JP4465870B2 - Audio signal processing device - Google Patents

Description

つまり、上記したように伝達関数Ｈ（ａ〜ｐ）によって、第一閉曲面１２内の聴取者にとっては第一閉曲面の外側には測定環境１の音場が再現されていると考えられるので、図６に示す第二閉曲面１４の測定用マイク１３（Ａ〜Ｅ）は、再現環境１１においては仮想音像とされる測定用スピーカ３から測定用音声が出力された状況と近似的に同じになる。
したがって、第二閉曲面１４の内側において、伝達係数ＴＡ、ＴＢ、ＴＣ、ＴＤ、ＴＥは、測定環境１において測定用スピーカ３から測定用音声を出力した場合の伝達関数と等価になる。
【００２６】
ところで、例えば測定環境１におけるＮ箇所の残響音場特性を、Ｍ種類の再生用スピーカの配置状態に対応させようとした場合、測定用スピーカ３をＮ箇所に配置して、再生用スピーカの配置状態に対応させてＮ回の測定を行う必要がある。すなわち、Ｎ×Ｍ回の測定を行う必要があった。しかし本発明では、再生用スピーカの配置状態に対応した測定は再現環境１１で行うので、測定環境１においてはＮ回の測定を行えばよい。
また、再生用スピーカの配置状態の種類が増えた場合、この配置状態に対応させてＮ箇所の測定環境１を再現させようとした場合、Ｎ箇所の測定環境１において測定を行う必要があった。したがって、測定環境１において測定を行う位置の数（Ｎ箇所）が多い場合は困難な作業となる、しかし、本発明では、Ｎカ所の測定環境１において伝達関数を求めておけば、再生用スピーカの配置状態の種類が増える毎に、再生環境１１において測定を行えばよい。したがって、新規の再生用スピーカの配置状態に対応させる場合でも、Ｎ箇所の測定環境に出かけていく必要がなく、効率のよい測定作業を行うことができるようになる。
【００２７】
次に、再現環境１１で得られた伝達関数ＴＡ、ＴＢ、ＴＣ、ＴＤ、ＴＥに基づいて、例えば家庭の部屋などとされる再生環境において、例えば５チャンネルステレオの音声を再生する例を説明する。
まず、図７にしたがい、再生環境における５チャンネルステレオ音声の再生系について説明する。
再生装置１５において、音声再生部１６は例えばディスクなどの記録媒体に記録されているコンテンツにおいて、音声信号を再生することができるようにされている。この音声再生部１６から出力された音声信号Ｓは、演算部１７ａ、１７ｂ、１７ｃ、１７ｄ、１７ｅに供給される。演算部１７（ａ〜ｅ）は、供給された音声信号Ｓに対して所要の演算処理を行って、再現環境１１で求められた伝達関数ＴＡ、ＴＢ、ＴＣ、ＴＤ、ＴＥに基づく演算処理を行う。これにより、演算部１７（ａ〜ｅ）からは、音声信号Ｓに対して伝達関数に基づいて演算処理を施された再現信号ＳＴＡ、ＳＴＢ、ＳＴＣ、ＳＴＤ、ＳＴＥが出力される。そして、各再生信号ＳＴ（Ａ〜Ｅ）は、再生用スピーカ１８Ａ、１８Ｂ、１８Ｃ、１８Ｄ、１８Ｅに供給される。したがって、各再生用スピーカ１８（Ａ〜Ｅ）からは、伝達関数Ｔ（Ａ〜Ｅ）に基づいた再生信号ＳＴＡ、ＳＴＢ、ＳＴＣ、ＳＴＤ、ＳＴＥによる音声が出力される。
【００２８】
再生環境における再生用スピーカ１８（Ａ〜Ｅ）は、図８に示されているように第二閉曲面１９上における所要の位置に配置される。なお、この第一閉曲面１９と再現環境１１における第一閉曲面１２は、同等の閉曲面とされるが、便宜上異なる符号を付している。
