JPH045758A - Computer systems using portable authentication - Google Patents

Abstract

PURPOSE:To improve safety by executing the checking of a user with using an identification code read from a portable certification and executing a fixed procedure based on the voice of the user. CONSTITUTION:When an IC card 1 is inserted into a slot 2, the slot 2 detects it, reads an identification code from the card 1 by a reading part 3 and makes it sent to a user identifying part 4. The identifying part 4 urges a user to input a keyword. When the user inputs the keyword, the identifying part 4 retrieves a data base and checks whether the keyword registered in a system and the keyword key-inputted by the user are coincident or not. Reference data are read, a reference data preparing part 6 prepares the reference data base 8 proper to the owner of the card 1 and urges the user to input by voice. The voice of the user is converted into a feature vector by a voice feature extracting part 10 and is converted into a phoneme code by a voice encoding part 11 respectively. A word column substituted from a phoneme column by a language analyzing part 12 is inputted to an executing part and a procedure is executed.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a computer system that allows voice interaction with a computer system and that allows users to be identified and security problems to be resolved. .

[Prior Art] In a computer system that uses conventional authentication, a user name and a secret keyword known only to the user are entered from the keyboard, and if the pair matches a pre-registered user name and keyword pair, the computer is authenticated. They were given permission to use the system.

[Problems and objectives to be solved by the invention] In recent years, computer systems have been used in ways that are more similar to normal conversations, rather than operating keys according to procedures determined by the computer system. The need for this is increasing. Voice input is an interface that meets this purpose.

To put speech input into practical use, a computer system used by many users must be able to recognize the voices of many users, but speech recognition for unspecified speakers is difficult to recognize for specific speakers. Recognition accuracy is inferior in comparison. Comparing the average recognition error rate, speech recognition systems for unspecified speakers are 3 to 5 times more expensive than those for specific speakers.
It will be about double. In particular, in the case of a speaker whose speech characteristics deviate greatly from the average, it is difficult to obtain recognition accuracy that can be put to practical use.

An object of the present invention is to enable reliable authentication and identification of users, and at the same time, to enable human voice recognition with high recognition accuracy.

[Means for solving the problem] A computer system using portable authentication of the present invention has the following features:
It is characterized by using portable authentication that stores reference data and identification code for recognizing the voice of a specific individual,
a data reading unit that reads data from the portable authentication; a user identification unit that uses the read identification code to check whether the user is a legitimate user; and a user identification unit that uses the reference data to identify the user's voice. It is characterized in that it includes as its constituent elements a speech recognition section that performs recognition, and an execution section that executes a predetermined procedure in response to the result of the speech recognition section.

[Embodiment] FIG. 1 is an embodiment of a computer system using portable authentication of the present invention. 1 is an IC card, and this I
An identification code unique to the C card is recorded. Also IC
The card 1 stores reference data for recognizing the phonemes uttered by a specific individual who is the owner of this card. The reference data is obtained by modeling phonemes using a hidden Markov model and adding parameters for a specific individual to this model. Parameters adapted to unspecified speakers are used as reference values, and differences from the reference values are stored in the IC card. This reduces the amount of reference data stored in the IC card.

When the IC card 1 is inserted into the slot 2, the slot detects this and starts the data reading section 3. The data reading section 3 reads the identification code from the IC card l and sends it to the user identification section 4. The user identification section prompts the user to input a keyword. When the user inputs a keyword from the keyboard 5, the user identification unit 4 searches the database and finds a match between the keyword registered in the system paired with this identification code and the keyword entered manually by the user. Check whether

If they match, use is permitted. If they do not match, use will not be permitted.

If the user's keyword is valid, the data reading unit reads the reference data described above.

The reference data creation unit 6 adds the read reference data for specific individuals and the generalized database 7 that has been generalized to support unspecified speakers for each corresponding parameter, and creates a reference unique to the owner of the IC card. Create database 8. At this point, the user is notified that the service is available, and voice assistance is encouraged.

The voice input by the user is collected by the microphone 9 and converted into a time series of feature vectors by the voice feature extractor 10. Each feature vector is compared with the reference database 8 and converted into a phoneme code by the speech encoding unit 11.

The language analysis unit 12 refers to the dictionary 13 and replaces the phoneme string with a word string. Finally, the execution unit 14 executes a predetermined procedure according to the input word string.

[Effects of the Invention] According to the configuration of the present invention, possession of authentication is required in order to use a computer system, and security is improved compared to the conventional case where verification is performed only by keywords. Furthermore, the voice input section of the computer system can be adapted to the individual user, and high recognition accuracy can be obtained.

All of the data stored in the portable authentication device is unique to a specific individual, and keeping it inseparable provides convenience to the user.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the configuration of one embodiment of the present invention. Below Up

Claims (1)

[Claims] Using a portable authentication device that stores reference data for recognizing the voice of a specific individual and an identification code, (a) a data reading unit that reads data from the portable authentication device; (b) a user identification unit that checks whether the user is a legitimate user using the read identification code; (c) a speech recognition unit that recognizes the user's voice using the reference data; (d) A computer system using portable authentication, comprising as a component an execution unit that executes a predetermined procedure in response to the result of the voice recognition unit.