-
公开(公告)号：JP2000242293A
公开(公告)日：2000-09-08
申请号：JP2000036105
申请日：2000-02-15
Applicant: MOTOROLA INC
Inventor： MEUNIER JEFFREY ARTHUR , POPPERT DANIEL CHARLES
Abstract: PROBLEM TO BE SOLVED: To make storable trace-back information needed for HMM tracing by using a smaller memory by dividing the signal representation of a pronounced voice into frames, allocating the respective frames to respective states by using adjusting algorithm, and discriminating state transition to respective states. SOLUTION: A sound signal received from a microphone 114 is converted by an analog-digital converter 202 into a digital signal. A speech call processor inplements a feature extracting function 204 for the processed digital signal representation of the analog signal output of the microphone 114 to generate one group of feature vectors representing the pronounced voice of the user. And, one feature vector is present for every frame. The processor performs speech recognition 206 or training 207 by making use of the feature vectors. In training mode, the feature vector of the pronounced voice is used to generate a template in HMM format to be stored in a memory 208.
-
公开(公告)号：BR0000894A
公开(公告)日：2000-09-12
申请号：BR0000894
申请日：2000-02-21
Applicant: MOTOROLA INC
Inventor： MEUNIER JEFFREY ARTHUR , POPPERT DANIEL CHARLES
-
Patent Agency Ranking
公开(公告)号：DE10006937C2
公开(公告)日：2003-02-13
申请号：DE10006937
申请日：2000-02-16
Applicant: MOTOROLA INC
Inventor： MEUNIER JEFFREY ARTHUR , POPPERT DANIEL CHARLES
-
公开(公告)号：GB2347254B
公开(公告)日：2001-08-22
申请号：GB0004281
申请日：2000-02-23
Applicant: MOTOROLA INC
Inventor： MEUNIER JEFFREY ARTHUR , POPPERT DANIEL CHARLES
-
公开(公告)号：DE10006937A1
公开(公告)日：2000-10-12
申请号：DE10006937
申请日：2000-02-16
Applicant: MOTOROLA INC
Inventor： MEUNIER JEFFREY ARTHUR , POPPERT DANIEL CHARLES
Abstract: The voice recognition device works on the basis of receiving a speech portion (104). A signal is then generated to represent the speech. This signal is divided (108) into frames and each of these frames is allocated to an alignment algorithm. A path is then stored to represent the frame to state allocations by storing in memory (110,122) for each state, a separate type to identify the transition. An independent claim is included for the method of voice recognition.
-
公开(公告)号：GB2347254A
公开(公告)日：2000-08-30
申请号：GB0004281
申请日：2000-02-23
Applicant: MOTOROLA INC
Inventor： MEUNIER JEFFREY ARTHUR , POPPERT DANIEL CHARLES
Abstract: A speech utterance is divided into frames representative of the utterance. Frames are allocated to states (S1-S5) using an alignment algorithm. A path representing frame to state allocations is stored in memory using state transition types identifying a state transition to each state. A 1 may be stored for a self-loop, and A 0 for a state change.
-
公开(公告)号：GB2333877B
公开(公告)日：2001-08-08
申请号：GB9900679
申请日：1999-01-14
Applicant: MOTOROLA INC
Inventor： SRENGER EDWARD , MEUNIER JEFFREY ARTHUR , KUSHNER WILLIAM M
Abstract: The present invention provides a method of calculating, within the framework of a speaker dependent system, a standard filler, or garbage model, for the detection of out-of-vocabulary utterances. In particular, the method receives new training data in a speech recognition system (202); calculates statistical parameters for the new training data (204); calculates global statistical parameters based upon the statistical parameters for the new training data (206); and updates a garbage model based upon the global statistical parameters (208). This is carried out on-line while the user is enrolling the vocabulary. The garbage model described in this disclosure is preferably an average speaker model, representative of all the speech data enrolled by the user to date. Also, the garbage model is preferably obtained as a by-product of the vocabulary enrollment procedure and is similar in it characteristics and topology to all the other regular vocabulary HMMs.
-
-
-
-
-
-
Search Results
7
records
(took 0.021 seconds)
