(12) United States Patent ao) Patent No.: us 6,347,296 Bi
Friedland (45) Date of Patent: Feb. 12,2002
(54) CORRECTING SPEECH RECOGNITION WITHOUT FIRST PRESENTING ALTERNATIVES
(75) Inventor: Steven J. Friedland, Boca Raton, FL (US)
(73) Assignee: International Business Machines Corp., Armonk, NY (US)
( * ) Notice: Subject to any disclaimer, the term ol this patent is extended or adjusted under 35 U.S.C. 154(b) by 0 days.
(21) Appl. No.: 09/338,699
(22) Filed: Jun. 23, 1999
(51) Int. CI.7 G10L 15/22; G10L 15/26;
G10L 21/00
(52) U.S. CI 704/231; 704/270; 704/251;
704/252
(58) Field of Search 704/270, 272,
704/278, 231, 251, 252
(56) References Cited
U.S. PATENT DOCUMENTS
4,797,855 A * 1/1989 Duncan, IV et al 400/63
A method and a system within a computer speech recognition system to automatically correct portions ol text. The invention receives text derived Irom a first user input for inclusion in a body ol text. Concurrently upon receipt ol the first user input, and based upon the first user input, the system identifies a list ol alternate text selections potentially intended by the user. Next, the system stores each ol the alternate text selections in a memory location associated with the text. In response to a second user input, and without displaying the list ol alternate text selections to the user, the system automatically retrieves a first one ol the alternate text selections Irom the memory location and inserts the first one ol the alternate text selections in place ol the text in the body ol text.
19 Claims, 5 Drawing Sheets
