CN112788184A - Method for connecting call center according to voice input - Google Patents
Method for connecting call center according to voice input Download PDFInfo
- Publication number
- CN112788184A CN112788184A CN202110062248.0A CN202110062248A CN112788184A CN 112788184 A CN112788184 A CN 112788184A CN 202110062248 A CN202110062248 A CN 202110062248A CN 112788184 A CN112788184 A CN 112788184A
- Authority
- CN
- China
- Prior art keywords
- function
- user
- voice
- frame
- functions
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M3/00—Automatic or semi-automatic exchanges
- H04M3/42—Systems providing special services or facilities to subscribers
- H04M3/487—Arrangements for providing information services, e.g. recorded voice services or time announcements
- H04M3/493—Interactive information services, e.g. directory enquiries ; Arrangements therefor, e.g. interactive voice response [IVR] systems or voice portals
- H04M3/4936—Speech interaction details
-
- 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
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
-
- 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
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- 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
- G10L15/28—Constructional details of speech recognition systems
- G10L15/30—Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/18—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being spectral information of each sub-band
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/21—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being power information
Abstract
The invention provides a method for connecting a call center according to voice input, which comprises the following steps: s100, according to the telephone number identified by the incoming call, carrying out preference sorting of the functions used by the user, and providing the corresponding function list sequence to the user by the preference sorting; s200, acquiring voice feedback of a user, performing recognition analysis, and matching the voice feedback with a fixed phrase in a response system; s300, acquiring the function associated with the matched fixed phrase in the response system, and automatically jumping to the associated function. The invention carries out preference sequencing of the functions used by the user in a targeted manner through the telephone number identified by the incoming call, provides a sequenced function sequence list, carries out identification analysis on the voice feedback after the voice feedback of the user is obtained, matches the voice feedback with the fixed phrase in the response system, obtains the corresponding function in the response system through the incidence relation of the fixed phrase, and directly jumps to the function, thereby providing rapid service function requirements for the user.
Description
Technical Field
The invention relates to the technical field of intelligent voice response, in particular to a method for connecting a call center according to voice input.
Background
With the development and progress of the technology, more and more units utilize the intelligent voice response system to provide response service for the telephone feedback of the user. As the user demands are more and more diversified, each unit has to continuously expand the service function of the intelligent voice response system, and has to classify the service function in order to shorten the prompt duration, so that the service function level in the intelligent voice response system is too much, and the user needs to perform multi-step complex operation to obtain the desired service function when using the intelligent voice response system.
At present, after a customer accesses a call center in a telephone mode, the customer navigates to different hierarchical menus through regular key input according to function hierarchy, more sequences and combinations are added to an interface along with the increase of the number of functions, and the user generally needs to wait for a complex prompt and then input a complex key to enter the function, so that the user experience is greatly reduced.
Disclosure of Invention
In order to solve the above technical problem, the present invention provides a method for connecting a call center according to a voice input, comprising the steps of:
s100, according to the telephone number identified by the incoming call, carrying out preference sorting of the functions used by the user, and providing the corresponding function list sequence to the user by the preference sorting;
s200, acquiring voice feedback of a user, performing recognition analysis, and matching the voice feedback with a fixed phrase in a response system;
s300, acquiring the function associated with the matched fixed phrase in the response system, and automatically jumping to the associated function.
Optionally, in step S100, the attribution is identified according to the phone number of the incoming call, if the answering system identifies the attribution of the phone number, the dialect of the attribution is used to prompt the voice response, otherwise, the standard mandarin is used to prompt the voice response.
Optionally, in step S100, the process of sorting the preferences of the user using the function includes:
s110, judging whether a previous call is received or not through a telephone number according to the call, if the answer system has a call record of the telephone number and the call record has a corresponding use function, adding the corresponding use function into a function list, and marking the level to be 3;
s120, all incoming calls and corresponding use functions of the telephone number attribution of the current incoming call are obtained from the answering system, the function with the most use frequency is determined, and the function list is added, wherein the marking level is 2;
s130, determining the function with the most frequent use of all user calls in the answering system, and adding the function into a function list with the marking level of 1;
s140 performs preference sorting of functions according to the mark level based on the function list obtained through the steps of S110, S120, and S130.
