1. Speech recognition overview
Speech Recognition, Automatic Speech Recognition (ASR), is a technology that allows machines to understand human language. Language is the most important, longest and most direct way for human beings to exchange information. Speech recognition technology is a major breakthrough in the realization of man-machine dialogue. It has developed rapidly in foreign countries in recent years, and its application has gradually been promoted. The IVR (Automatic Telephone Answering), which has become popular in recent years, has handled a lot of simple and repetitive consultations, saving a lot of manpower, but this push-button voice auto-answer allows customers to spend a lot of time to follow the guidelines. The query is inconvenient for the user.
Speech recognition can undoubtedly solve this problem. The successful development of the speech recognition system has fully utilized the advantages of computer technology and network technology. Using advanced human-machine dialogue methods to get rid of the shackles of telephone buttons, people simply say the required service items to the phone as usual. Easily access the information you need from an automated system.
   Speech recognition system structure
2. Speech recognition application
Nuance is a leader in natural voice interface software. With natural voice interface software, people can easily and securely access information, services and transactions over the phone. Every day, thousands of people make travel reservations, stock transactions, interactions with other communication media, businesses and the Internet system by dialing a phone that runs Nuance's speech recognition, language understanding and voiceprint authentication software. NUANCE applications: American Airlines, Bell Atlantic, Charles Schwab, Home Shopping Network, Lloyds TSB, Sears, UPS.
3.NUANCE speech recognition features
(1) Massive vocabulary, independent of the speaker's robust recognition function
The Nuance system reliably identifies large vocabularies in multiple languages ​​and provides confidence in the recognition results. The system provides the most accurate speech recognition technology for a large number of vocabularies used commercially. Applications developed with the Nuance system have the highest accuracy 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 architecture designed specifically for the robustness and scalability required by large applications. The caller's speech is collected by the client, and the load of identification and authentication processing is evenly distributed across multiple separate servers on the network.
(3) N-Best processing
For some applications, it may be necessary to identify the engine to produce a possible set of recognition results, rather than a best result. This feature is available in the Nuance system's N-best recognition processing method, which provides a list of possible recognition results and ranks them from high to low.
(4) Grammatical probability
The Nuance system allows for the specification of the probability of a particular word or phrase spoken by the caller in the grammar. Useful when the probability of a spoken word or phrase can be estimated based on actual usage. 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 identification server more accurately identified. The recognition server enhances the incoming utterance to effectively filter noise such as tone, squeak, squeak, and snoring. This mechanism works well if a significant number of phones contain stable background noise, such as hands-free calling on a car.
4. Recognition-based applications
Voice SMS itself business, company phone book, personal phone book, smart songs, stock inquiry and transaction, intelligent information on demand, train time inquiry
(1) Corporate phone book based on speech recognition
Company phone book application description
The system can support multiple access numbers and virtualize the switchboards of multiple companies. And rely on voice recognition technology, intelligent transfer to the corresponding users.
Process
The user dials the access code, the system finds the corresponding company database according to the access code, and reminds the user to speak the corresponding user. The system queries the database according to the corresponding user, and obtains the number of the user, and notifies the switch to connect the number.
Company phone book features
â—The system supports telephone access
Users can modify personal passwords, personal work calls and non-work calls by phone
â—The system supports WEB access mode
â— System administrator can modify all information
â— Each company administrator can add, delete, and modify the company's telephone information.
(2) Voice recognition based on voice recognition
Functional description
The user accesses the system, prompts the user to name the song, 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 name the song or the name of the artist, and relies on the voice recognition technology to find the name of the song or the name of the artist. If the name is queried for the database, the path of the song is obtained and played to the user. If it is a singer name, the singer's album is read out, and the user is prompted to select, after the user selects, the song name of the album is read, and the user is prompted to select, and after the user selects, the user is played the song.
Solar Panel,Portable Big Power Bank,Lithium Batteries Power Station,Portable Battery Powered Generator
Guangzhou Fengjiu New Energy Technology Co.,Ltd , https://www.flashfishbatteries.com