CN107689225A - A kind of method for automatically generating minutes - Google Patents

A kind of method for automatically generating minutes Download PDF

Info

Publication number
CN107689225A
CN107689225A CN201710907548.8A CN201710907548A CN107689225A CN 107689225 A CN107689225 A CN 107689225A CN 201710907548 A CN201710907548 A CN 201710907548A CN 107689225 A CN107689225 A CN 107689225A
Authority
CN
China
Prior art keywords
speaker
vocal print
print feature
sound bite
label
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.)
Granted
Application number
CN201710907548.8A
Other languages
Chinese (zh)
Other versions
CN107689225B (en
Inventor
刘维
王贤俊
高刚强
郑文侃
邱霖恺
何逞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
FUJIAN SHIDA COMPUTER EQUIPMENT Co Ltd
Original Assignee
FUJIAN SHIDA COMPUTER EQUIPMENT Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by FUJIAN SHIDA COMPUTER EQUIPMENT Co Ltd filed Critical FUJIAN SHIDA COMPUTER EQUIPMENT Co Ltd
Priority to CN201710907548.8A priority Critical patent/CN107689225B/en
Publication of CN107689225A publication Critical patent/CN107689225A/en
Application granted granted Critical
Publication of CN107689225B publication Critical patent/CN107689225B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/26—Speech to text systems
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L17/00—Speaker identification or verification techniques
    • G10L17/02—Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L17/00—Speaker identification or verification techniques
    • G10L17/22—Interactive procedures; Man-machine interfaces

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Computational Linguistics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Telephonic Communication Services (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present invention provides a kind of method for automatically generating minutes, including step 1）The voice of recorded meeting process；2）Some sound bites for content of speaking only are included by the extraction of VAD technologies；3）Sound bite is converted to by word by speech recognition technology；4）Speaker's vocal print feature is extracted from sound bite, and speaker clustering is carried out to sound bite according to vocal print feature, while speaker's label is set；5）Sound bite, corresponding word and speaker's label are stored；6）It is specific speaker by speaker's tag replacement, obtains final minutes.Phonetic feature, voiceprint of the present invention without collecting personnel participating in the meeting in advance, carry out speaker clustering automatically.