Optionally, if the number of previous incoming calls of the current incoming call user is large and the number of functions used by the incoming call is more than two, sorting the incoming calls according to the number of function use times, wherein the number of use times is arranged in front; if a plurality of functions with the same use times exist, sorting is carried out according to the time from the near to the far away from the current incoming call.
Optionally, in step S200, the process of performing recognition analysis on the voice feedback of the user is as follows:
s210, framing the voice feedback according to the time length to ensure that the time length of each frame can contain enough information for performing initial consonant and vowel judgment;
s220, converting each frame into a feature vector, respectively identifying by adopting short-time Fourier analysis, and sequentially determining the initial consonants and the vowels to which the frames belong;
and S230, matching with the fixed phrases in the response system according to the initial and final voice characteristics of the sub-frames.
Optionally, if the matching with the fixed phrase in the response system fails, a prompt is sent to the user, and the sequence list of the corresponding functions is fed back to the user again by preference sorting, so that the user is required to perform voice feedback again.
Optionally, in step S200, the voice feedback of the user is stored as an audio file.
Optionally, in step S220, the short-time fourier analysis includes analyzing each frame of the speech signal by using the following fourier transform function:
in the above formula, the first and second carbon atoms are,
representing signal spectrum information extracted from the speech signal of the nth frame; n represents the framing of the speech feedback; x (n) represents the nth frame sampling speech source signal of the speech feedback; w (nt-m) represents a window function; t represents a frame shift duration; e represents a natural constant; j denotes a virtual unit, i.e. j2-1; k represents angular frequency, k is more than or equal to 0 and less than or equal to N-1; n represents the frame length;
and then, performing initial consonant and vowel identification on the frequency spectrum information of each frame of signal obtained by Fourier transform.
Optionally, in step S210, after framing the voice feedback according to the time length, and before performing the identification analysis, preprocessing each frame of the voice feedback, where the preprocessing process is as follows:
filtering each frame of voice by adopting a filter with 32 channels;
solving the spectral energy of each frame of voice by adopting the following formula:
in the above equation, Q (i, j) represents the spectral energy of the jth spectral coefficient vector in the ith frame; u denotes the number of channels of the filter, i.e., u is 32; envM(k, i) an envelope spectrum representing the ith frame of speech for the kth channel;
and extracting a 12-dimensional spectral coefficient vector of each frame of voice according to the calculation result, and taking the obtained spectral coefficient vector as a feature vector of voice recognition.
Optionally, in step S100, the user preference value of the function is calculated by using the following formula:
in the above formula, τiA user preference value indicating an ith function; k is a radical of1All users representing the ith function use the weight; m represents the number of users; n represents the number of functions in the answering system; w is aijRepresenting the number of times that the jth user uses the ith function; k is a radical of2Representing the current incoming call user use weight of the ith function; w is aiRepresenting the number of times that the current incoming call user uses the ith function;
and according to the user preference values of the functions obtained by calculation, sorting the preference of the functions according to the user preference values from big to small to obtain a function list.
According to the method for connecting the call center according to the voice input, the preference ordering of the use functions is carried out on the user of the telephone number through the telephone number identified by the incoming call, the functions preferred by the user are arranged in front, the ordered function sequence list is provided, after the voice feedback of the user is obtained, the voice feedback is identified and analyzed, the voice feedback is matched with the fixed phrase in the response system, the corresponding function in the response system is obtained through the incidence relation of the fixed phrase, and the function is directly jumped to, so that the rapid service function requirement is provided for the user; the user is prevented from carrying out complex operation of a plurality of steps, user dislike and mood dysphoria caused by the complexity of the operation are prevented, the user satisfaction and good experience are improved, and the user loss is reduced.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a flow chart of a method for connecting a call center based on voice input in an embodiment of the present invention;
FIG. 2 is a diagram illustrating a preference ranking process for user usage functions in an embodiment of a method for connecting a call center based on voice input in accordance with the present invention;
FIG. 3 is a schematic diagram illustrating a process of performing recognition analysis on a user's voice feedback according to an embodiment of a method for connecting a call center according to a voice input;
fig. 4 shows an example of an application in which the method of the invention is used.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
As shown in fig. 1, an embodiment of the present invention provides a method for connecting a call center according to a voice input, including the following steps:
s100, according to the telephone number identified by the incoming call, carrying out preference sorting of the functions used by the user, and providing the corresponding function list sequence to the user by the preference sorting;
s200, acquiring voice feedback of a user, performing recognition analysis, and matching the voice feedback with a fixed phrase in a response system;
s300, acquiring the function associated with the matched fixed phrase in the response system, and automatically jumping to the associated function.
