CN106656767A - Method and system for increasing new anchor retention - Google Patents

Method and system for increasing new anchor retention Download PDF

Info

Publication number
CN106656767A
CN106656767A CN201710014790.2A CN201710014790A CN106656767A CN 106656767 A CN106656767 A CN 106656767A CN 201710014790 A CN201710014790 A CN 201710014790A CN 106656767 A CN106656767 A CN 106656767A
Authority
CN
China
Prior art keywords
main broadcaster
entropy
increase
direct broadcasting
broadcasting room
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
Application number
CN201710014790.2A
Other languages
Chinese (zh)
Inventor
赵连超
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.)
Wuhan Douyu Network Technology Co Ltd
Original Assignee
Wuhan Douyu Network Technology 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 Wuhan Douyu Network Technology Co Ltd filed Critical Wuhan Douyu Network Technology Co Ltd
Priority to CN201710014790.2A priority Critical patent/CN106656767A/en
Publication of CN106656767A publication Critical patent/CN106656767A/en
Pending legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L51/00—User-to-user messaging in packet-switching networks, transmitted according to store-and-forward or real-time protocols, e.g. e-mail
    • H04L51/02—User-to-user messaging in packet-switching networks, transmitted according to store-and-forward or real-time protocols, e.g. e-mail using automatic reactions or user delegation, e.g. automatic replies or chatbot-generated messages
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/04—Segmentation; Word boundary detection
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L15/00—Speech recognition
    • G10L15/26—Speech to text systems
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/47—End-user applications
    • H04N21/478—Supplemental services, e.g. displaying phone caller identification, shopping application
    • H04N21/4788—Supplemental services, e.g. displaying phone caller identification, shopping application communicating with other users, e.g. chatting

Abstract

The present invention discloses a method and system for increasing new anchor retention. The method includes: 1) monitoring barrage amount of a live room in which an anchor is located in real time, and dynamically increasing or decreasing the number of simulation robots in the live room; 2) collecting a voice data stream of the anchor, and intelligently punctuating the voice data stream, to segment the voice data stream into small voice segments by punctuation; 3) invoking a voice recognition API to recognize the small voice segments in the previous step, and converting the intelligently punctuated small voice segments into text information M; 4) taking the text information M as an input and sending same to an interface of an intelligent chat robot, so that the intelligent chat robot gives response text information R; 5) according to the text information R obtained in step 4), generating a barrage, and pushing same to the anchor; and 6) judging whether the anchor makes a voice response, and performing barrage interaction with the anchor. The method and system can effectively increase the live passion of a new anchor, and enrich the diversity of barrage contents of the live room.

