PROBLEM TO BE SOLVED: To solve the problem that biosignal detection accuracy is not sufficiently high in an identification apparatus using a biosignal such as an utterance supporting device.SOLUTION: A person A to be supported wears a biosignal sensor 1 detecting a biosignal. A waveform reshaping section 2 outputs an analysis signal in which noise removal is made for the signal to a dividing section 13. An acquiring section 5 is composed of the biosignal sensor 1 and the waveform reshaping section 2. The dividing section 13 receives the analysis signal and carries out feature sampling and outputs signals passing through the feature sampling and signals not passing through the feature sampling to a selecting and identifying section. The selecting and identifying section receives the signals passing through the feature sampling and the signals not passing through the feature sampling and outputs the result of identification to an output section 6 by selectively using signals effective for identification or by greatly weighting the effective signals from among the signals and using them for identification. The output section carries out output according to the result of identification by an output device such as a speaker, a display section, a network connector (computer for example), or wireless communication equipment.COPYRIGHT: (C)2013,JPO&INPIT【課題】発話支援装置などの生体信号を用いた識別装置において、生体信号の検出精度が十分に高くないという課題がある。【解決手段】被支援者Aは、生体信号を検出する生体信号センサ部1を装着している。波形整形部2は、その信号に対して、雑音除去処理などを行った解析信号を分割加工部13に出力する。取得部5は、生体信号センサ1と波形整形部2で構成される。分割加工部13は、解析信号を受けて特徴抽出を行い、特徴抽出を経た信号と特徴抽出を経ない信号を選択識別部に出力する。選択識別部は特徴抽出を経た信号と特徴抽出を経ない信号を受けて、これらの信号から、識別のために有効な信号を選択的に用いて、もしくは有効な信号に高い重み付け(ウェイト)をかけて識別に用い、識別結果を出力部6に出力する。出力部は、スピーカー、表示部、ネットワーク接続器(例えば、コンピューター)、無線通信器などの出力装置で、識別結果に応じた出力をする。【選択図】 図1