The working principle and the beneficial effects of the technical scheme are as follows: according to the scheme, preference sorting of using functions is carried out on users of the telephone numbers according to the telephone numbers identified by incoming calls, the functions preferred by the users are arranged in front, a sorted function sequence list is provided, after voice feedback of the users is obtained, the voice feedback is identified and analyzed, the voice feedback is matched with fixed phrases in a response system, corresponding functions in the response system are obtained through the incidence relation of the fixed phrases, the functions are directly skipped to, and rapid service function requirements are provided for the users; the user is prevented from carrying out complex operation of a plurality of steps, user dislike and mood dysphoria caused by the complexity of the operation are prevented, the user satisfaction and good experience are improved, and the user loss is reduced.
In one embodiment, in step S100, a home location is identified according to the phone number of the incoming call, and if the answering system identifies the home location of the phone number, the answering system uses the dialect of the home location to prompt the voice response, otherwise, the answering system uses the standard mandarin to prompt the voice response.
The working principle and the beneficial effects of the technical scheme are as follows: according to the scheme, the home location identification is carried out on the telephone number of the incoming call, the common language of the user is judged, the common language of the user is adopted in the communication of the response system, the language habit of the user is respected, the use requirements of users in areas with limited education degrees or different official wording areas can be met, the home location feeling and good experience of the user are enhanced, and the loyalty of the user can be improved.
In one embodiment, as shown in fig. 2, the performing of the preference ranking process of the user usage function in the S100 step includes:
s110, judging whether a previous call is received or not through a telephone number according to the call, if the answer system has a call record of the telephone number and the call record has a corresponding use function, adding the corresponding use function into a function list, and marking the level to be 3;
s120, all incoming calls and corresponding use functions of the telephone number attribution of the current incoming call are obtained from the answering system, the function with the most use frequency is determined, and the function list is added, wherein the marking level is 2;
s130, determining the function with the most frequent use of all user calls in the answering system, and adding the function into a function list with the marking level of 1;
s140, sorting the preference of the functions according to the marking level according to the function list obtained after the steps of S110, S120 and S130;
if the previous incoming call times of the current incoming call user are more and the number of the used functions is more than two, sorting the incoming call according to the number of the used functions, wherein the used functions are arranged in front; if a plurality of functions with the same use times exist, sorting the functions from near to far away from the current incoming call time; if the function list is required to be included for the functions contained in the response system except the functions already added into the function list, the functions are ranked after the ranking; the response system provides the voice to the user according to the sequence of the function list, allows the user to perform voice feedback in the function list providing process (namely when the voice prompt is not complete), stops subsequent function voice prompts in the function list once the voice feedback of the user is received, and enters the next step to perform recognition analysis on the voice feedback of the user, so that the service efficiency is improved.
The working principle and the beneficial effects of the technical scheme are as follows: the method provides a preferred preference sequencing mode of the user use functions, the function with the highest use frequency of the user counted in the answering system is taken as a first preferred level, the function with the highest use frequency of the user at the current calling user attribution counted in the answering system is taken as a second preferred level, the function used by the calling user through the calling before is taken as a third preferred level, preference sequencing is carried out according to the preferred levels, and the function with the high calling user selection probability is ranked in front, so that the user can quickly select, the user service efficiency is improved, and the user experience is improved.
In one embodiment, as shown in fig. 3, in step S200, the process of performing recognition analysis on the voice feedback of the user is as follows:
s210, framing the voice feedback according to the time length to ensure that the time length of each frame can contain enough information for performing initial consonant and vowel judgment;
s220, converting each frame into a feature vector, respectively identifying by adopting short-time Fourier analysis, and sequentially determining the initial consonants and the vowels to which the frames belong;
s230, matching with a fixed phrase in the response system according to the initial and final voice characteristics of the sub-frames;
and if the matching fails, sending a prompt to the user, feeding back the sequence list of the corresponding functions to the user again by preference sorting, and requesting the client to perform voice feedback again.