このように、第二閉曲面１９上にその内側に向けて再生用スピーカ１８（Ａ〜Ｅ）を配置して、再生信号ＳＴＡ、ＳＴＢ、ＳＴＣ、ＳＴＤ、ＳＴＥによる音声を出力することで、第二閉曲面１９の内側に居る聴取者にとっては、破線で示す第一閉曲面１２上に配された再現用スピーカ８（ａ〜ｐ）による音場と同等な音場を再現することができる。したがって、第二閉曲面１９内を聴取位置とした場合、図９に示されているように、測定環境１内における残響音場、及び音像定位を得ることができるようになる。これにより、例えば家庭の部屋に居ながら、例えばホールなどの残響音場及び音像定位により、コンテンツの音声を聴取することができるようになる。
また、本発明は測定環境、再現環境で実際に測定したインパルス応答に基づいて伝達関数を求めているので、例えば４点近接法などのように解析結果を参照する方法よりも高品位の音声により、残響、音像定位を再現することが可能である。
【００２９】
なお、再生環境における再生用スピーカの数は再生システムの構成によって異なる場合があり、さらに配置位置に関しても家庭事情などによって同一にすることは容易ではない。このような場合、従来では、測定環境において再生用スピーカの配置状況に応じた測定を行う必要があったが、本実施の形態では測定環境での伝達関数をマスターデータとして、その後は無響室などで再生環境に応じた測定を行うことができる。つまり、測定を行うためのホールの使用料などを削減することができると共に、容易に再生環境に対応した伝達関数を得ることができるようになる。
また、特徴のある残響音場とされる測定環境が、例えば遠隔地にあった場合などでも、上記したようにマスターデータを得てしまえば、その測定環境に行く必要は無くなるので、このような点でも測定環境における測定を効率よく行うことが可能になる。
【００３０】
さらに、本実施の形態では、再生装置１５の構成として音声信号処理装置を示したが、例えば家庭などで使用されるＡＶ（Audio Visual）アンプ、サラウンドプロセッサ、あるいはＣＤやＤＶＤプレーヤなどに適用することができる。
【００３１】
ところで、前記式１に示されているように、行列ＧとベクトルＨｖは分離することが可能となる。したがって、再現環境１１においては例えば図４、図５で説明したような測定方法をとらなくても伝達関数Ｔ（Ａ〜Ｅ）を求めることができる。
例えば、ホールなどの残響特性データとして、測定環境においてベクトルＨｖに相当するデータ（伝達関数Ｈを求めるためのインパルス応答）を測定する。一方、再生環境における再生用スピーカの配置状態に対応するデータとして、例えば図１０に示されているように第一閉曲面１２において、測定用マイク１３（Ａ〜Ｅ）までの伝達関数Ｇを、測定信号ＴＳＰに基づいて別途測定しておく。そして、個別に求められたベクトルＨｖと行列Ｇによって、式１により伝達関数Ｔ（Ａ〜Ｅ）を求めることが可能となる。
したがって、ベクトルＨｖと行列Ｇを音声信号処理装置において個別に管理して必要に応じて行列演算を行うことができるようにすることで、伝達関数の測定回数を削減することができる。さらに、残響特性と再生用スピーカの配置を独立して選択することもできるようになる。
【００３２】
なお、本実施の形態では、測定環境１において測定用スピーカ３を再生環境における仮想スピーカであるとして説明したが、再生環境で再生されるソースが、例えば人間が知覚しやすい自然音である場合や、例えば１種類の音から構成される音である場合、または測定用スピーカ３の特性がフラットに近い特性である場合などでは、前記ソース自体が仮想音像として認識される。但し、前記ソース自体の仮想音像を正確に表現するためには、スピーカの逆特性を含めて補正することがより望ましい。
【００３３】
図１１は、上記したように伝達関数を求め、再生環境において測定環境の音場を再現する場合の手順を説明するフローチャートを示している。