Description

It is a kind of to increase the method and system that new main broadcaster retains
Technical field
The present invention relates to the Internet direct seeding technique, more particularly to a kind of method and system for increasing new main broadcaster's retention.
Background technology
With the fast development of live industry, butcher,baker,and the candlestick maker initially enters live industry, but the general popularity of new main broadcaster All very low, nobody's barrage together with main broadcaster is interactive, and main broadcaster is easy to lose live interest, so new main broadcaster is easy to stream Lose.In order to increase the live enthusiasm of new main broadcaster, new main broadcaster is allowed smoothly to spend the popularity relatively low initial live phase, this programme description It is a kind of to increase the solution that new main broadcaster retains.The program by increase in new main broadcaster's direct broadcasting room intelligent chat robots and The method of speech recognition, carries out barrage interactive with main broadcaster.
The content of the invention
The technical problem to be solved in the present invention is for defect of the prior art, there is provided one kind increases new main broadcaster and retains Method and system.
The technical solution adopted for the present invention to solve the technical problems is：It is a kind of to increase the method that new main broadcaster retains, including：
1) the barrage amount of real-time monitoring main broadcaster place direct broadcasting room, intelligent chat robots quantity in dynamic increase and decrease direct broadcasting room；
2) audio data stream of main broadcaster is gathered, and intelligent punctuate is carried out to audio data stream, by making pauses in reading unpunctuated ancient writings speech data Flow point is cut into voice segment；
3) speech recognition API is called to be identified the voice segment in previous step, by the voice segment after intelligence punctuate Be converted to text message M.
4) text message M is sent as input to into intelligent chat robots interface, what intelligent chat robots were provided returns Multiple text R；
5) according to step 4) the text message R that obtains, barrage is generated, it is pushed to main broadcaster；
6) judge whether main broadcaster has speech answering, if it has, then repeat step 1) to 5), directly carry out main broadcaster and intelligence chat Barrage between its robot is interactive；If it is determined that main broadcaster does not have speech answering, then the barrage of initial setting up is issued.
By such scheme, the step 2) in that intelligence punctuate is carried out to audio data stream is specific as follows：Using based on entropy The Method of Speech Endpoint Detection, constructs a voice messaging entropy function, to the voice for extracting main broadcaster from one section of voice flow Information, cancelling noise information；When standardization amplitude spectrum | the X (i) | speech frame regards a probability distribution as, the entropy meter of spectral domain Calculation can take the probability of i-th amplitude spectrum to replace taking the probability P (S of i-th source symboli), i.e., represented with below equation：
So, the entropy of speech manual can be expressed as follows：
Calculate and intelligent punctuate is carried out to audio data stream according to entropy size after entropy：For background noise, The absolute value of the voice segments amplitude of variation in voice signal is more than given threshold, and entropy is big；And quiet section of amplitude of variation is exhausted Given threshold is less than to value, Relatively centralized is distributed, entropy is little.
By such scheme, the step 5) in generate the method for barrage and be：Dummy robot's account is created first, and At random from pre-set ID and pet name storehouse, the pet name as the dummy robot is selected；Then dummy robot Simulate the direct broadcasting room that normal spectators log into main broadcaster；The barrage transmission interface that live application is carried finally is called, by step 4) The text message R for obtaining is sent in chatroom by interface.
By such scheme, the dummy robot has when logining direct broadcasting room and following logins effect：For dummy robot's increase Random grade and website title, when dummy robot enters direct broadcasting room, increase the direct broadcasting room that enters to dummy robot and welcome Sentence, increases the welcome specially good effect to special title.
By such scheme, the step 1) in following methods are adopted to the dynamic increase and decrease of direct broadcasting room robot quantity：Work as master Direct broadcasting room popularity value is broadcast less than popularity threshold value T1When, increase the intelligent chat robots of setting quantity, when direct broadcasting room popularity value reaches To or more than popularity threshold value T1When, reduce the intelligent chat robots of setting quantity；Direct broadcasting room popularity value P, by rear number of units Safeguard according to storehouse, when popularity value P<T1,T1When=100, direct broadcasting room people's QI-insufficiency is represented, need to introduce a number of intelligence chat Robot；As popularity value P >=T1,T1When=100, represent that direct broadcasting room popularity value rises, it is desirable to reduce a number of intelligence is chatted Its robot；
The quantity for introducing robot is determined by following rule：
Wherein, P be direct broadcasting room popularity value, the acquisition modes of popularity value：Each direct broadcasting room in one data base of background maintenance, The quantity of the true spectators between current live is added up in real time, is obtained the popularity value between current live.
By such scheme, the step 6) in determine whether main broadcaster's speech answering, the method for using is based on the language of entropy Message end-point detecting method, judgement currently has the entropy in speech manual to be more than certain threshold value T2, you can judge current anchor There are speech answering, wherein T2Represent the threshold value of speech manual entropy.
By such scheme, threshold value T of the speech manual entropy2For 50.
The present invention also provides a kind of system for increasing new main broadcaster's retention, including：
Dummy robot's quantity determining module, by the barrage amount of real-time monitoring main broadcaster place direct broadcasting room, dynamic increase and decrease is straight Middle intelligent chat robots quantity between broadcasting；