The working principle and the beneficial effects of the technical scheme are as follows: the scheme divides the signal of the voice feedback into shorter frames, and the frame size is usually between 20ms and 40ms, for example, the frame size can be set to 25 ms; extracting a characteristic vector for each frame after framing, then respectively identifying by adopting short-time Fourier analysis, determining the corresponding initial consonant and vowel, and matching with a fixed phrase in a response system; through the processing process of the voice feedback, the meaning or the idea to be expressed by the voice feedback is analyzed, and the corresponding fixed phrase is found in the response system, so that the requirement that a client repeatedly and repeatedly feeds back the content is avoided, and the user dislike is prevented from being caused.
In one embodiment, in the S200 step, the voice feedback of the user is stored as an audio file; in step S220, the short-time fourier analysis includes analyzing each frame of the speech signal by using the following fourier transform function:
in the above formula, the first and second carbon atoms are,
representing signal spectrum information extracted from the speech signal of the nth frame; n represents the framing of the speech feedback; x (n) represents the nth frame sampling speech source signal of the speech feedback; w (nt-m) represents a window function; t represents a frame shift duration; e represents a natural constant; j denotes a virtual unit, i.e. j2-1; k represents angular frequency, k is more than or equal to 0 and less than or equal to N-1; n represents the frame length;
and then, performing initial consonant and vowel identification on the frequency spectrum information of each frame of signal obtained by Fourier transform.
The working principle and the beneficial effects of the technical scheme are as follows: according to the scheme, the Fourier transform is carried out on the voice feedback sub-frames of the user to obtain the signal frequency spectrum information of each frame, and the initial and final sound recognition is carried out according to the signal frequency spectrum information of each frame, so that the accuracy of the voice feedback recognition can be improved, the functions required by the user expression can be quickly and accurately found, the error rate of the provided service functions is reduced, and the good experience of the user is improved.
In one embodiment, in step S210, after framing the voice feedback according to the time length, and before performing the recognition analysis, preprocessing each frame of the voice feedback, where the preprocessing process is as follows:
filtering each frame of voice by adopting a filter with 32 channels;
solving the spectral energy of each frame of voice by adopting the following formula:
in the above equation, Q (i, j) represents the spectral energy of the jth spectral coefficient vector in the ith frame; u denotes the number of channels of the filter, i.e., u is 32; envM(k, i) an envelope spectrum representing the ith frame of speech for the kth channel;
and extracting a 12-dimensional spectral coefficient vector of each frame of voice according to the calculation result, and taking the obtained spectral coefficient vector as a feature vector of voice recognition.
The working principle and the beneficial effects of the technical scheme are as follows: according to the scheme, each frame of voice feedback is preprocessed before recognition and analysis, the preprocessing comprises filtering processing by adopting a 32-channel filter, noise or noise interference can be eliminated, the formula is adopted to solve the spectrum energy of each frame of voice, the spectrum coefficient vector is extracted through the spectrum energy and is used as the feature vector of each frame of recognition in the voice feedback, the accuracy of the feature vector is improved, a good foundation is provided for follow-up recognition, and the recognition accuracy is improved.
In one embodiment, in step S100, the user preference value of the function is calculated using the following formula:
in the above formula, τiA user preference value indicating an ith function; k is a radical of1All users representing the ith function use the weight; m represents the number of users; n represents the number of functions in the answering system; w is aijRepresenting the number of times that the jth user uses the ith function; k is a radical of2Representing the current incoming call user use weight of the ith function; w is aiRepresenting the number of times that the current incoming call user uses the ith function;
and according to the user preference values of the functions obtained by calculation, sorting the preference of the functions according to the user preference values from big to small to obtain a function list.
The working principle and the beneficial effects of the technical scheme are as follows: the scheme provides another optional user preference sorting mode, the user preference values of all functions are calculated through the formula, the function utilization rates of all users and the function utilization rate of the incoming call user are introduced, weight distribution is carried out on the function utilization rates of all users and the function utilization rate of the incoming call user, the function utilization rates of all users and the function utilization rate of the incoming call user are summed according to the weight to serve as the user preference values of the corresponding functions, and the function preference sorting is carried out on the calculated user preference values from large to small to obtain a function list; by adopting the formula, the scheme can calculate and complete the function use preference sequencing of the user at one time, and provides efficient function service.
The invention is further illustrated by the following application example.
After the user dials in, as shown in fig. 4, the answering system proceeds as follows:
1. and identifying the attribution according to the incoming call number of the user, if the attribution can be correctly identified by the response system, prompting by using the dialect of the attribution, and otherwise, prompting by using standard Mandarin.