まず、測定環境１の第一閉曲面１０において測定用スピーカ３から測定用音声を出力してインパルス応答を測定し（Ｓ００１）、測定したインパルス応答に基づいて伝達関数Ｈを求める（Ｓ００２）。次に、再現環境１１の第一閉曲面１２において伝達関数Ｈに基づく演算処理を施された音声信号によって再現用音声を出力して（Ｓ００３）、第二閉曲面１４において再現用音声のインパルス応答を測定し（Ｓ００４）、さらに第二閉曲面１４で測定したインパルス応答に基づいて伝達関数Ｇを求める（Ｓ００５）。そして、再生環境において音声の再生を行う場合に、再生された音声再生信号に対して伝達関数Ｇに基づく演算処理を行って出力する（Ｓ００６）。
これにより、再生環境では図９に示したように測定環境と同等の残響効果、音像定位を再現することができるようになる。
【００３４】
また、伝達関数Ｈと伝達関数Ｇを個別に管理して行列演算を行う場合の手順は、図１２のフローチャートに示されているようになる。
図１１に示したフローチャートと同様に、測定環境１の第一閉曲面１０において測定用スピーカ３から測定用音声を出力してインパルス応答を測定し（Ｓ１０１）、測定したインパルス応答に基づいて伝達関数Ｈを求める（Ｓ１０２）。そして、再現環境１１に測定用スピーカ３を配置して、第一閉曲面１２において、測定用スピーカ３から測定用音声を出力してインパルス応答を測定し（Ｓ１０３）、さらに第二閉曲面１４で測定したインパルス応答に基づいて伝達関数Ｇを求める（Ｓ１０４）。
そして、再生環境において音声の再生を行う場合に、再生された音声再生信号に対して伝達関数Ｈに基づく演算処理を行い（Ｓ１０５）、さらに、伝達関数Ｈに基づいて演算処理された音声再生信号に伝達関数Ｇに基づく演算処理を施す（Ｓ１０６）。
このような手順により、伝達関数Ｈと伝達関数Ｇを個別に管理して必要に応じて行列演算を行うことで、残響特性と再生用スピーカの配置を独立して選択することができる。
【００３５】
２．複数の仮想スピーカの再現
次に、例えばディスクリートな音声信号からなる５チャンネルステレオを実現することができる再生システムに対応した伝達関数を生成する例を説明する。
図１３は、５チャンネルステレオの再生環境を説明する図である。
この図に示す再生環境は、例えば、映画館２１などの広い空間とされ、ディスクリートな５チャンネルを実現するために実線で示す５個のスピーカ２０Ａ、２０Ｂ、２０Ｃ、２０Ｄ、２０Ｅが配置される。各スピーカ２０（Ａ〜Ｅ）に対して、センターフロント信号ＤＡ、フロントライト信号ＤＢ、リアライト信号ＤＣ、リアレフト信号ＤＤ、フロントレフト信号ＤＥが供給されて、各信号に対応した音声が再生される。つまり、この図に示す再生環境としての映画館２１では、広い空間の残響効果を得ることができる。
【００３６】
しかし、例えば家庭の部屋などの狭い空間では、図１３に示した映画館２１と同じ間隔で各スピーカ２０（Ａ〜Ｅ）を配置することが困難なので、再生用スピーカ１８（Ａ〜Ｅ）は破線で示されているように配置されることが考えられる。したがって、第二閉曲面１９における再生用スピーカ１８（Ａ〜Ｅ）によって映画館２１のような、音像定位、残響効果が得られる音場を構築するには、前記各信号Ｄ（Ａ〜Ｅ）に伝達関数Ｔ（Ａ〜Ｅ）に基づいて演算処理を施すようにすればよい。
【００３７】
図１４は、前記各信号Ｄ（Ａ〜Ｅ）に伝達関数Ｔ（Ａ〜Ｅ）に基づく演算処理を行う音声信号処理装置の構成例を説明する図である。
図１４に示されている音声信号処理装置２２は、音声再生部２３から再生音声信号として、センターフロント信号ＤＡ、フロントライト信号ＤＢ、リアライト信号ＤＣ、リアレフト信号ＤＤ、フロントレフト信号ＤＥを出力することができるようにされている。