Intelligent punctuate module, by the Mike of main broadcaster the audio data stream of main broadcaster is gathered, and to audio data stream intelligence is carried out Punctuate, voice segment is divided into according to punctuate result by audio data stream；
Sound identification module, for calling speech recognition API to be identified the voice segment split, after intelligence punctuate Voice segment be converted to text message M；
Text generation module is replied, for text message M to be sent as input to into intelligent chat robots interface, intelligence The reply text R that chat robots are given；
Barrage module, for according to the text message R that text generation module is obtained is replied, generating barrage, is pushed to main broadcaster, Interaction is carried out with main broadcaster；
Main broadcaster replys judge module, for judging whether main broadcaster has speech answering, if it has, then proceeding main broadcaster and being Barrage between system is interactive；If it is determined that main broadcaster does not have speech answering, then the barrage of initial setting up is issued, improve main broadcaster and reply bullet The enthusiasm of curtain.
By such scheme, the dynamic increase and decrease of direct broadcasting room robot quantity is adopted in dummy robot's quantity determining module Using the following method：When main broadcaster's direct broadcasting room popularity value is less than popularity threshold value T1When, increase the intelligent chat robots of setting quantity, when Direct broadcasting room popularity value meets or exceeds popularity threshold value T1When, reduce the intelligent chat robots of setting quantity；
The quantity for introducing robot is determined by following rule：
Wherein, P is direct broadcasting room popularity value, and the acquisition modes of popularity value are as follows：Each direct broadcasting room is in the number of background maintenance one According to storehouse, the quantity of the true spectators between current live is added up in real time, obtained the popularity value between current live.
By such scheme, intelligent punctuate method is carried out to audio data stream in the intelligent punctuate module as follows：Using base In the Method of Speech Endpoint Detection of entropy, a voice messaging entropy function is constructed, to extract main broadcaster from one section of voice flow Voice messaging, cancelling noise information；Standardization amplitude spectrum | the X (i) | of speech frame is regarded as a probability distribution, the entropy of spectral domain Calculating can take the probability of i-th amplitude spectrum to replace taking the probability P (S of i-th source symboli), i.e., represented with below equation：
So, the entropy of speech manual can be expressed as follows：
Calculate and intelligent punctuate is carried out to audio data stream according to entropy size after entropy：For background noise, The absolute value of the voice segments amplitude of variation in voice signal is more than given threshold, and entropy is big；And quiet section of amplitude of variation is exhausted Given threshold is less than to value, Relatively centralized is distributed, entropy is little.
The beneficial effect comprise that：The present invention wants issue to be resolved to be that main broadcaster can run into popularity at the live initial stage Inadequate problem, when starting live, not enough, without lasting audience interaction, the live enthusiasm of main broadcaster has popularity Very big damage, so the present invention proposes, using with reference to the interactive mode of speech recognition, intelligent chat robots, barrage, to increase Main broadcaster increases interactive enjoyment when people's QI-insufficiency, lifts the retention ratio of new main broadcaster so as to actively carry out live.
Description of the drawings
Below in conjunction with drawings and Examples, the invention will be further described, in accompanying drawing：
Fig. 1 is the method flow diagram of the embodiment of the present invention.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, with reference to embodiments, to the present invention It is further elaborated.It should be appreciated that specific embodiment described herein is not used to limit only to explain the present invention The fixed present invention.
As shown in figure 1, a kind of increase the method that new main broadcaster retains, including：
1) the barrage amount of real-time monitoring main broadcaster place direct broadcasting room, in dynamic increase and decrease direct broadcasting room chat robots quantity is simulated；
When main broadcaster's direct broadcasting room popularity is less, increase a number of dummy robot.When direct broadcasting room popularity reaches necessarily Value T1When (popularity threshold value), a number of dummy robot is reduced.Direct broadcasting room popularity value P, is safeguarded by background data base, works as people Gas value P<T1,T1When=100, direct broadcasting room people's QI-insufficiency is represented, need to introduce a number of dummy robot；When popularity value P >= T1,T1When=100, represent that direct broadcasting room popularity value rises, it is desirable to reduce a number of dummy robot；
The quantity for introducing dummy robot is determined by following rule：
The acquisition modes of popularity value：Each direct broadcasting room in one data base of background maintenance, wherein comprising currently in direct broadcasting room Whether the id information of spectators, spectators are the data such as robot, and backstage in real time enters the quantity of the true spectators between current live Row is cumulative, obtains the popularity value between current live.
2) audio data stream of main broadcaster is gathered by the Mike of main broadcaster, and is stored in voice flow caching, then to voice number Intelligent punctuate is carried out according to stream, audio data stream is divided into by voice segment by punctuate；
It is that to carry out Speech processing (such as speech recognition, speaker identification) important using the end-point detection of voice signal And the first step of key.Used as the embodiment of this example, intelligence punctuate is using the Method of Speech Endpoint Detection based on entropy.Root According to the definition of comentropy, in applying it to voice, a voice messaging entropy function is constructed, to extract from one section of voice flow The voice messaging of main broadcaster, cancelling noise information.When standardization amplitude spectrum | the X (i) | speech frame regards a probability distribution as, The entropy of spectral domain is calculated can be taken the probability of i-th amplitude spectrum to replace taking the probability P (S of i-th source symboli), i.e., with following Formula is represented：
