Telephonic communications – Audio message storage – retrieval – or synthesis – Voice activation or recognition
Reexamination Certificate
1998-05-04
2001-11-06
Tsang, Fan (Department: 2645)
Telephonic communications
Audio message storage, retrieval, or synthesis
Voice activation or recognition
C379S088010, C379S088040
Reexamination Certificate
active
06314166
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method presented in the preamble of the appended claim
1
for dialling a telephone number by voice commands. Further, the invention relates to a telecommunication terminal presented in the preamble of the appended claim
5
.
2. Description of the Prior Art
A telephone number is usually dialled by using the selecting disc or dialling keys of a telecommunication terminal. However, there may be situations when it would be necessary to dial a telephone number e.g. by speaking the telephone number. Particularly for use in automobiles, so-called hands-free modes have been developed, wherein the driver of the vehicle does not need to loose hold of the steering wheel for dialling a telephone number. The dialling of the telephone number is thus conducted by speaking the telephone number to be dialled. This kind of a hands-free mode is practical also e.g. in offices. In the simplest way, this kind of dialling of a telephone number by voice commands works in a way that the user of the telecommunication terminal says the telephone number to be dialled digit by digit, wherein after saying the whole number, a speech recognizer unit in the telecommunication terminal searches the number sequence that the speech recognizer unit interprets to best correspond to the number sequence recited by the user. The interpretation is based e.g. on the fact that the speech recognizer unit calculates one or several feature vectors on the basis of the audio signal received. The speech pattern memory of the voice recognizer unit contains e.g. speech patterns corresponding to numeral digits, and these are used in combination with the feature vectors calculated from the audio signal for calculating the recognition result by using methods known as such.
Telephone numbers are usually very long, commonly number sequences containing at least seven numeral digits, wherein the number of various combinations is very large: with a sequence of seven numerals, there are 10,000,000 alternative combinations. When making calls to mobile stations, to telecommunication terminals of different directory areas, or from one country to another, the length of a telephone number can be as long as 15 numeral digits. The large number of alternative combinations sets high demands on the operation of a device to be controlled by voice commands, so that the number of incorrect recognitions could be reduced to the minimum. In the above-mentioned situation of dialling a telephone number consisting of seven numeral digits, incorrect recognition of even one numeral digit will lead to incorrect dialling.
U.S. Pat. No. 4,870,686 discloses a telecommunication terminal controlled by speech recognition, wherein the telephone number to be dialled can be uttered in one or several digit strings. Each digit string is recognized separately, wherein the number of alternatives for each digit string is considerably smaller than in a situation when the whole telephone number is recognized as one digit string. For example, the telephone phone number “1234567” can thus be uttered e.g. in digit strings “12”, “34”, “567”, wherein the number of different dialling alternatives is one hundred for the first and second digit strings and one thousand for the third digit string. Thus, the probability that each digit string will be recognized correctly is considerably higher than if the whole telephone number were recognized as one number sequence. However, also this dialling method has the disadvantage that if any of the digit strings is recognized incorrectly the first time, in which case the user will control the speech recognizer unit to recognize this numeral sequence again, the speech recognizer unit can make the same incorrect interpretation also the next time. In the worst case, the user must repeat the whole number sequence, and even this will not guarantee that the speech recognizer unit can recognize the uttered telephone number correctly. This unreliability of recognition is due to a number of various factors. For example, the recognition of the telephone number can be interfered by noise conditions. Furthermore, many recognizer units are advantageously programmed at the manufacturing stage so that an average speech pattern for each number from zero to nine is stored in the speech pattern memory of the recognizer unit. However, different users will pronounce the numbers in slightly different ways, which will not necessarily always result in equally good recognition for different users, wherein the error rate can be different when different persons use such a telecommunication terminal controlled by voice commands. In these situations, it is possible to use recognizer units which can be taught to recognize the user's voice, i.e. the user pronounces the numbers from zero to nine, wherein the speech recognizer unit stores the speech patterns corresponding to the numbers in the speech pattern memory. Nevertheless, this will not eliminate all incorrect recognitions, e.g. under the influence of noise or the user's voice which is changed for any reason.
Further, such voice-controlled telecommunication terminals have been developed in which the user can store the telephone numbers desired and an identification corresponding to these, such as the name of a firm and/or a person. Thus the selection of the telephone number can be made by uttering the identification, on the basis of which the speech recognizer unit compares the identifications stored and conducts dialling on the basis of this comparison. In such a device, where the identification can be divided into sub-identifications, the recognizer unit conducts a comparison of sub-identifications and after correct recognition of the sub-identification, the user utters the next sub-identification. When a sufficient number of sub-identifications have been uttered to identify the telephone number, the telecommunication terminal conducts dialling of the telephone number. Also in this kind of a telecommunication terminal, the problem may occur that the identification or sub-identification is continually recognized incorrectly and the correct telephone number cannot be dialled.
OBJECTS OF THE INVENTION
One purpose of the present invention is to eliminate the disadvantages mentioned above to a great extent and to provide dialling of a telephone phone number by voice commands as accurately as possible. The invention is based on the idea that when the user utters a telephone number, a part of the telephone number or an identification, after an incorrect recognition this recognition is marked incorrect by the recognizer unit, wherein when the user repeats said number sequence or identification, the recognizer unit will no longer offer the recognized incorrect alternative but the alternative which according to the calculation by the recognizer unit is next probable and which is not marked incorrect. Thus, the number of alternative combinations is reduced after each incorrect recognition, wherein the correct number sequence or identification is worked out at the latest when there is only one alternative left. The method according to the present invention is characterized in what will be presented in the characterizing part of the appended claim
1
. Further, the telecommunication terminal according to the present invention is characterized in what will be presented in the characterizing part of the appended claim
5
.
The present invention gives significant advantages. Using the method of the invention, the operation of speech-controlled telecommunication terminals can be improved to a significant extent e.g. for the reason that the recognizer unit will not offer the same alternative again after an incorrect recognition. Also in disturbance conditions, the telecommunication terminal of the invention is more reliable than speech-controlled telecommunication terminals of prior art. Moreover, the method of the invention improves the operation of such telecommunication terminals in which the speech recognizer unit is not a so-call
Laurila Kari
Mettälä Markku
Escalante Ovidio
Nokia Mobile Phones Limited
Perman & Green LLP
Tsang Fan
LandOfFree
Method for dialling a telephone number by voice commands and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for dialling a telephone number by voice commands and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for dialling a telephone number by voice commands and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2617387