音声再生部２３から出力された各信号Ｄ（Ａ〜Ｅ）は、それぞれ演算部２４（ａ〜ｅ）、２６（ａ〜ｅ）、２８（ａ〜ｅ）、３０（ａ〜ｅ）、３２（ａ〜ｅ）に供給される。例えば演算部２４（ａ〜ｅ）では、センターフロント信号ＤＡに対して伝達係数ＴＡＡ、フロントライト信号ＤＢに対して伝達係数ＴＢＡ、リアライト信号ＤＣに対して伝達係数ＴＣＡ、リアレフト信号ＤＤに対して伝達係数ＴＤＡ、フロントレフト信号ＤＥに対して伝達係数ＴＥＡに基づく演算処理を行う。そして、演算処理が施された各信号を加算部２５、２７、２９、３１、３３で加算して再生信号ＶＡを生成する。
【００３８】
ところで、図１４に示す演算部２４，２６、２８、３０、３２において用いられる伝達関数ＴＡ（Ａ〜Ｅ）、ＴＢ（Ａ〜Ｅ）、ＴＣ（Ａ〜Ｅ）、ＴＤ（Ａ〜Ｅ）、ＴＥ（Ａ〜Ｅ）は、先に図６で説明した方法によって求めることができる。
すなわち図１３に示すスピーカ２０Ａを、図６に示す再現環境における仮想的な音像とされる測定用スピーカ３に対応させることができる。つまり、図６に示す伝達関数ＴＡ、ＴＢ、ＴＣは、図１４に示す伝達関数ＴＡＡ、ＴＡＢ、ＴＡＣ・・・に対応したものとなる。
この場合、伝達関数ＴＡ（Ａ〜Ｅ）、ＴＢ（Ａ〜Ｅ）、ＴＣ（Ａ〜Ｅ）、ＴＤ（Ａ〜Ｅ）、ＴＥ（Ａ〜Ｅ）は、式２に示されているように表すことができる。
【数２】

【００３９】
さらに、図６に示す仮想的な測定用スピーカ３を、図１３に示すスピーカ２０Ｂ、スピーカ２０Ｃ、２０Ｄ、２０Ｅに対応させることで、各スピーカ２０（Ｂ〜Ｅ）から第二閉曲面１９の測定用マイク１３（Ａ〜Ｅ）間での伝達関数ＴＢ（Ａ〜Ｅ）、ＴＣ（Ａ〜Ｅ）、ＴＤ（Ａ〜Ｅ）、ＴＥ（Ａ〜Ｅ）を求めればよい。ここで、図１５に示されているように測定環境としての映画館２１においてスピーカ２０Ａから測定用マイク４（ａ〜ｐ）までの伝達関数をＨａ（１〜１６）、スピーカＢから測定用マイク４（ａ〜ｐ）までの伝達関数をＨｂ（１〜１６）、スピーカ２０Ｃから測定用マイク４（ａ〜ｐ）までの伝達関数をＨｃ（１〜１６）、スピーカ２０Ｄから測定用マイク４（ａ〜ｐ）までの伝達関数をＨｄ（１〜１６）、スピーカ２０Ｅから測定用マイク４（ａ〜ｐ）までの伝達関数をＨｅ（１〜１６）として、これらの伝達関数を行列Ｈとすると、式２に示した行列Ｔは式３に示されているように表すことができる。
【数３】

【００４０】
この式３に示されているようにして伝達関数Ｈを求め、さらに例えば図４、図５で説明したようにして伝達関数Ｇを求めることで、最終的に伝達関数ＴＡ（Ａ〜Ｅ）、ＴＢ（Ａ〜Ｅ）、ＴＣ（Ａ〜Ｅ）、ＴＤ（Ａ〜Ｅ）、ＴＥ（Ａ〜Ｅ）が求められる。
そして、図１４に示した音声信号処理装置２２によって、伝達関数ＴＡ（Ａ〜Ｅ）、ＴＢ（Ａ〜Ｅ）、ＴＣ（Ａ〜Ｅ）、ＴＤ（Ａ〜Ｅ）、ＴＥ（Ａ〜Ｅ）に基づく演算処理を施された音声信号Ｖ（Ａ〜Ｅ）を、図１６に示されている再生環境において再生用スピーカ１８（Ａ〜Ｅ）に供給して、音声を出力することで、再生用スピーカ１８（Ａ〜Ｅ）からなる第二閉曲面１９内には、仮想スピーカとしてのスピーカ２０（Ａ〜Ｅ）を想定した音場が形成される。つまり、例えば映画館などのような広い空間の残響、及び音像定位を、再生環境とされる部屋などで再現することができる。聴取者は第二閉曲面１９内であれば、いずれの位置で聴取しても、元の測定環境における音場と同等な音場感を得ることができ、より豊かな臨場感に浸ることができる。
なお、ディスクリートな音声信号に音声再生システムとして、例えばＤＶＤ−Ｖｉｄｅｏ（Digital Versatile Disc Video）等に適用されている５．１チャンネルシステムが知られている。このような再生システムに本発明を適用すると、聴取者が居る再生環境よりも広い部屋、響きの良い部屋を再現した鑑賞を実現することができるようになる。これにより、再生用スピーカの存在を意識させるような不自然な音声を自然な音声に緩和させることが可能になる。
【００４１】
３．円形対称性を利用する例
