CN108877807A - A kind of intelligent robot for telemarketing - Google Patents
A kind of intelligent robot for telemarketing Download PDFInfo
- Publication number
- CN108877807A CN108877807A CN201810726777.4A CN201810726777A CN108877807A CN 108877807 A CN108877807 A CN 108877807A CN 201810726777 A CN201810726777 A CN 201810726777A CN 108877807 A CN108877807 A CN 108877807A
- Authority
- CN
- China
- Prior art keywords
- voice
- speech
- client
- voice signal
- module
- 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.)
- Withdrawn
Links
Classifications
-
- 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
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J11/00—Manipulators not otherwise provided for
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0281—Customer communication at a business location, e.g. providing product or service information, consulting
Abstract
The invention discloses a kind of intelligent robot for telemarketing, which includes：Voice acquisition module, speech processing module, speech recognition module and voice broadcast module.Voice acquisition module, for acquiring the voice messaging of client；Speech processing module obtains the first voice signal for handling the voice messaging of acquisition；Speech recognition module, for confirming the identity information of client and the business demand of client according to the first voice signal；Voice broadcast module, for the business demand according to client, to the corresponding business information of pushes customer.It is customer service that intelligent robot provided by the invention, which can substitute telemarketing personnel, avoid previous telemarketing personnel attitude when carrying out telemarketing it is unfriendly and pointedly for offering customers service the shortcomings that, greatly improve the success rate of telemarketing, and human resources have been saved, there is application value.
Description
Technical field
The present invention relates to intelligent robot technology field, specially a kind of intelligent robot for telemarketing.
Background technique
Traditional telemarketing method needs sales force to promote the sale of goods for client, and this method not only needs to put into largely
Fund, equipment, and need to be equipped with more sales force, also to carry out the staff training of long period, while Service Quality
Amount is also deep to be restricted by the quality and portfolio size of sales force.
Summary of the invention
In view of the above-mentioned problems, the present invention is intended to provide a kind of intelligent robot for telemarketing.
The purpose of the present invention is realized using following technical scheme：
A kind of intelligent robot for telemarketing, the intelligent robot include：Voice acquisition module, speech processes mould
Block, speech recognition module and voice broadcast module.
Voice acquisition module, for acquiring the voice messaging of client；Speech processing module, for the voice messaging to acquisition
It is handled, obtains the first voice signal；Speech recognition module, for confirming the identity letter of client according to the first voice signal
The business demand of breath and client；Voice broadcast module, for the business demand according to client, to the corresponding business of pushes customer
Information.
Beneficial effect：The present invention provides a kind of intelligent robot for telemarketing, which can be replaced
For telemarketing personnel be customer service, avoid previous telemarketing personnel carry out telemarketing when attitude it is unfriendly with
And not pointedly be offering customers service the shortcomings that, greatly improve the success rate of telemarketing, and saved people
Power resource has application value.
Detailed description of the invention
The present invention will be further described with reference to the accompanying drawings, but the embodiment in attached drawing is not constituted to any limit of the invention
System, for those of ordinary skill in the art, without creative efforts, can also obtain according to the following drawings
Other attached drawings.
Fig. 1 is frame construction drawing of the invention；
Fig. 2 is the frame construction drawing of speech processing module of the present invention；
Fig. 3 is the frame construction drawing of speech recognition module of the present invention.
Appended drawing reference：
Voice acquisition module 1；Speech processing module 2；Speech recognition module 3；Voice broadcast module 4；Database 5；Client
Identity recognizing unit 31；Voice content recognition unit 32；Speech enhancement unit 21；Speech de-noising unit 22；Pre-process subelement
221；Detection sub-unit 222；Update subelement 223.
Specific embodiment
In conjunction with following application scenarios, the invention will be further described.
Fig. 1 shows a kind of intelligent robot for telemarketing, the intelligent robot include voice acquisition module 1,
Speech processing module 2, speech recognition module 3 and voice broadcast module 4.
Voice acquisition module 1, for acquiring the voice messaging of client；Speech processing module 2, for the voice letter to acquisition
Breath is handled, and the first voice signal is obtained；Speech recognition module 3, for confirming the identity of client according to the first voice signal
Information and the business demand of client；Voice broadcast module 4, it is corresponding to pushes customer for the business demand according to client
Business information.
Beneficial effect：The present invention provides a kind of intelligent robot for telemarketing, which can be replaced
For telemarketing personnel be customer service, avoid previous telemarketing personnel carry out telemarketing when attitude it is unfriendly with
And not pointedly be offering customers service the shortcomings that, greatly improve the success rate of telemarketing, and saved people
Power resource has application value.
Preferably, which further includes database 5, and database 5 is used to store various businesses information and client
Identity information.
Preferably, referring to fig. 2, speech recognition module 3 includes client identity recognition unit 31 and voice content recognition unit
32, client identity recognition unit 31, for extracting the voiceprint of client, confirming the identity of client according to the first voice signal
Information, and the identity information of its client is stored to database 5；Voice content recognition unit 32, for according to first language
Sound signal extracts the characteristic parameter that can describe client traffic demand, and determines that client traffic needs in first voice signal
The matching degree of the characteristic parameter of the business information prestored in the characteristic parameter and the database 5 asked, when the matching journey of the two
When degree is greater than presetting threshold value, by the voice broadcast module 4 to the corresponding business information of pushes customer, specifically,
It is the business information for being greater than presetting threshold value to pushes customer matching degree, the business information in the present embodiment can be commodity
Or service.
Beneficial effect：In the above-described embodiments, client identity recognition unit 31 and voice content recognition unit is respectively set
32, confirm for the identity to client, while the client identity information that will confirm that is stored into database, which has
Conducive to the company identity information of client and the demand information of client in hand, be conducive to maintain the cooperation between company and client
Relationship, facilitating company is the business tine that pushes customer is adapted therewith.
Preferably, referring to Fig. 3, speech processing module 2 includes speech enhancement unit 21 and speech de-noising unit 22, and voice increases
Strong unit 21 is used to carry out the voice signal of acquisition speech enhan-cement processing, and speech de-noising unit 22 is for enhancing, treated
Voice signal carries out denoising, obtains the first voice signal.
Preferably, speech enhan-cement processing is carried out to the voice signal of acquisition, including：
(1) voice signal of acquisition is transformed from the time domain in frequency domain using Fourier transformation；
(2) speech enhan-cement processing is carried out to the voice signal in frequency domain according to customized speech enhan-cement formula, is increased
Voice signal after strong, wherein customized speech enhan-cement formula is：
In formula,Indicate that enhanced treated voice signal, D (t, d) refer to prior weight, Y (t, d) table
The voice signal in frequency domain before showing enhancing processing, p (t, d) indicate probability value existing for voice, G at t frame d frequency bandminTable
Show that minimum gain value in the state of non-speech frame, t indicate that frame number, d indicate frequency, GthIt is the state in speech frame of setting
Under yield value；
(3) inverse Fourier transform is used from frequency-domain transform to time domain, to obtain enhanced voice signal in time domain
Enhanced voice signal.
Beneficial effect：It is transformed in frequency domain using the voice signal that Fourier transformation will acquire and is believed to voice in frequency domain
Number carry out enhancing processing, wherein in customized speech enhan-cement function, by introducing p (t, d) so that non-speech frame when
It waits, it is also considered that estimate the amplitude of clean speech, while introducing GminVoice distortion is also further reduced, while residual error being made an uproar
Sound is changed into white noise, allows people to sound smoother, more comfortably；The algorithm can preferably retain the main letter of voice signal
Breath, removes the ambient noise of low frequency part.When carrying out enhancing processing to voice signal using speech enhan-cement function, G is introducedthInto
One step corrects the voice signal in speech frame, which further can further enhance processing to efficient voice signal, have
Conducive to the subsequent extraction to characteristic parameter in efficient voice signal.
Preferably, further to obtain purer voice signal, using following formula to customized speech enhan-cement function
In prior weight D (t, d) be modified, and by revised D (t, d) replace speech enhan-cement function in D (t, d),
In, the calculation formula for correcting prior weight is：
In above formula,Indicate that revised prior weight, θ (t, d) indicate that posteriori SNR, v are the tune of setting
The factor is saved, for being balanced control to noise.
When carrying out enhancing processing to voice signal, by introducing revised prior weightTo replace priori
Signal-to-noise ratio D (t, d), the way can further decrease error, improve the progress speech enhan-cement processing under low signal-to-noise ratio scene
When reinforcing effect, while to prior weightWhen being modified, noise, which was easy to appear, to be estimated and owes the case where estimating,
By introducing regulatory factor v (usual v choose 0.75), it is subsequent noise power spectrum is estimated when, can effectively prevent
The generation for the case where estimating is estimated and owes, so that voice distortion is minimum.
Preferably, speech de-noising unit 22 includes pretreatment subelement 221, detection sub-unit 222 and updates subelement
223。
It is multiple subbands in Time Domain Decomposition by enhanced voice signal that subelement 221, which is pre-processed, using acoustic filter group
Signal；Detection sub-unit 222 is used to carry out energy measuring to each subband signal, estimates region existing for effective audio；More
New subelement 223 is used for the testing result according to detection sub-unit 222, is updated to the estimation of noise energy value of present frame,
And updated estimation of noise energy value is applied in detection sub-unit 222, and then realizes to region existing for effective audio
Update.
In one embodiment, described that energy measuring is carried out to each subband signal, including：
(1) sub-frame processing is carried out to the signal of each subband；
(2) using energy balane formula calculate frame signal of each subband signal after sub-frame processing energy estimators and
Initial noisc energy estimators, wherein the energy balane formula of j subband is：
In formula, S (k, i, j)2It is the amplitude of k-th of sampled point in the i-th frame of j subband, LenfrFor the length of a frame signal,
Efr(i, j) is the energy estimators of the i-th frame signal, Enoise(i, j) is the initial noisc energy estimation of the i-th frame signal of j subband
Value；M is the preceding m frame signal of subband j；
(2) according to obtaining energy estimators and initial noisc energy estimators, using determine formula to present frame whether be
Effective audio is determined, wherein determines that formula is：
In formula, η1And η2For threshold value weighting coefficient, for adjusting threshold size, and meet η1< η2；Work as Efr(i, j) < η1Enoise(i, j), Vcross(i, j)=0, the i-th frame is noise frame at this time, works as Efr(i, j) > η2EnoiseWhen (i, j), Vcross(i, j)
=1, the i-th frame is speech frame, in the other cases, V at this timecross(i, j) is consistent with the differentiation result of previous frame；
(3) all frames in all subbands are traversed, region existing for effective audio in each subband is obtained.
(4) union is sought into region existing for obtained effective audio, effective sound of enhanced audio signal can be obtained
Region existing for frequency.
Beneficial effect：This algorithm sets absolute threshold η by using dual threshold discriminant approach2Enoise(i, j) and protection
Threshold value η1Enoise(i, j) two decision thresholds, can be avoided a possibility that judging by accident when determining using single threshold value, the differentiation
The accuracy that mode detects is higher than single threshold value detection algorithm, can better discriminate between speech frame and noise frame.And the detection list
Member is detected for each subband, which can ignore that the otherness that noise energy itself is distributed in different frequency sections,
The noise range used is wider.
In one embodiment, the testing result according to detection sub-unit 222, to the estimation of noise energy of present frame
Value is updated, and updated estimation of noise energy value is applied in the detection unit, and then is existed to effective audio
Region be updated, work as satisfaction | Enoise k(i, j)-Enoise k-1(i, j) | when≤σ, then stop estimating the noise energy of the i-th frame
Evaluation is updated, and σ is preset constant, and k is the number of iterations, wherein is updated to the estimation of noise energy value of present frame
More new formula be：
In formula, R (i, i-1) indicates the related coefficient of the estimation of noise energy value of the i-th frame and the (i-1)-th frame；R (i, i-2) table
Show the related coefficient of the estimation of noise energy value of the i-th frame and the i-th -2 frame.
Beneficial effect：It is updated by the estimation of noise energy value in real time to present frame, and then by updated noise
Energy estimators are applied among detection sub-unit 222, further correct the range of effective audio region, which improves
Adaptability under different occasions, that is, improve robustness.And when the estimation of noise energy value to present frame is updated, when
Determine when the i-th frame is noise frame as the result is shown, it is contemplated that the estimation of noise energy value of the (i-1)-th frame and the i-th -2 frame, and with i-th
The degree of correlation of frame can more accurately be updated estimation of noise energy value.
Finally it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than the present invention is protected
The limitation of range is protected, although explaining in detail referring to preferred embodiment to the present invention, those skilled in the art are answered
Work as analysis, it can be with modification or equivalent replacement of the technical solution of the present invention are made, without departing from the reality of technical solution of the present invention
Matter and range.
Claims (5)
1. a kind of intelligent robot for telemarketing, which is characterized in that including：Voice acquisition module, speech processing module,
Speech recognition module and voice broadcast module；
The voice acquisition module, for acquiring the voice messaging of client；
The speech processing module obtains the first voice signal for handling the voice messaging of acquisition；
The speech recognition module, for confirming the identity information of client and the industry of client according to first voice signal
Business demand；
The voice broadcast module, for the business demand according to the client, to the corresponding business information of pushes customer.
2. intelligent robot according to claim 1, which is characterized in that further include database, the database is for depositing
Store up the characteristic parameter information of various businesses information and the identity information of client.
3. intelligent robot according to claim 2, which is characterized in that the speech recognition module includes that client identity is known
Other unit and voice content recognition unit；
The client identity recognition unit confirms client for extracting the voiceprint of client according to first voice signal
Identity information, and the identity information for the client that will confirm that is stored to the database；
The voice content recognition unit, for according to first voice signal, extraction can to describe client traffic demand
Characteristic parameter, and determine the industry prestored in the characteristic parameter and the database of client traffic demand in first voice signal
The matching degree of the characteristic parameter for information of being engaged in is broadcast when the matching degree of the two is greater than presetting threshold value by the voice
Report module to the corresponding business information of pushes customer.
4. intelligent robot according to claim 3, which is characterized in that the speech processing module includes speech enhan-cement list
Member and speech de-noising unit；
The speech enhancement unit, for carrying out speech enhan-cement processing to the voice signal of acquisition；
The speech de-noising unit obtains the first voice signal for carrying out denoising to enhancing treated voice signal.
5. intelligent robot according to claim 4, which is characterized in that the voice signal of described pair of acquisition carries out voice increasing
Strength reason, including：
(1) voice signal of acquisition is transformed from the time domain in frequency domain using Fourier transformation；
(2) speech enhan-cement processing is carried out to the voice signal in frequency domain according to customized speech enhan-cement formula, after obtaining enhancing
Voice signal, wherein customized speech enhan-cement formula is：
In formula,Indicate that enhanced treated voice signal, D (t, d) refer to that prior weight, Y (t, d) indicate to increase
The voice signal in frequency domain before the reason of strength, p (t, d) indicate probability value existing for voice, G at t frame d frequency bandminIt indicates
Minimum gain value in the state of non-speech frame, t indicate that frame number, d indicate frequency, GthIt is setting in the state of speech frame
Yield value；
(3) use inverse Fourier transform that enhanced voice signal from frequency-domain transform to time domain, is obtained the enhancing in time domain
Voice signal afterwards.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810726777.4A CN108877807A (en) | 2018-07-04 | 2018-07-04 | A kind of intelligent robot for telemarketing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810726777.4A CN108877807A (en) | 2018-07-04 | 2018-07-04 | A kind of intelligent robot for telemarketing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108877807A true CN108877807A (en) | 2018-11-23 |
Family
ID=64299161
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810726777.4A Withdrawn CN108877807A (en) | 2018-07-04 | 2018-07-04 | A kind of intelligent robot for telemarketing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108877807A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109773806A (en) * | 2019-02-28 | 2019-05-21 | 利哲科技(厦门)股份有限公司 | A kind of electricity Xiao AI robot based on artificial intelligence |
| CN110060098A (en) * | 2019-04-04 | 2019-07-26 | 秒针信息技术有限公司 | The acquisition methods and device of marketing term |
| CN112530453A (en) * | 2020-11-27 | 2021-03-19 | 五邑大学 | Voice recognition method and device suitable for noise environment |
| CN112947468A (en) * | 2021-03-11 | 2021-06-11 | 深圳市九码云科技有限公司 | Robot for telephone sales |
-
2018
- 2018-07-04 CN CN201810726777.4A patent/CN108877807A/en not_active Withdrawn
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109773806A (en) * | 2019-02-28 | 2019-05-21 | 利哲科技(厦门)股份有限公司 | A kind of electricity Xiao AI robot based on artificial intelligence |
| CN110060098A (en) * | 2019-04-04 | 2019-07-26 | 秒针信息技术有限公司 | The acquisition methods and device of marketing term |
| CN112530453A (en) * | 2020-11-27 | 2021-03-19 | 五邑大学 | Voice recognition method and device suitable for noise environment |
| CN112530453B (en) * | 2020-11-27 | 2022-04-05 | 五邑大学 | Voice recognition method and device suitable for noise environment |
| CN112947468A (en) * | 2021-03-11 | 2021-06-11 | 深圳市九码云科技有限公司 | Robot for telephone sales |
| CN112947468B (en) * | 2021-03-11 | 2023-08-22 | 深圳进行时电子科技有限公司 | Robot for telephone sales |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108877807A (en) | A kind of intelligent robot for telemarketing | |
| EP0689194B1 (en) | Method of and apparatus for signal recognition that compensates for mismatching | |
| Lin et al. | Adaptive noise estimation algorithm for speech enhancement | |
| EP0470245B1 (en) | Method for spectral estimation to improve noise robustness for speech recognition | |
| Govindan et al. | Adaptive wavelet shrinkage for noise robust speaker recognition | |
| CN101154383B (en) | Method and device for noise suppression, phonetic feature extraction, speech recognition and training voice model | |
| CN109616139A (en) | Pronunciation signal noise power spectral density estimation method and device | |
| CN108899034A (en) | It is a kind of with the intellectual nursing bed of voice control function | |
| Borgström et al. | HMM-based reconstruction of unreliable spectrographic data for noise robust speech recognition | |
| Alam et al. | Robust feature extraction for speech recognition by enhancing auditory spectrum | |
| Fu et al. | Perceptual wavelet adaptive denoising of speech. | |
| WO2007041789A1 (en) | Front-end processing of speech signals | |
| US6687672B2 (en) | Methods and apparatus for blind channel estimation based upon speech correlation structure | |
| de Veth et al. | Acoustic backing-off as an implementation of missing feature theory | |
| Van Hamme | Robust speech recognition using cepstral domain missing data techniques and noisy masks | |
| CN108922529A (en) | A kind of vehicle-mounted voice control system | |
| Fu et al. | A novel speech enhancement system based on wavelet denoising | |
| Bolisetty et al. | Speech enhancement using modified wiener filter based MMSE and speech presence probability estimation | |
| KR101124712B1 (en) | A voice activity detection method based on non-negative matrix factorization | |
| Zheng et al. | SURE-MSE speech enhancement for robust speech recognition | |
| Loweimi et al. | Channel Compensation in the Generalised Vector Taylor Series Approach to Robust ASR. | |
| Surendran et al. | Perceptual subspace speech enhancement with ssdr normalization | |
| KR100570873B1 (en) | Voice recognition apparatus using feature requital of parallel combined mixture model and the method | |
| Delcroix et al. | Combined static and dynamic variance adaptation for efficient interconnection of speech enhancement pre-processor with speech recognizer | |
| Schless et al. | Adaptive model combination for robust speech recognition in car environments |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20181123 |
|
| WW01 | Invention patent application withdrawn after publication |