VOICE RECOGNITION DEVICE, SEARCH DEVICE, VOICE RECOGNITION METHOD, SEARCH METHOD AND PROGRAM

• Main Authors: KASAHARA KANAME, SUGIZAKI MASAYUKI, AMAKASU TETSUO, HIKICHI TAKAFUMI
• Format: Patent
• Language: English; Japanese
• Published: 20.08.2020
• Subjects: ACOUSTICS; MUSICAL INSTRUMENTS; PHYSICS; SPEECH ANALYSIS OR SYNTHESIS; SPEECH OR AUDIO CODING OR DECODING; SPEECH OR VOICE PROCESSING; SPEECH RECOGNITION

To acquire highly accurate voice recognition results for conversation message while suppressing an increase in an amount of calculation.SOLUTION: The voice recognition device 10 according to the present invention includes: a first voice recognition unit 11 which performs voice recognition processing by a first method on voice data of conversation of a plurality of speakers, and outputs a voice recognition result for each utterance section of each of the plurality of speakers; a determination unit 13 for determining a message section, based on the result of the voice recognition processing by the first voice recognition unit 11; and a second voice recognition unit 14 which performs voice recognition processing by a second method which is more accurate than the first method, on the voice data of the section determined to be the message section, and outputs the voice recognition result as a message text.SELECTED DRAWING: Figure 1 【課題】計算量の増大を抑制しつつ、会話の用件について高精度な音声認識結果を取得する。【解決手段】本発明に係る音声認識装置１０は、複数の話者による会話の音声データに対して、第１の方式により音声認識処理を行い、複数の話者それぞれの発話区間ごとの音声認識結果を出力する第１の音声認識部１１と、第１の音声認識部１１による音声認識処理の結果に基づき、用件区間を判定する判定部１３と、用件区間であると判定された区間の音声データに対して、第１の方式よりも高精度の第２の方式により音声認識処理を行い、音声認識結果を用件テキストとして出力する第２の音声認識部１４と、を備える。【選択図】図１