ところで、上記実施の形態では、測定環境において例えば図１６に示した複数の仮想スピーカ（例えばスピーカ２０（Ａ〜Ｅ））全てを想定した位置に、測定用スピーカを配置させて伝達関数の測定を行う例を挙げた。したがって、仮想スピーカの数に応じて伝達関数の測定を行う必要があり、作業効率がよいものではない場合がある。
【００４２】
例えば、仮想スピーカの位置、再現用スピーカの位置、及び第一閉曲面の中心の３つの相対的幾何学位置が共通している場合、既知の伝達関数を用いることができるようになる。
例えば図１７に示されているように、仮想スピーカ３５Ａ、３５Ｂは、上記した３つの位置関係が共通している。つまり、仮想スピーカ３５Ａに対する再現用スピーカ８ｐ、８ａ、８ｂの配置関係と、仮想スピーカ３５Ｂに対する再現用スピーカ８ｂ、８ｃ、８ｄの配置関係は同じである。したがって、測定環境において、反射音などに方向性が強くない場合には、これらの伝達関数はそれぞれほぼ同じものとなる。これにより、上記式３に示した行列Ｈに関して、２列目をベクトルとみて、
【数４】

とすることができ、既知の伝達係数の行列を転置させて使用することが可能となる。
【００４３】
なお、図１７においては、仮想スピーカ３５Ｃは、仮想スピーカ３５Ａ、３５Ｂと上記した３つの位置関係が共通していない。このような状況が生じることを想定して、測定環境１においては第一閉曲面１０において配置される測定用マイク４の数を増やし、第一閉曲面１２における再現用スピーカ８の数を増やすことで、位置関係の対応をとることができるようになる。もちろん、仮想スピーカ３５Ｃに対して、隣接配置される再現用スピーカの配置関係から補間処理などの手法で対応することも可能であるが、処理の簡便さにより上記の配置とすることが好ましい。
【００４４】
このような方法を、先に図１５に示した例に対応させることができる。つまり、例えばスピーカ２０Ａからの出力に対応した伝達関数を求め、この伝達関数を、例えばスピーカ２０Ｂを想定した伝達関数として用いることが可能となる。
さらに、再現環境１１において再生用スピーカ８の配置位置も第一閉曲面１２における再現用スピーカ８の位置、再生用スピーカ１８の位置、及び第一閉曲面１２の中心の３つの相対的幾何学位置が共通していれば、行列Ｇの特定行の値をベクトル単位で適宜転置させることで、音場を近似させることができるようになる。したがって、再現環境１１における測定回数を削減することができるようになる。
また、この場合も、行列Ｇと行列Ｈを個別に管理して必要に応じて行列演算を行うことができるようにすることで、残響特性と再生用スピーカを独立して選択することもできる。そして、行列Ｇ、行列Ｈに関して、位置の対称性を利用して、複数の伝達関数を転置させて使用することで、例えば音声信号処理装置において伝達関数を管理する記憶手段の容量を小さくすることができるという利点がある。
【００４５】
また、図１８に示されているように、測定環境（例えばホールなど）１の壁からの反射情報も仮想スピーカと同じように回転することになる。したがって、測定用マイク４（ａ〜ｐ）が配置される第一閉曲面１０は、壁の影響が大きくなく、障害物などが近くに無い、測定環境１の中心部分に近い位置に設定して測定を行うことが望ましい。なお、この例においては、入力としての複数チャンネルからなる音声信号、及びその音声信号に対応した複数のスピーカを対象としているので、伝達関数を共通して使用することで、音像が狭くなる、パンニングの効果になる等とされる本来の残響効果が薄れてしまうことはない。
【００４６】
４．仮想音源の移動において円形対称性を利用する例
図１９は、第二閉曲面１９内において、聴取者から距離Ｒ＿ｈｍ離れた場所を回転しながら移動する仮想的な音像を作成する例を示す模式図である。