Description

A kind of method for automatically generating minutes
Technical field
The present invention relates to technical field of information processing, and in particular to a kind of method for automatically generating minutes.
Background technology
In general minutes are to pass through the equipment such as recording pen, video camera by special scribe to shoot with video-corder in meeting Hold, then carry out manual sorting and obtain, with the development of science and technology people begin one's study how automatically to be quickly obtained meeting note Record, including spokesman, speech content etc..
Patent application CN201510530579 discloses a kind of minutes method, and software is converted by voice signal by voice Corresponding text information is changed into, then the text information of mistake is recognized, edlin of going forward side by side is shown；Patent application CN201410839533 discloses a kind of minutes device and its method for automatically generating minutes, is received according in meeting Voice signal and memory in the user vocal feature table that stores, recognize voice signal corresponding to user, then voice is believed Number being converted to word obtains minutes.
Above-mentioned prior art, it is special in the voice for need to collect in advance during spokesman's identification and store the related personnel that attends a meeting Reference ceases, and is then contrasted by the voice messaging and the user vocal feature information of storage of recording so as to pick out use of speaking Family.But the personnel participating in the meeting actually having a meeting is not often fixed even to have a meeting together with cooperative venture personnel, and this often leads to can not The voice characteristics information of personnel participating in the meeting, while the temporary variations of personnel participating in the meeting are collected in advance（Increase personnel）Existing skill can not be used The workflow of art, and the voice characteristics information of personnel participating in the meeting is stored to the potential safety hazard that itself leakage be present（It is important If the voice characteristics information leakage of personage may be used by crime one's share of expenses for a joint undertaking）.
The content of the invention
In view of this, the present invention provides a kind of method for automatically generating minutes, without collecting storage participant in advance The phonetic feature of member can be classified speaker；Minutes arrangement personnel can will readily appreciate that meeting after carrying out speaker clustering The speech content of speaker and arranged in view.
To achieve the above object, the technical scheme is that：A kind of method for automatically generating minutes, including it is following Step：
Step S1：The voice of recorded meeting process；
Step S2：Some sound bites for content of speaking only are included by the extraction of VAD technologies；
Step S3：Cross speech recognition technology and sound bite is converted into word；
Step S4：Speaker's vocal print feature is extracted from sound bite, and sound bite is spoken according to vocal print feature People classifies, while sets speaker's label；
Step S5：Sound bite, corresponding word and speaker's label are stored；
Step S6：It is specific speaker by speaker's tag replacement, obtains final minutes.
Further, the step S4 is specific as follows：
According to the order of sound bite, the vocal print feature in sound bite is extracted by i-vector and GMM-UBM voice technologies；
Extract first vocal print feature is subjected to interim storage, the text that sound bite corresponding to the vocal print feature is changed It is meter digital that word, which sets speaker's label speakerX, X, X=1；
Vocal print feature is extracted successively, and is matched one by one with the vocal print feature of interim storage, and such as the match is successful, then by extraction The word that sound bite corresponding to vocal print feature is changed sets identical speaker's label speakerX；Such as match it is unsuccessful, Vocal print feature speaker's label counting position of extraction is then added 1, and carries out interim storage.
Further, in addition to step S7：Delete the vocal print feature data of interim storage.
Compared with prior art, the present invention has beneficial effect：Without collecting phonetic feature, the vocal print of personnel participating in the meeting in advance Information, so the temporary variations of personnel participating in the meeting do not interfere with the workflow of the present invention；Although it is extracted in processing procedure The vocal print feature information of people is talked about, but is soon deleted after processing terminates, it is ensured that the safety of speaker's vocal print feature information Property.
Brief description of the drawings
Fig. 1 is a kind of method flow diagram for automatically generating minutes of the present invention；
Fig. 2 is the processing procedure schematic diagram of one embodiment of the invention.
Embodiment
Below in conjunction with the accompanying drawings and embodiment the present invention will be further described.
As shown in figure 1, a kind of method for automatically generating minutes, comprises the following steps：
Step S1：The voice of recorded meeting process；
Step S2：Some sound bites for content of speaking only are included by the extraction of VAD technologies；
Step S3：Sound bite is converted to by word by speech recognition technology；
Step S4：Speaker's vocal print feature is extracted from sound bite, and sound bite is spoken according to vocal print feature People classifies, while sets speaker's label；
It is specific as follows：
According to the order of sound bite, the vocal print feature in sound bite is extracted by i-vector and GMM-UBM voice technologies；
Extract first vocal print feature is subjected to interim storage, the text that sound bite corresponding to the vocal print feature is changed It is meter digital that word, which sets speaker's label speakerX, X, X=1；
As shown in Fig. 2 in the present embodiment, the word that sound bite corresponding to first vocal print feature of extraction is changed is " welcoming everybody to participate in the meeting of today ", speaker's label speaker1 is set；
Vocal print feature is extracted successively, and is matched with the vocal print feature of interim storage, and such as the match is successful, then by the vocal print of extraction The word that sound bite corresponding to feature is changed sets identical speaker's label speakerX；Such as match it is unsuccessful, then will Vocal print feature speaker's label counting position of extraction adds 1, and carries out interim storage；
In the present embodiment, second vocal print feature of extraction can not match with the speaker1 of interim storage vocal print feature Work(, then speaker's label counting position is set to add 1 the word that sound bite corresponding to second vocal print feature is changed, i.e., speaker2；Then the vocal print feature of the 3rd vocal print feature of extraction and the speaker1 and speaker2 of interim storage is entered Row matching, matching is unsuccessful, then the word changed the 3rd sound bite corresponding to vocal print feature sets speaker's label Meter digital adds 1, i.e. speaker3；In the present embodiment, the speaker1 of the 4th vocal print feature of extraction and interim storage, Speaker2, speaker3 are matched one by one, and the match is successful with speaker1, then by the 4th voice corresponding to vocal print feature The word that fragment is changed sets identical speaker label speaker1, and the rest may be inferred, and all sound bites are handled；
Step S5：Sound bite, corresponding word and speaker's label are stored；
Step S6：It is specific speaker by speaker's tag replacement, obtains final minutes；
Step S7：Delete the vocal print feature data of interim storage.
Phonetic feature, voiceprint of the present invention without collecting personnel participating in the meeting in advance, it is local directly to extract voiceprint simultaneously Contrasted, the automatic classification for carrying out speaker, finally only need tag along sort replacing with actual speaker and can reach to meeting Discuss the arrangement of record.
The foregoing is only presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with Modification, it should all belong to the covering scope of the present invention.