So, the entropy of speech manual can be expressed as follows：
For background noise, the dynamic range of the voice segments amplitude in voice signal than larger, therefore intuitively Say, it is believed that voice segments random event in the signal is more, therefore average information is big, that is, entropy is big.And quiet section Amplitude of variation is little, is distributed Relatively centralized, thus entropy is little.
Thus, it is possible to the voice flow in caching is divided into into voice segment, wherein per section of some voices for all including main broadcaster Reply.
Compared with traditional punctuate mode, voice messaging entropy can preferably distinguish voice segment and noise speech information, With preferable distinction.
3) speech recognition API is called to be identified the voice segment in previous step, by the voice segment after intelligence punctuate Be converted to text message M.
Call speech recognition API to be identified the voice segment in previous step, intelligent punctuate voice messaging is converted to Word message.As the embodiment of this example, speech recognition SDK can be flown using current Baidu's speech recognition SDK or news, be led to Access speech recognition API is crossed, the voice of main broadcaster is identified, convert speech into text message M.
4) text message M is sent as input to into intelligent chat robots interface, what intelligent chat robots were provided returns Multiple text R；As the embodiment of this example, the chat robots of third party manufacturer, such as SimSimiAPI can be directly invoked, be schemed Clever robot API.
5) according to step 4) the text message R that obtains, barrage is generated, main broadcaster is pushed to, carry out interaction with main broadcaster.Generate bullet The method of curtain：First an account is created for dummy robot, and it is random from pre-set ID and pet name storehouse, select one The individual pet name as the robot, secondly simulates the direct broadcasting room that normal spectators log into main broadcaster, finally calls live application certainly The barrage transmission interface of band, the text message R that 4. step is obtained is sent in chatroom by interface, and this process is simulated completely Normal users send the process of barrage, and unique difference is to send barrage with dummy robot.
Dummy robot logins the effect of direct broadcasting room, specifically has the following effects：Increase random grade for dummy robot With website title, when dummy robot enters direct broadcasting room, increase the direct broadcasting room that enters to dummy robot and welcome sentence, it is right to increase Welcome specially good effect of special title etc..
6) judge whether main broadcaster has speech answering, if it has, then repeat step 1) to 5), directly carry out main broadcaster and analog machine Barrage between device people is interactive；If it is determined that main broadcaster does not have speech answering, then the barrage of initial setting up is issued, improve main broadcaster and reply The enthusiasm of barrage.The barrage information of initial setting up has similar：" where you are", " you this year how old" etc. first meeting trick Exhale term.
Main broadcaster's speech answering is determined whether, the method for using is that the previously described speech sound signal terminal point based on entropy is detected Method, judgement currently has the entropy in speech manual to be more than certain threshold value T2, T2=50 (according to platform empirical values), you can judge Current anchor has speech answering.T2Represent the threshold value of speech manual entropy.
The present invention also can determine whether to increase and decrease machine of intelligently chatting in direct broadcasting room by the number of the true spectators of direct broadcasting room People's quantity, it is specific as follows：Judge whether there are true spectators to enter direct broadcasting room between current live, if true spectators enter live Between and number reaches threshold value, then gradually reduce the quantity of intelligent chat robots, the method for reducing the quantity of robot be by Robot is released into this room.Judge whether to have between current live the true spectators be into the mode of direct broadcasting room, rear number of units There is individual labelling to each user according to storehouse, whether labelling active user is the spectators of true registration, if subsequently increase people being so incensed that The robotic user for arriving, then will not generate the labelling, subsequently only need to judge the labelling, you can judge whether it is that real user is entered Enter direct broadcasting room.
The present invention also provides a kind of system for increasing new main broadcaster's retention, including：
Dummy robot's quantity determining module, by the barrage amount of real-time monitoring main broadcaster place direct broadcasting room, dynamic increase and decrease is straight Middle dummy robot's quantity between broadcasting；
Intelligent punctuate module, by the Mike of main broadcaster the audio data stream of main broadcaster is gathered, and to audio data stream intelligence is carried out Punctuate, voice segment is divided into according to punctuate result by audio data stream；
Sound identification module, for calling speech recognition API to be identified the voice segment split, after intelligence punctuate Voice segment be converted to text message M；
Text generation module is replied, for text message M to be sent as input to into intelligent chat robots interface, intelligence The reply text R that chat robots are given；
Barrage module, for according to the text message R that text generation module is obtained is replied, generating barrage, is pushed to main broadcaster, Interaction is carried out with main broadcaster；
Main broadcaster replys judge module, for judging whether main broadcaster has speech answering, if it has, then proceeding main broadcaster and being Barrage between system is interactive；If it is determined that main broadcaster does not have speech answering, then the barrage of initial setting up is issued, improve main broadcaster and reply bullet The enthusiasm of curtain.
It should be appreciated that for those of ordinary skills, can according to the above description be improved or be converted, And all these modifications and variations should all belong to the protection domain of claims of the present invention.