図１９において第二閉曲面１９は、例えば８個の再生用スピーカ３６Ａ、３６Ｂ、３６Ｃ、３６Ｄ、３６Ｅ、３６Ｆ、３６Ｇ、３６Ｈによって形成され、これらの再生用スピーカ３６（Ａ〜Ｈ）の内側が聴取位置となる。
このように配置されている再生用スピーカ３６（Ａ〜Ｈ）に供給する音声信号に対して、畳み込む伝達関数を変更することにより、例えば足音などの音像を回転移動させることができる。
この図に示す例では、仮想音像を回転させる位置として、例えば第一閉曲面１０と同心円となる半径Ｒ＿ｈｍとされる円周を例えば３２分割した、仮想位置ＶＨｍ１、ＶＨｍ２、ＶＨｍ３・・・までの各位置とする。つまり、測定環境では、これらの仮想位置ＶＨｍ（１〜３２）に対応した位置に測定用マイク４を配置して、伝達関数Ｈの測定を行うものとなる。そして、この３２箇所に対応した伝達係数Ｈを切り替えて用いることで、再生用スピーカ３６（Ａ〜Ｈ）から出力される音声によって音像が回転する音場を再現する。
なお、例えば仮想位置ＶＨｍ１のように、末尾に奇数の符号が付されている位置は測定環境１における測定用マイク４、及び再現環境１１における再現用スピーカ８と同一半径の位置とし、例えば仮想位置ＶＨｍ２のように、末尾に偶数の符号が付されている位置は測定用マイク４、再現用スピーカ８の間に対応した半径位置とされている。
【００４７】
再生用スピーカ３６（Ａ〜Ｈ）に供給する音声信号は、例えば図２０に示されている音声信号処理装置４０によって生成される。
音声再生部４１によって再生された音声信号ＳＶは、測定スピーカ逆特性処理部４２を介して、演算部４３に供給される。演算部４３（ａ〜ｈ）は、供給された音声信号ＳＶに対して、伝達関数ＴＡ〜ＴＨに基づく所要の演算処理を行い、再生信号ＳＶａ、ＳＶｂ、ＳＶｃ・・・、ＳＶｈを出力する。そして、出力された再生信号ＳＶ（ａ〜ｈ）は、再生用スピーカ３６（Ａ〜Ｈ）に供給される。
【００４８】
演算部４３において用いられる伝達関数Ｔ（Ａ〜Ｈ）は、測定環境１における伝達関数Ｈ、及び再現環境１１における伝達関数Ｇに基づいて求められる。但し、伝達関数Ｈについては、図１７で説明した例と同様にして第一閉曲面１０の円の対称性を利用することができる。すなわち、測定スピーカ３を、例えば仮想位置ＶＨｍ１、ＶＨｍ２に対応させて２回の測定を行い、この２箇所の測定によって得られた伝達関数Ｈを転置して代用することが可能となる。これにより、式３に示した行列Ｈを求めることができる。
また、伝達関数Ｇについては、図示しない再現環境において、再生用スピーカ３６（Ａ〜Ｈ）に対応した位置に測定用マイクを配置して、仮想位置ＶＨｍ（１〜３２）に対応した位置に配した再現用スピーカから出力される再現信号に基づいて求めればよい。
【００４９】
演算部４３ではこのようにして求められた伝達関数Ｔ（Ａ〜Ｈ）に基づいて再生信号ＳＶに演算処理を行うが、音像位置コントローラ４４では、仮想位置ＶＨｍ１、ＶＨｍ２・・・を想定して、伝達関数Ｔ（Ａ〜Ｈ）を制御することで、例えば図１９に示したように第二閉曲面１９内において音像の仮想位置を回転させることができるようになる。
【００５０】
なお、以上説明してきた実施の形態においては、第一閉曲面１０が円形となる例を挙げたが、必ずしも円形である必要はない。すなわち、閉曲面の外周上で複数の測定用マイク４をその外側に向けて測定を行うことができるようにされていればよい。但し、対称性を利用して測定回数を減らす場合などは、第一閉曲面は例えば円形または正多角形であることが必要となる。
【００５１】
また、第一閉曲面１０に配置される測定用マイク４の数は、本実施の形態で説明した数に限定されるものではない。したがって、測定環境１における測定データ（伝達関数）のデータ量を増やして、第一閉曲面１０の内側の音圧分布をできるだけ正確なものにする場合は、できるだけ多くの測定用マイク４を配置すればよい。