Claims (3)

  1. A kind of 1. method for automatically generating minutes, it is characterised in that comprise the following steps：
    Step S1：The voice of recorded meeting process；
    Step S2：Some sound bites for content of speaking only are included by the extraction of VAD technologies；
    Step S3：Sound bite is converted to by word by speech recognition technology；
    Step S4：Speaker's vocal print feature is extracted from sound bite, and sound bite is spoken according to vocal print feature People classifies, while sets speaker's label；
    Step S5：Sound bite, corresponding word and speaker's label are stored；
    Step S6：It is specific speaker by speaker's tag replacement, obtains final minutes.
  2. 2. a kind of method for automatically generating minutes according to claim 1, it is characterised in that the step S4 is specific It is as follows：
    According to the order of sound bite, the vocal print feature in sound bite is extracted by i-vector and GMM-UBM voice technologies；
    Extract first vocal print feature is subjected to interim storage, the text that sound bite corresponding to the vocal print feature is changed It is meter digital that word, which sets speaker's label speakerX, X, X=1；
    Vocal print feature is extracted successively, and is matched one by one with the vocal print feature of interim storage, and such as the match is successful, then by extraction The word that sound bite corresponding to vocal print feature is changed sets identical speaker's label speakerX；Such as match it is unsuccessful, Vocal print feature speaker's label counting position of extraction is then added 1, and carries out interim storage.
  3. 3. a kind of method for automatically generating minutes according to claim 2, it is characterised in that also including step S7： Delete the vocal print feature data of interim storage.
CN201710907548.8A 2017-09-29 2017-09-29 A method of automatically generating minutes Active CN107689225B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710907548.8A CN107689225B (en) 2017-09-29 2017-09-29 A method of automatically generating minutes

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710907548.8A CN107689225B (en) 2017-09-29 2017-09-29 A method of automatically generating minutes

Publications (2)

Publication Number Publication Date
CN107689225A true CN107689225A (en) 2018-02-13
CN107689225B CN107689225B (en) 2019-11-19

Family

ID=61153780

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710907548.8A Active CN107689225B (en) 2017-09-29 2017-09-29 A method of automatically generating minutes

Country Status (1)

Country Link
CN (1) CN107689225B (en)

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108447502A (en) * 2018-03-09 2018-08-24 福州米鱼信息科技有限公司 A kind of memo method and terminal based on voice messaging
CN108510992A (en) * 2018-03-22 2018-09-07 北京云知声信息技术有限公司 The method of voice wake-up device
CN109767757A (en) * 2019-01-16 2019-05-17 平安科技（深圳）有限公司 A kind of minutes generation method and device
CN110335612A (en) * 2019-07-11 2019-10-15 招商局金融科技有限公司 Minutes generation method, device and storage medium based on speech recognition
CN110428853A (en) * 2019-08-30 2019-11-08 北京太极华保科技股份有限公司 Voice activity detection method, Voice activity detection device and electronic equipment
WO2020073633A1 (en) * 2018-10-12 2020-04-16 深圳海翼智新科技有限公司 Conference loudspeaker box, conference recording method, device and system, and computer storage medium
CN111145903A (en) * 2019-12-18 2020-05-12 东北大学 Method and device for acquiring vertigo inquiry text, electronic equipment and inquiry system
CN111180025A (en) * 2019-12-18 2020-05-19 东北大学 Method and device for representing medical record text vector and inquiry system
WO2020147256A1 (en) * 2019-01-16 2020-07-23 平安科技（深圳）有限公司 Conference content distinguishing method and apparatus, and computer device and storage medium
CN112017632A (en) * 2020-09-02 2020-12-01 浪潮云信息技术股份公司 Automatic conference record generation method
CN113113022A (en) * 2021-04-15 2021-07-13 吉林大学 Method for automatically identifying identity based on voiceprint information of speaker
CN113299279A (en) * 2021-05-18 2021-08-24 上海明略人工智能(集团)有限公司 Method, apparatus, electronic device and readable storage medium for associating voice data and retrieving voice data
CN113536257A (en) * 2021-07-27 2021-10-22 南京邮电大学盐城大数据研究院有限公司 Multi-party conference admission method and system based on block chain
CN115394304A (en) * 2021-03-30 2022-11-25 北京百度网讯科技有限公司 Voiceprint determination method, apparatus, system, device and storage medium

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008032825A (en) * 2006-07-26 2008-02-14 Fujitsu Fsas Inc Speaker display system, speaker display method and speaker display program
CN102985965A (en) * 2010-05-24 2013-03-20 微软公司 Voice print identification
CN103165131A (en) * 2011-12-17 2013-06-19 富泰华工业（深圳）有限公司 Voice processing system and voice processing method
WO2015024413A1 (en) * 2013-08-22 2015-02-26 中兴通讯股份有限公司 Conference summary extraction method and device
CN106657865A (en) * 2016-12-16 2017-05-10 联想(北京)有限公司 Method and device for generating conference summary and video conference system
CN106782545A (en) * 2016-12-16 2017-05-31 广州视源电子科技股份有限公司 A kind of system and method that audio, video data is changed into writing record

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008032825A (en) * 2006-07-26 2008-02-14 Fujitsu Fsas Inc Speaker display system, speaker display method and speaker display program
CN102985965A (en) * 2010-05-24 2013-03-20 微软公司 Voice print identification
CN103165131A (en) * 2011-12-17 2013-06-19 富泰华工业（深圳）有限公司 Voice processing system and voice processing method
WO2015024413A1 (en) * 2013-08-22 2015-02-26 中兴通讯股份有限公司 Conference summary extraction method and device
CN106657865A (en) * 2016-12-16 2017-05-10 联想(北京)有限公司 Method and device for generating conference summary and video conference system
CN106782545A (en) * 2016-12-16 2017-05-31 广州视源电子科技股份有限公司 A kind of system and method that audio, video data is changed into writing record

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108447502A (en) * 2018-03-09 2018-08-24 福州米鱼信息科技有限公司 A kind of memo method and terminal based on voice messaging
CN108447502B (en) * 2018-03-09 2020-09-22 福州米鱼信息科技有限公司 Memorandum method and terminal based on voice information
CN108510992A (en) * 2018-03-22 2018-09-07 北京云知声信息技术有限公司 The method of voice wake-up device
WO2020073633A1 (en) * 2018-10-12 2020-04-16 深圳海翼智新科技有限公司 Conference loudspeaker box, conference recording method, device and system, and computer storage medium
WO2020147256A1 (en) * 2019-01-16 2020-07-23 平安科技（深圳）有限公司 Conference content distinguishing method and apparatus, and computer device and storage medium
CN109767757A (en) * 2019-01-16 2019-05-17 平安科技（深圳）有限公司 A kind of minutes generation method and device
CN110335612A (en) * 2019-07-11 2019-10-15 招商局金融科技有限公司 Minutes generation method, device and storage medium based on speech recognition
CN110428853A (en) * 2019-08-30 2019-11-08 北京太极华保科技股份有限公司 Voice activity detection method, Voice activity detection device and electronic equipment
CN111180025A (en) * 2019-12-18 2020-05-19 东北大学 Method and device for representing medical record text vector and inquiry system
CN111145903A (en) * 2019-12-18 2020-05-12 东北大学 Method and device for acquiring vertigo inquiry text, electronic equipment and inquiry system
CN112017632A (en) * 2020-09-02 2020-12-01 浪潮云信息技术股份公司 Automatic conference record generation method
CN115394304A (en) * 2021-03-30 2022-11-25 北京百度网讯科技有限公司 Voiceprint determination method, apparatus, system, device and storage medium
CN113113022A (en) * 2021-04-15 2021-07-13 吉林大学 Method for automatically identifying identity based on voiceprint information of speaker
CN113299279A (en) * 2021-05-18 2021-08-24 上海明略人工智能(集团)有限公司 Method, apparatus, electronic device and readable storage medium for associating voice data and retrieving voice data
CN113536257A (en) * 2021-07-27 2021-10-22 南京邮电大学盐城大数据研究院有限公司 Multi-party conference admission method and system based on block chain

Also Published As

Publication number Publication date
CN107689225B (en) 2019-11-19

Similar Documents

Publication Publication Date Title
CN107689225B (en) A method of automatically generating minutes
CN106657865B (en) Conference summary generation method and device and video conference system
CN106782545B (en) A kind of system and method that audio, video data is converted to writing record
CN107274916B (en) Method and device for operating audio/video file based on voiceprint information
WO2020211354A1 (en) Speaker identity recognition method and device based on speech content, and storage medium
US8676586B2 (en) Method and apparatus for interaction or discourse analytics
JP4466564B2 (en) Document creation / viewing device, document creation / viewing robot, and document creation / viewing program
US8219404B2 (en) Method and apparatus for recognizing a speaker in lawful interception systems
CN112037791B (en) Conference summary transcription method, apparatus and storage medium
CN105975569A (en) Voice processing method and terminal
CN107154257A (en) Customer service quality evaluating method and system based on customer voice emotion
US20130158992A1 (en) Speech processing system and method
TWI590240B (en) Meeting minutes device and method thereof for automatically creating meeting minutes
CN105957514A (en) Portable deaf-mute communication equipment
JP2010060850A (en) Minute preparation support device, minute preparation support method, program for supporting minute preparation and minute preparation support system
CA2652970A1 (en) System and method for sorting objects using ocr and speech recognition techniques
CN106982344A (en) video information processing method and device
CN111223487B (en) Information processing method and electronic equipment
CN206672635U (en) A kind of voice interaction device based on book service robot
JP2017167726A (en) Conversation analyzer, method and computer program
CN113744742A (en) Role identification method, device and system in conversation scene
CN113327619B (en) Conference recording method and system based on cloud-edge collaborative architecture
CN111010484A (en) Automatic quality inspection method for call recording
CN112468753B (en) Method and device for acquiring and checking record data based on audio and video recognition technology
CN112562644A (en) Customer service quality inspection method, system, equipment and medium based on human voice separation

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant