1. Voice recognition overview
Speech recognition technology, AutomaTIc Speech RecogniTIon, or ASR for short, is a technology that allows machines to understand human language. Language is the most important, longest and most direct way for humans to communicate information. Speech recognition technology is a major breakthrough in realizing human-machine dialogue. It has developed rapidly abroad in recent years, and its application has been gradually promoted. IVR (Automatic Telephone Answering), which has become popular in recent years, has handled many simple and repeated consultation work, saving a lot of manpower, but this button-type voice automatic answering allows customers to spend a lot of time following the guidelines to complete the simple Query, which makes users feel inconvenient.
Speech recognition can undoubtedly solve this problem. The successful development of the speech recognition system fully utilizes the advantages of computer technology and network technology, adopts advanced man-machine dialogue methods, and gets rid of the shackles of the phone buttons. As long as people simply speak the required service items to the phone as usual, namely The required information provided by the automatic system can be easily obtained.
Speech recognition system structure
2. Voice recognition application
Nuance is a leader in natural voice interface software. Using natural voice interface software, people can easily and securely obtain information, services and conduct transactions through the phone. Every day, tens of millions of people call Nuance ’s voice recognition, language understanding, and voiceprint identification software to make travel reservations, stock transactions, and interact with other communication media, businesses, and Internet systems. NUANCE applications: American Airlines, Bell AtlanTIc, Charles Schwab, home shopping network, Lloyds TSB, Sears, UPS.
3. Features of NUANCE voice recognition
(1) Massive vocabulary, robust recognition function independent of speaker
The Nuance system can reliably recognize large vocabulary in multiple languages ​​and provide confidence in the recognition results. The system provides the most accurate speech recognition technology for a large number of words used commercially. The application developed by Nuance system has the highest accuracy rate in the market. The application in production has been tested with an accuracy of over 96%.
(2) Host-based client / server structure
The Nuance system is based on an open client / server machine structure, specifically designed for the robustness and scalability required by large applications. The caller's speech is collected by the client, and the load of the identification and authentication process is evenly distributed to multiple separate servers on the network.
(3) N-Best processing
For some applications, the recognition engine may need to generate a possible recognition result set, rather than a best result. The N-best recognition processing method of the Nuance system has this function. It provides a list of possible recognition results and ranks them from high to low.
(4) Grammatical probability
The Nuance system allows the grammar probabilities of specific words or phrases spoken by the caller to be specified. It is very useful when the probability of the spoken word or phrase can be estimated based on actual use. Increasing the probability of grammar can improve the accuracy and speed of recognition.
(5) Reduce noise
When the incoming call contains stable background noise, the Nuance system uses a mechanism to make the recognition server recognize more accurately. The recognition server enhances the incoming utterances to effectively filter noises such as tone, hum, hum, and boo. If a considerable number of phones contain stable background noise, such as hands-free phone calls in a car, this mechanism works well.
4. Identification-based applications
Voice SMS itself business, company phone book, personal phone book, smart song, stock query and transaction, smart information on demand, train time query
(1) Company phone book based on voice recognition
Company Phone Book Application Description
The system can support multiple access numbers and virtual switchboards of multiple companies. And rely on voice recognition technology to intelligently transfer to the corresponding user.
Process
When the user dials the access code, the system finds the corresponding company database based on the access code, and at the same time reminds the user to speak the corresponding user. The system queries the database based on the corresponding user, obtains the user's number, and notifies the switch to connect the number.
Company Phone Book Features
• The system supports telephone access
Users can modify their personal passwords by phone, personal work calls and non-work calls
• The system supports WEB access
• System administrator can modify all information
• The administrators of each company can add, delete and modify the company's telephone information
(2) The song selection function based on voice recognition
Functional description
The user accesses the system, prompts the user to say the song name, and relies on voice recognition technology to find the song name and play it to the user.
Application process
The user dials the access code, prompts the user to say the song name or artist name, and relies on voice recognition technology to find the song name or artist name, if the song name queries the database, get the path of the song stored, and play to the user, If it is the name of the singer, the album of the singer is read out, and the user is reminded to choose. After the user selects, the title of the album is read out, and the user is reminded of the choice. After the user chooses, the song is played to the user.
With Perkins Engine Generator,Container Type Diesel,Soundproof Generator,Super Silent Type Diesel
Shanghai Kosta Electric Co., Ltd. , https://www.shkostagenerator.com