user-defined speech recognition candidates quantity, as 2 or 3 some name/address entrys, the user selects one therein.
Wherein the voice telephone number book is revised subsystem and can be adopted two kinds of working methods, as shown in Figure 4:
1. browsing mode begins to read entries of phone book from a certain position, when running into the clauses and subclauses that need modification, deletes by the function key of phone, revises work such as telephone number.
2. inquiry mode is identical with pullout method, name in the phonetic entry numbering directory or place name, and the user can be confirmed to be Outgoing Number and still revise these clauses and subclauses.
When the telephone set caller ID be certain bar as a result during the telephone number of project correspondence voice operation demonstrator will play this result items purpose synthetic speech, thereby realize that phone ringing is the tinkle of bells that has name, this is the caller identification of another mode, has increased the added value of phone.
Claims (10)
1. the method for a voice typing phone number entry receiving the result who carries out phonetic entry after user speech is imported, is characterized in that described method comprises the following steps:
(a) name that the prompting user will be to be imported or ground noun phrase are decomposed into isolated word and read in a voice telephone number book device;
(b) the prompting user is this word filling tone after reading in the voice of described isolated word;
(c) a plurality of speech recognition candidates of providing according to certain rule of monitoring speech recognition device and feed back to the user, the prompting user selects one as a result of in above-mentioned a plurality of speech recognition candidates;
(d) after complete typing of phone number entry, control described voice telephone number book device and will form the name/place name of these clauses and subclauses and feed back to the user once more by phonetic synthesis output, the user confirms errorless back input telephone number.
2. the method for a kind of voice typing phone number entry according to claim 1 is characterized in that in the step (b), and the user is by the telephone key input tone of several definition.
3. the method for a kind of voice typing phone number entry according to claim 1 is characterized in that feedback described in the step (c) comprises that the literal of phone set display shows or the speech play of voice operation demonstrator.
4. the method for a kind of voice typing phone number entry according to claim 1 is characterized in that certain rule refers to the recognition confidence score descending sort from high to low that a plurality of speech recognition candidates provide according to speech recognition device described in the step (c).
5. the method for a kind of voice typing phone number entry according to claim 1 is characterized in that the quantity of the speech recognition candidates that user in the step (c) can be self-defined be provided by speech recognition device.
6. the method for a kind of voice typing phone number entry according to claim 1 is characterized in that the mode of input telephone number in the step (c) is the digital keys typing.
7. the method for a kind of voice typing phone number entry according to claim 1 is characterized in that the mode of input telephone number in the step (c) is the voice typing.
8. the method for a kind of voice typing phone number entry according to claim 1 is characterized in that described method comprises that further the user revises and deletion the result items purpose.
9. the method for a kind of voice typing phone number entry according to claim 1, it is characterized in that described method further comprise the telephone set caller ID for certain bar as a result during the telephone number of project correspondence voice operation demonstrator with this project as a result of speech play.
10. voice telephone number book device, it is characterized in that being provided with in the device central processing unit (6) that speech data is handled, described central processing unit (6) connects voice typing/output units (1) by versatile interface card (5), connect telephone number key dish (2) and telephone set (3) by interface card (4), the intermediate operations process of described central processing unit (6) is carried out in internal storage (7), and the voice recognition processing program that need call and address-book data are kept in the flash memory (8).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2003101082388A CN1612209A (en) | 2003-10-29 | 2003-10-29 | Method and device for voice recording telephone number entry |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2003101082388A CN1612209A (en) | 2003-10-29 | 2003-10-29 | Method and device for voice recording telephone number entry |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1612209A true CN1612209A (en) | 2005-05-04 |
Family
ID=34758527
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2003101082388A Pending CN1612209A (en) | 2003-10-29 | 2003-10-29 | Method and device for voice recording telephone number entry |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1612209A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102467362A (en) * | 2010-11-18 | 2012-05-23 | 联想(北京)有限公司 | Voice input method and input device |
| CN104580723A (en) * | 2014-12-31 | 2015-04-29 | 深圳市金立通信设备有限公司 | Terminal |
| CN106710585A (en) * | 2016-12-22 | 2017-05-24 | 上海语知义信息技术有限公司 | Method and system for broadcasting polyphonic characters in voice interaction process |
-
2003
- 2003-10-29 CN CNA2003101082388A patent/CN1612209A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102467362A (en) * | 2010-11-18 | 2012-05-23 | 联想(北京)有限公司 | Voice input method and input device |
| CN102467362B (en) * | 2010-11-18 | 2015-04-29 | 联想(北京)有限公司 | Voice input method and input device |
| CN104580723A (en) * | 2014-12-31 | 2015-04-29 | 深圳市金立通信设备有限公司 | Terminal |
| CN106710585A (en) * | 2016-12-22 | 2017-05-24 | 上海语知义信息技术有限公司 | Method and system for broadcasting polyphonic characters in voice interaction process |
| CN106710585B (en) * | 2016-12-22 | 2019-11-08 | 云知声（上海）智能科技有限公司 | Polyphone broadcasting method and system during interactive voice |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1184841C (en) | Mobile communication terminal and its voice distinguishing method and record medium | |
| CN1249667C (en) | Voice-operated services | |
| US6163596A (en) | Phonebook | |
| CN1705326A (en) | Method for performing functions using telephone number searched for in mobile terminal | |
| CN101616212B (en) | Mobile phone and method for automatically searching same | |
| CN1925515A (en) | Method and system for enquiring telephone number | |
| JP2002540731A (en) | System and method for generating a sequence of numbers for use by a mobile phone | |
| US7269563B2 (en) | String matching of locally stored information for voice dialing on a cellular telephone | |
| CN102025834A (en) | Mobile terminal voice operation method and device | |
| CN1165889C (en) | Method and system for voice dialling | |
| CN1863236A (en) | Mobile phone capable of transmitting preset speech information and relative method thereof | |
| CN101212505A (en) | Quick phone book searching method and device on communication terminal | |
| CN1612209A (en) | Method and device for voice recording telephone number entry | |
| CN1893482A (en) | System and method for realizing voice dialling by intelligent network | |
| JPS6059846A (en) | Voice recognition automatic dial device | |
| CN1598754A (en) | Mobile terminal and method for providing a user-interface using a voice signal | |
| CN1231038C (en) | Fast speech dialing apparatus and method | |
| CN1564569A (en) | Serching method of telephone number and its serching engine | |
| EP1076979A1 (en) | Personalized information system | |
| CN1492654A (en) | Method for quickly inputting telephone numbers | |
| CN1649368A (en) | Voice identification telephone set | |
| CN1182694C (en) | Voice command system for automatic dialing | |
| CN1278207C (en) | Chinese-character 'speech recognition' input method for mobile telephone | |
| CN1663309A (en) | Telephone apparatus | |
| CN2660813Y (en) | Two-staged speech recognition telephone number calling device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |