US 7,321,852 B2
System and method for transcribing audio files of various languages
Joey Stanford, Longmont, Colo. (US)
Assigned to International Business Machines Corporation, Armonk, N.Y. (US)
Filed on Oct. 28, 2003, as Appl. No. 10/696,113.
Prior Publication US 2005/0091274 A1, Apr. 28, 2005
Int. Cl. G10L 15/00 (2006.01); G10L 21/00 (2006.01)
U.S. Cl. 704—235  [704/231; 704/277] 5 Claims
OG exemplary drawing
 
1. A method for transcribing an audio file included in or referenced by a web page, said method comprising the steps of:
determining a language of text in said web page, wherein the determining step comprises:
(i) determining whether language identifying metadata is contained within said web page and if so, said language is determined to be as indicated by said language identifying metadata,
(ii) if said language identifier metadata is not contained within said web page, determining a domain extension of said web page, and an official language of said domain extension, wherein said language is determined to be said official language of said domain extension,
responsive to determining said language of said text in said web page, selecting a voice recognition software program of said language of said text from a plurality of voice recognition software programs that are each operable for transcribing a different language,
using said selected voice recognition software to transcribe said audio file,
responsive to using said selected voice recognition software program to transcribe said audio file, determining whether said audio file was successfully transcribed by said selected voice recognition software program, and if not, selecting another voice recognition software program from said plurality of voice recognition software programs and using said selected another voice recognition software program to transcribe said audio file, wherein said another voice recognition software program is selected based upon a country of origin for said domain extension as determined by using a domain name registrar service to lookup said country of origin using said domain extension,
responsive to using said selected another voice recognition software program to transcribe said audio file, determining whether said audio file was successfully transcribed by said selected another voice recognition software program, and if not, selecting an alternative voice recognition software program from said plurality of voice recognition software programs and using said selected alternative voice recognition software program to transcribe said audio file, wherein said alternative voice recognition software program is selected based upon comparing said text to a multi-lingual data base to identify words of said text which match words in said multi-lingual data base,
responsive to using said selected alternative voice recognition software program to transcribe said audio file, determining whether said audio file was successfully transcribed by said selected alternative voice recognition software program, and if not, selecting another alternative voice recognition software program from said plurality of voice recognition software programs and using said selected another alternative voice recognition software program to transcribe said audio file, wherein said another alternative voice recognition software program is selected based upon a related language that is related to said language, and
responsive to using said selected another alternative voice recognition software program to transcribe said audio file, determining whether said audio file was successfully transcribed by said selected another alternative voice recognition software program, and if not, continuing to (i) successively select another remaining voice recognition software program from said plurality of voice recognition software programs and (ii) use said selected another remaining voice recognition software program to transcribe said audio file until said audio file is successfully transcribed.