Claims (10)

1. it is a kind of to increase the method that new main broadcaster retains, it is characterised in that to include：
1) the barrage amount of real-time monitoring main broadcaster place direct broadcasting room, dummy robot's quantity in dynamic increase and decrease direct broadcasting room；
2) audio data stream of main broadcaster is gathered, and intelligent punctuate is carried out to audio data stream, by making pauses in reading unpunctuated ancient writings speech data flow point It is cut into voice segment；
3) to step 2) in voice segment be identified, by intelligence punctuate after voice segment be converted to text message M；
4) text message M is sent as input to into intelligent chat robots interface, the reply text that intelligent chat robots are provided This R；
5) according to step 4) the text message R that obtains, barrage is generated, it is pushed to main broadcaster；
6) judge whether main broadcaster has speech answering, if it has, then repeat step 1) to 5), carry out between main broadcaster and dummy robot Barrage it is interactive；If it is determined that main broadcaster does not have speech answering, then the barrage of initial setting up is issued.
2. it is according to claim 1 to increase the method that new main broadcaster retains, it is characterised in that the step 2) in voice number It is as follows intelligent punctuate method to be carried out according to stream：Using the Method of Speech Endpoint Detection based on entropy, a voice messaging entropy is constructed Function, to the voice messaging for extracting main broadcaster from one section of voice flow, cancelling noise information；The standardization amplitude spectrum of speech frame | X (i) | regards a probability distribution as, and the entropy of spectral domain is calculated can be taken the probability of i-th amplitude spectrum to replace taking i-th information source symbol Number probability P (Si), i.e., represented with below equation：
P ( | X ( i ) | 2 ) = | X ( i ) | 2 &Sigma; i | X ( i ) | 2
So, the entropy of speech manual can be expressed as follows：
H = - &Sigma; i P ( | X ( i ) | 2 ) * log 2 P ( | X ( i ) | 2 )
Calculate and intelligent punctuate is carried out to audio data stream according to entropy size after entropy：For background noise, voice The absolute value of the voice segments amplitude of variation in signal is more than given threshold, and entropy is big；And the absolute value of quiet section of amplitude of variation Less than given threshold, Relatively centralized is distributed, entropy is little.
3. it is according to claim 1 to increase the method that new main broadcaster retains, it is characterised in that the step 5) middle generation barrage Method be：First an account is created for dummy robot, and it is random from pre-set ID and pet name storehouse, select one The individual pet name as the dummy robot；Then dummy robot simulates the direct broadcasting room that normal spectators log into main broadcaster；Finally The barrage transmission interface that live application is carried being called, by step 4) the text message R that obtains is sent in chatroom by interface.
4. it is according to claim 3 to increase the method that new main broadcaster retains, it is characterised in that the dummy robot logins directly Have when broadcasting and following login effect：Increase random grade and website title for dummy robot, increase to dummy robot's Enter direct broadcasting room and welcome sentence, increase the welcome specially good effect to special title.
5. it is according to claim 1 to increase the method that new main broadcaster retains, it is characterised in that the step 1) in direct broadcasting room The dynamic increase and decrease of dummy robot's quantity adopts following methods：When main broadcaster's direct broadcasting room popularity value is less than popularity threshold value T1When, increase The dummy robot of setting quantity, when direct broadcasting room popularity value meets or exceeds popularity threshold value T1When, reduce the simulation of setting quantity Robot；
The quantity for introducing dummy robot is determined by following rule：
f ( x ) = P - T 1 , T 1 < P T 1 - P , T 1 &GreaterEqual; P
Wherein, P is direct broadcasting room popularity value, and the acquisition modes of popularity value are as follows：Each direct broadcasting room is in one data of background maintenance Storehouse, is in real time added up the quantity of the true spectators between current live, obtains the popularity value between current live.
6. it is according to claim 1 to increase the method that new main broadcaster retains, it is characterised in that the step 6) in judge whether There is main broadcaster's speech answering, the method for using is that, based on the Method of Speech Endpoint Detection of entropy, judgement currently has in speech manual Entropy is more than certain threshold value T2, you can judge that current anchor has speech answering, wherein T2Represent the threshold value of speech manual entropy.
7. it is according to claim 6 to increase the method that new main broadcaster retains, it is characterised in that threshold value T of speech manual entropy2For 50。
8. it is a kind of to increase the system that new main broadcaster retains, it is characterised in that to include：
Dummy robot's quantity determining module, by the barrage amount of real-time monitoring main broadcaster place direct broadcasting room, dynamic increase and decrease direct broadcasting room Middle dummy robot's quantity；
Intelligent punctuate module, by the audio data stream for gathering main broadcaster, to audio data stream intelligent punctuate is carried out, according to punctuate knot Audio data stream is divided into voice segment by fruit；
Sound identification module, for calling speech recognition API to be identified the voice segment split, by the language after intelligence punctuate Sound segment is converted to text message M；
Text generation module is replied, for text message M to be sent as input to into intelligent chat robots interface, intelligence chat The reply text R that robot is given；
Barrage module, for according to the text message R that text generation module is obtained is replied, generating barrage, is pushed to main broadcaster；
Main broadcaster replys judge module, for judging whether main broadcaster has speech answering, if it has, then proceed main broadcaster and system it Between barrage it is interactive；If it is determined that main broadcaster does not have speech answering, then the barrage of initial setting up is issued.
9. it is according to claim 8 to increase the method that new main broadcaster retains, it is characterised in that dummy robot's quantity is true Following methods are adopted to the dynamic increase and decrease of direct broadcasting room dummy robot's quantity in cover half block：When main broadcaster's direct broadcasting room popularity value is less than people Gas threshold value T1When, increase the dummy robot of setting quantity, when direct broadcasting room popularity value meets or exceeds popularity threshold value T1When, reduce The dummy robot of setting quantity；
The quantity for introducing robot is determined by following rule：
f ( x ) = P - T 1 , T 1 < P T 1 - P , T 1 &GreaterEqual; P
Wherein, P is direct broadcasting room popularity value, and the acquisition modes of popularity value are as follows：Each direct broadcasting room is in one data of background maintenance Storehouse, is in real time added up the quantity of the true spectators between current live, obtains the popularity value between current live.
10. it is according to claim 8 to increase the method that new main broadcaster retains, it is characterised in that in the intelligent punctuate module Intelligent punctuate method is carried out to audio data stream as follows：Using the Method of Speech Endpoint Detection based on entropy, a language is constructed Message ceases entropy function, to the voice messaging for extracting main broadcaster from one section of voice flow, cancelling noise information；The standard of speech frame Change amplitude spectrum | X (i) | and regard a probability distribution as, the entropy of spectral domain is calculated can be taken the probability of i-th amplitude spectrum to replace taking i-th Probability P (the S of individual source symboli), i.e., represented with below equation：
P ( | X ( i ) | 2 ) = | X ( i ) | 2 &Sigma; i | X ( i ) | 2
So, the entropy of speech manual can be expressed as follows：
H = - &Sigma; i P ( | X ( i ) | 2 ) * log 2 P ( | X ( i ) | 2 )
Calculate and intelligent punctuate is carried out to audio data stream according to entropy size after entropy：For background noise, voice The absolute value of the voice segments amplitude of variation in signal is more than given threshold, and entropy is big；And the absolute value of quiet section of amplitude of variation Less than given threshold, Relatively centralized is distributed, entropy is little.
CN201710014790.2A 2017-01-09 2017-01-09 Method and system for increasing new anchor retention Pending CN106656767A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710014790.2A CN106656767A (en) 2017-01-09 2017-01-09 Method and system for increasing new anchor retention

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710014790.2A CN106656767A (en) 2017-01-09 2017-01-09 Method and system for increasing new anchor retention

Publications (1)

Publication Number Publication Date
CN106656767A true CN106656767A (en) 2017-05-10

Family

ID=58843515

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710014790.2A Pending CN106656767A (en) 2017-01-09 2017-01-09 Method and system for increasing new anchor retention

Country Status (1)

Country Link
CN (1) CN106656767A (en)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107241614A (en) * 2017-07-25 2017-10-10 广州久邦世纪科技有限公司 A kind of live broadcast system
CN108933946A (en) * 2017-05-26 2018-12-04 武汉斗鱼网络科技有限公司 Live streaming concern method, storage medium, electronic equipment and system based on acoustic control
CN109672898A (en) * 2018-11-27 2019-04-23 北京潘达互娱科技有限公司 Sending method, device and the electronic equipment of message in a kind of direct broadcasting room
CN109754808A (en) * 2018-12-13 2019-05-14 平安科技（深圳）有限公司 Method, apparatus, computer equipment and the storage medium of voice conversion text
CN110730390A (en) * 2019-09-23 2020-01-24 杭州铠麒网络科技有限公司 Interaction method and device for intelligently activating atmosphere of live broadcast room
CN111556375A (en) * 2020-04-24 2020-08-18 西交利物浦大学 Video barrage generation method and device, computer equipment and storage medium
CN111695974A (en) * 2020-06-18 2020-09-22 上海适享文化传播有限公司 Method for using retail knowledge database at any time and any place in live broadcast
CN110267052B (en) * 2019-06-19 2021-04-16 云南大学 Intelligent barrage robot based on real-time emotion feedback
CN112887751A (en) * 2021-02-03 2021-06-01 北京城市网邻信息技术有限公司 Data processing method and device for live broadcast room
CN113365096A (en) * 2021-07-05 2021-09-07 湖南快乐阳光互动娱乐传媒有限公司 Live broadcast room warm-up method and device, storage medium and electronic equipment

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101415257A (en) * 2007-10-16 2009-04-22 康佳集团股份有限公司 Man-machine conversation chatting method
CN104598445A (en) * 2013-11-01 2015-05-06 腾讯科技（深圳）有限公司 Automatic question-answering system and method
CN105721891A (en) * 2016-03-30 2016-06-29 宁波元鼎电子科技有限公司 Method and system for interaction between anchor and audiences in webcast process
CN105872620A (en) * 2016-03-30 2016-08-17 宁波元鼎电子科技有限公司 Authority and level distinguished display method and authority and level distinguished display system in network broadcasting process

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101415257A (en) * 2007-10-16 2009-04-22 康佳集团股份有限公司 Man-machine conversation chatting method
CN104598445A (en) * 2013-11-01 2015-05-06 腾讯科技（深圳）有限公司 Automatic question-answering system and method
CN105721891A (en) * 2016-03-30 2016-06-29 宁波元鼎电子科技有限公司 Method and system for interaction between anchor and audiences in webcast process
CN105872620A (en) * 2016-03-30 2016-08-17 宁波元鼎电子科技有限公司 Authority and level distinguished display method and authority and level distinguished display system in network broadcasting process

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
广州日报: "网红猎头揭行业内幕：粉丝靠买、流量靠刷、送礼靠托", 《信息与电脑》 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108933946A (en) * 2017-05-26 2018-12-04 武汉斗鱼网络科技有限公司 Live streaming concern method, storage medium, electronic equipment and system based on acoustic control
CN108933946B (en) * 2017-05-26 2021-01-01 武汉斗鱼网络科技有限公司 Voice-control-based live broadcasting attention method, storage medium, electronic equipment and system
CN107241614A (en) * 2017-07-25 2017-10-10 广州久邦世纪科技有限公司 A kind of live broadcast system
CN109672898A (en) * 2018-11-27 2019-04-23 北京潘达互娱科技有限公司 Sending method, device and the electronic equipment of message in a kind of direct broadcasting room
CN109754808A (en) * 2018-12-13 2019-05-14 平安科技（深圳）有限公司 Method, apparatus, computer equipment and the storage medium of voice conversion text
CN110267052B (en) * 2019-06-19 2021-04-16 云南大学 Intelligent barrage robot based on real-time emotion feedback
CN110730390A (en) * 2019-09-23 2020-01-24 杭州铠麒网络科技有限公司 Interaction method and device for intelligently activating atmosphere of live broadcast room
CN111556375A (en) * 2020-04-24 2020-08-18 西交利物浦大学 Video barrage generation method and device, computer equipment and storage medium
CN111556375B (en) * 2020-04-24 2022-10-11 西交利物浦大学 Video barrage generation method and device, computer equipment and storage medium
CN111695974A (en) * 2020-06-18 2020-09-22 上海适享文化传播有限公司 Method for using retail knowledge database at any time and any place in live broadcast
CN112887751A (en) * 2021-02-03 2021-06-01 北京城市网邻信息技术有限公司 Data processing method and device for live broadcast room
CN113365096A (en) * 2021-07-05 2021-09-07 湖南快乐阳光互动娱乐传媒有限公司 Live broadcast room warm-up method and device, storage medium and electronic equipment

Similar Documents

Publication Publication Date Title
CN106656767A (en) Method and system for increasing new anchor retention
CN110381388B (en) Subtitle generating method and device based on artificial intelligence
WO2017084197A1 (en) Smart home control method and system based on emotion recognition
CN107437415B (en) Intelligent voice interaction method and system
JP2020034895A (en) Responding method and device
CN103514883B (en) A kind of self-adaptation realizes men and women&#39;s sound changing method
CN112863547A (en) Virtual resource transfer processing method, device, storage medium and computer equipment
WO2015107741A1 (en) Information processing device, information processing method, and program
CN106847305B (en) Method and device for processing recording data of customer service telephone
CN107645523A (en) A kind of method and system of mood interaction
CN101834809B (en) Internet instant message communication system
CN109785838B (en) Voice recognition method, device, equipment and storage medium
JP2017522626A (en) Information transmitting method and transmitting apparatus
CN108010541A (en) Method and device, the storage medium of pitch information are shown in direct broadcasting room
CN114938360A (en) Data processing method and device based on instant messaging application
CN109327614B (en) Global simultaneous interpretation mobile phone and method
CN111050023A (en) Video detection method and device, terminal equipment and storage medium
CN106899486A (en) A kind of message display method and device
JP2012181358A (en) Text display time determination device, text display system, method, and program
CN107451131A (en) A kind of audio recognition method and device
KR20170135598A (en) System and Method for Voice Conversation using Synthesized Virtual Voice of a Designated Person
CN111429895A (en) Semantic understanding method and device for multi-round interaction and computer storage medium
CN105727572B (en) A kind of self-learning method and self study device based on speech recognition of toy
CN112309365A (en) Training method and device of speech synthesis model, storage medium and electronic equipment
CN101460994A (en) Speech differentiation

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20170510