2. Judging whether the number calls once according to the incoming call number of the user, and if the incoming call record exists and the function is used, adding the previously used function into a function list with the level of [ 3 ].
3. And acquiring the most frequently used function of the current home from the response system, and adding the most frequently used function into a function list with the level of [ 2 ].
4. And acquiring the most frequently used functions of all users from the response system, and adding the most frequently used functions into a function list with the level of [ 1 ].
5. And sequencing according to the levels of the previously formed function lists, judging the most possible functions used by the user by the response system, feeding the function lists back to the user in a voice prompt mode, and waiting for the feedback of the user.
6. After the voice feedback of the user is acquired, the voice feedback of the user is uploaded to a response system and stored as an audio file, for example: wav format for subsequent identification.
7. The speech feedback is divided into frames, each frame occupies a relatively short fixed time (for example, 25ms), and if such a frame is long enough (enough information can be contained to judge which initial and final sound belongs to) and stable (short-time Fourier analysis is convenient), each frame is converted into a feature vector, and (in turn) the initial and final sound belonging to each frame is respectively identified.
8. After the voice characteristics of the initial consonants and the vowels after the framing are available, matching fixed phrases in the response system according to the initial consonants and the vowels, wherein the association relationship between the fixed phrases and the functions is established in the response system in advance, and after the corresponding association functions are found out successfully through the matched fixed phrases, the response system automatically jumps to the associated functions to provide services for users.
The method has the advantages that the response system gives corresponding prompts to the user through the method, and based on the voice input of the client, the functions required by the client are quickly located and found and entered, so that the waiting time of the client is reduced.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A method for connecting a call center based on a voice input, comprising the steps of:
s100, according to the telephone number identified by the incoming call, carrying out preference sorting of the functions used by the user, and providing the corresponding function list sequence to the user by the preference sorting;
s200, acquiring voice feedback of a user, performing recognition analysis, and matching the voice feedback with a fixed phrase in a response system;
s300, acquiring the function associated with the matched fixed phrase in the response system, and automatically jumping to the associated function.
2. The method of claim 1, wherein in the step S100, the home location is identified according to the incoming phone number, and if the answering system identifies the home location of the phone number, the voice answering system prompts the caller using a home dialect, otherwise, the caller using a standard mandarin.
3. The method of claim 2, wherein the performing a preference ranking process of the user usage function in the step S100 comprises:
s110, judging whether a previous call is received or not through a telephone number according to the call, if the answer system has a call record of the telephone number and the call record has a corresponding use function, adding the corresponding use function into a function list, and marking the level to be 3;
s120, all incoming calls and corresponding use functions of the telephone number attribution of the current incoming call are obtained from the answering system, the function with the most use frequency is determined, and the function list is added, wherein the marking level is 2;
s130, determining the function with the most frequent use of all user calls in the answering system, and adding the function into a function list with the marking level of 1;
s140 performs preference sorting of functions according to the mark level based on the function list obtained through the steps of S110, S120, and S130.
4. The method of claim 3, wherein if the number of previous incoming calls of the current caller is large and the number of functions used by the caller is two or more, the functions are sorted according to the number of times of use, and the number of times of use is arranged in front; if a plurality of functions with the same use times exist, sorting is carried out according to the time from the near to the far away from the current incoming call.
5. The method for connecting a call center according to a voice input of claim 1, wherein the voice feedback of the user is subjected to recognition analysis in the step S200 as follows:
s210, framing the voice feedback according to the time length to ensure that the time length of each frame can contain enough information for performing initial consonant and vowel judgment;
s220, converting each frame into a feature vector, respectively identifying by adopting short-time Fourier analysis, and sequentially determining the initial consonants and the vowels to which the frames belong;
and S230, matching with the fixed phrases in the response system according to the initial and final voice characteristics of the sub-frames.
6. The method of claim 5, wherein if the matching with the fixed phrase in the answering system fails, a prompt is sent to the user, and the sequential list of the corresponding functions is fed back to the user again in the order of preference, requiring the user to perform voice feedback again.
7. The method of claim 1, wherein the voice feedback of the user is stored as an audio file in the S200 step.
8. The method of claim 5, wherein in step S220, the short-time Fourier analysis comprises analyzing each frame of the voice signal using the following Fourier transform function:
in the above formula, the first and second carbon atoms are,
representing signal spectrum information extracted from the speech signal of the nth frame; n represents the framing of the speech feedback; x (n) represents the nth frame sampling speech source signal of the speech feedback; w (nt-m) represents a window function; t represents a frame shift duration; e represents a natural constant; j denotes a virtual unit, i.e. j2-1; k represents angular frequency, k is more than or equal to 0 and less than or equal to N-1; n represents the frame length;
and then, performing initial consonant and vowel identification on the frequency spectrum information of each frame of signal obtained by Fourier transform.
9. The method of claim 5, wherein in step S210, after the voice feedback is time-sliced, each frame of the voice feedback is preprocessed before the recognition analysis, and the preprocessing comprises:
filtering each frame of voice by adopting a filter with 32 channels;
solving the spectral energy of each frame of voice by adopting the following formula:
in the above equation, Q (i, j) represents the spectral energy of the jth spectral coefficient vector in the ith frame; u denotes the number of channels of the filter, i.e., u is 32; envM(k, i) an envelope spectrum representing the ith frame of speech for the kth channel;
and extracting a 12-dimensional spectral coefficient vector of each frame of voice according to the calculation result, and taking the obtained spectral coefficient vector as a feature vector of voice recognition.
10. The method of claim 1, wherein in the step of S100, the user preference value of the function is calculated using the following formula:
in the above formula, τiA user preference value indicating an ith function; k is a radical of1All users representing the ith function use the weight; m represents the number of users; n represents the number of functions in the answering system; w is aijRepresenting the number of times that the jth user uses the ith function; k is a radical of2Representing the current incoming call user use weight of the ith function; w is aiRepresenting the number of times that the current incoming call user uses the ith function;
and according to the user preference values of the functions obtained by calculation, sorting the preference of the functions according to the user preference values from big to small to obtain a function list.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110062248.0A CN112788184A (en) | 2021-01-18 | 2021-01-18 | Method for connecting call center according to voice input |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110062248.0A CN112788184A (en) | 2021-01-18 | 2021-01-18 | Method for connecting call center according to voice input |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112788184A true CN112788184A (en) | 2021-05-11 |
Family
ID=75756351
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110062248.0A Pending CN112788184A (en) | 2021-01-18 | 2021-01-18 | Method for connecting call center according to voice input |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112788184A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113645363A (en) * | 2021-08-11 | 2021-11-12 | 广东电网有限责任公司广州供电局 | Multi-service-scene-oriented internal call sharing service platform for power enterprises |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6798876B1 (en) * | 1998-12-29 | 2004-09-28 | At&T Corp. | Method and apparatus for intelligent routing of incoming calls to representatives in a call center |
| CN1659624A (en) * | 2002-04-05 | 2005-08-24 | 英特尔公司 | Dynamic and adaptive selection of vocabulary and acoustic models based on a call context for speech recognition |
| CN105744057A (en) * | 2016-01-21 | 2016-07-06 | 平安科技（深圳）有限公司 | Intelligent voice dialogue interaction method and intelligent voice dialogue interaction device |
| CN106598955A (en) * | 2015-10-20 | 2017-04-26 | 阿里巴巴集团控股有限公司 | Voice translating method and device |
| CN107800895A (en) * | 2016-08-31 | 2018-03-13 | 南京中兴软件有限责任公司 | A kind of interactive voice answering method and device |
| CN107995376A (en) * | 2017-11-07 | 2018-05-04 | 平安科技（深圳）有限公司 | A kind of user reports a case to the security authorities processing method and terminal device |
| CN109089001A (en) * | 2018-09-14 | 2018-12-25 | 广东尚通科技发展有限公司 | A kind of intelligent customer service method of calling and its calling system |
| CN109272129A (en) * | 2018-09-20 | 2019-01-25 | 重庆先特服务外包产业有限公司 | Call center's business management system |
| CN110110881A (en) * | 2019-03-21 | 2019-08-09 | 贵州电网有限责任公司 | Power customer requirement forecasting analysis method and system |
| CN111432081A (en) * | 2020-03-31 | 2020-07-17 | 中国银行股份有限公司 | Recommendation method and device for interactive voice response menu |
-
2021
- 2021-01-18 CN CN202110062248.0A patent/CN112788184A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6798876B1 (en) * | 1998-12-29 | 2004-09-28 | At&T Corp. | Method and apparatus for intelligent routing of incoming calls to representatives in a call center |
| CN1659624A (en) * | 2002-04-05 | 2005-08-24 | 英特尔公司 | Dynamic and adaptive selection of vocabulary and acoustic models based on a call context for speech recognition |
| CN106598955A (en) * | 2015-10-20 | 2017-04-26 | 阿里巴巴集团控股有限公司 | Voice translating method and device |
| CN105744057A (en) * | 2016-01-21 | 2016-07-06 | 平安科技（深圳）有限公司 | Intelligent voice dialogue interaction method and intelligent voice dialogue interaction device |
| CN107800895A (en) * | 2016-08-31 | 2018-03-13 | 南京中兴软件有限责任公司 | A kind of interactive voice answering method and device |
| CN107995376A (en) * | 2017-11-07 | 2018-05-04 | 平安科技（深圳）有限公司 | A kind of user reports a case to the security authorities processing method and terminal device |
| CN109089001A (en) * | 2018-09-14 | 2018-12-25 | 广东尚通科技发展有限公司 | A kind of intelligent customer service method of calling and its calling system |
| CN109272129A (en) * | 2018-09-20 | 2019-01-25 | 重庆先特服务外包产业有限公司 | Call center's business management system |
| CN110110881A (en) * | 2019-03-21 | 2019-08-09 | 贵州电网有限责任公司 | Power customer requirement forecasting analysis method and system |
| CN111432081A (en) * | 2020-03-31 | 2020-07-17 | 中国银行股份有限公司 | Recommendation method and device for interactive voice response menu |
Non-Patent Citations (1)
| Title |
|---|
| 张志霞: "语音识别中个人特征参数提取研究", 《中国优秀硕士学位论文全文数据库 (信息科技辑)》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113645363A (en) * | 2021-08-11 | 2021-11-12 | 广东电网有限责任公司广州供电局 | Multi-service-scene-oriented internal call sharing service platform for power enterprises |
| CN113645363B (en) * | 2021-08-11 | 2022-05-17 | 广东电网有限责任公司广州供电局 | Multi-service-scene-oriented internal call sharing service platform device for power enterprises |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6996525B2 (en) | Selecting one of multiple speech recognizers in a system based on performance predections resulting from experience | |
| US7450698B2 (en) | System and method of utilizing a hybrid semantic model for speech recognition | |
| EP2122610B1 (en) | Customizable method and system for emotional recognition | |
| US7966176B2 (en) | System and method for independently recognizing and selecting actions and objects in a speech recognition system | |
| US20030191639A1 (en) | Dynamic and adaptive selection of vocabulary and acoustic models based on a call context for speech recognition | |
| US8144838B2 (en) | Automated task classification system | |
| US6601029B1 (en) | Voice processing apparatus | |
| US7366668B1 (en) | Voice interface for a search engine | |
| EP0797823B1 (en) | Automated call router system and method | |
| US7398212B2 (en) | System and method for quality of service management with a call handling system | |
| US8824659B2 (en) | System and method for speech-enabled call routing | |
| US7206389B1 (en) | Method and apparatus for generating a speech-recognition-based call-routing system | |
| CN110059182A (en) | Art recommended method and device towards customer service | |
| US20060133587A1 (en) | System and method for speech recognition-enabled automatic call routing | |
| CN112131359A (en) | Intention identification method based on graphical arrangement intelligent strategy and electronic equipment | |
| US20040042591A1 (en) | Method and system for the processing of voice information | |
| CN112788184A (en) | Method for connecting call center according to voice input | |
| CN111986651A (en) | Man-machine interaction method and device and intelligent interaction terminal | |
| CN115017289A (en) | Method and device for serving customers based on knowledge base | |
| CN112087726B (en) | Method and system for identifying polyphonic ringtone, electronic equipment and storage medium | |
| CN107800895A (en) | A kind of interactive voice answering method and device | |
| JP7304627B2 (en) | Answering machine judgment device, method and program | |
| US6845356B1 (en) | Processing dual tone multi-frequency signals for use with a natural language understanding system | |
| CN114067842B (en) | Customer satisfaction degree identification method and device, storage medium and electronic equipment | |
| Natarajan et al. | Speech-enabled natural language call routing: BBN Call Director |
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 |
Application publication date: 20210511 |
|
| RJ01 | Rejection of invention patent application after publication |