VOICE PROCESSOR, MUSICAL SOUND EXTRACTION METHOD FOR VOICE PROCESSOR, AND PROGRAM

• Main Author: FURUKAWA YOSHIHISA
• Format: Patent
• Language: English; Japanese
• Published: 08.10.2015
• Subjects: ACOUSTICS; MUSICAL INSTRUMENTS; PHYSICS; SPEECH ANALYSIS OR SYNTHESIS; SPEECH OR AUDIO CODING OR DECODING; SPEECH OR VOICE PROCESSING; SPEECH RECOGNITION

PROBLEM TO BE SOLVED: To provide a voice processor, a musical sound extraction method of the voice processor and a program capable of preventing the deterioration of the volume and quality of an extracted sound.SOLUTION: A musical sound data extraction part 17 includes: an information acquisition part 31 for acquiring information indicating frequency transition on the time axis of an object musical sound; a frequency width determination part 32 for determining frequency width to be extracted for each frame in accordance with a frequency variation amount between the adjacent frames of the object musical sound; and a musical sound extraction part 33 for extracting the object musical sound in accordance with the frequency width determined by the frequency width determination part 32. The frequency width determination part 32 is characterized to, when the frequency variation amount is equal to or more than a predetermined amount, widen the frequency width to be extracted in comparison with the case that the frequency variation amount is less than the predetermined amount. 【課題】抽出音の音量低下および音質劣化を防止可能な音声処理装置、音声処理装置の楽音抽出方法およびプログラムを提供することを課題とする。【解決手段】本発明の楽音データ抽出部１７は、対象楽音の時間軸上における周波数推移を示す情報を取得する情報取得部３１と、対象楽音の隣接フレーム間における周波数変動量に応じて、抽出する周波数幅をフレームごとに決定する周波数幅決定部３２と、周波数幅決定部３２により決定された周波数幅にしたがって、対象楽音を抽出する楽音抽出部３３と、を備え、周波数幅決定部３２は、周波数変動量が所定量以上の場合、周波数変動量が所定量未満の場合と比較して、抽出する周波数幅を広くすることを特徴とする。【選択図】図４