Speech recognition error analysis apparatus, method, program, and recording medium therefor

摘要

P To specify a part being likely to cause error in recognition in a language model. PSOLUTION: In a method, voice recognition processings are performed for a voice signal by using the language model and recognized word strings being the results of voice recognition are assigned. The erroneously recognized word strings constituted by one or a plurality of continuous words which do not agree with the correct word strings corresponding to the recognized word strings in the recognized word strings is extracted out of the recognized word strings. All the sets of two erroneous words in a section being the sets of two continuous words in the erroneously recognized word strings are extracted out of the erroneously recognized word strings. A set of two words returned to correct words constituted by the first word of the set of two erroneous words in the section and the word in the correct word strings having a START being later in time than a START of the first word and being the closest to a terminal of the first word in time is extracted per set of two erroneous words in each section. A set of two erroneous words in the section having higher word chain probability than word chain probability of the set of two words returned to correct words is extracted. PCOPYRIGHT: (C)2009 and JPO& INPIT