【００５２】
さらに、第一閉曲面１０の面積（本実施の形態では半径）については、例えば第一閉曲面１０、１２の外周が、第二閉曲面１４、１９の外側となるようにされていればよい。
【００５３】
また、測定用マイク４（ａ〜ｐ）、測定用マイク１３（Ａ〜Ｅ）は、例えば指向性マイクとして説明したが、指向性マイクとはしない場合でも、例えば複数の無指向性マイクを用いてそれらの出力に対して遅延あるいは減算処理などをして相応の効果を得ることができればよい。
【００５４】
また、行列Ｇ２に関しては、再現環境１１における測定を行わずに、幾何学的な構図に基づいて振幅や遅延特性などを用いるようにしてもよい。
また、第一閉曲面１０を本実施の形態で説明したように例えば円形で構成した場合、第一閉曲面１０における或る軸に関して対称となる一方側の伝達関数を他方側に転用することも可能である。例えば図１に示されているように、測定用マイク４ｂに対する伝達関数をＨｂとした場合、この伝達関数Ｈｂを測定用マイク４ａを軸として対称的な位置に配されている測定用マイク４ｐと共通化することが可能になる。
同様に第二閉曲面１４においても、例えば図４に示されているように再現用スピーカ８ｂから測定用マイク１３Ａに対する伝達関数Ｇ２Ａとした場合、この伝達関数Ｇ２Ａを再現用スピーカ８ｐから測定用マイク１３Ａと共通化することができる。
【００５５】
また、上記実施の形態では第一閉曲面、第二閉曲面ともに例えば平面上を想定した例を挙げたが、立体的な閉曲面上に測定用マイクを配置することで、立体的に音場の再現を行うことも可能である。
【００５６】
さらに、インパルス応答の測定時、また伝達関数を求める場合において、第一閉曲面内、第二閉曲面内の音圧分布近似を、対象となる音場における仮想スピーカを音源とした音圧分布により近づけるために、上記式１に示した行列に対して、さらに正方行列で示される補償処理を施すようにしてもよい。この補償処理は、音圧分布の近似を行うために十分ではない数のスピーカ、マイクロホンを使う場合や、スピーカ、マイクロホンの指向性などを補償するための処理となる。
【００５７】
５．カーオーディオシステム
次に、本発明をカーオーディオシステムに適用する場合の例を説明する。
この例では、図２１に示されているように自動車５０の室内に配置される再生用スピーカ５１（ａ〜ｊ）によって、例えば自動車５０の前方に配置されていることを想定したスピーカ５２Ｌ、５２Ｒ（後述する測定用スピーカに相当する）から、ステレオ音声が再生されている環境を室内にいる聴取者に対して与えることができるようにすることを目的としている。
このため、図２２に示されている測定環境５３においては自動車５０の例えば外観を囲むことができるような立体（本実施の形態では直方体）を想定し、この立体を第一閉曲面５５として、その外周上に例えば２１個の測定用マイク５６ａ〜５６ｕを配置している。測定用マイク５６（ａ〜ｕ）の指向性の主軸は、矢印で示されているように第一閉曲面５５の中心から、当該第一閉曲面５５を形成する面における法線方向外向き発散する延長方向に対応させる。なお、測定用マイク５６（ａ〜ｕ）の配置位置が前記面上ではなく、２以上の面に接する線、または点であれば接している前記面上の各法線方向のベクトル和として、指向性の主軸の向きを定めることとする。
【００５８】
測定用スピーカ５２Ｌ、５２Ｒからは第一閉曲面５５に対して自動車５０の前方を想定した位置に配置され、測定用スピーカ５２Ｌからは左音声用の測定音声、また測定用スピーカ５２Ｒからは右音声用の測定音声を出力することができるようにされている。そして、測定用スピーカ５２Ｌ、５２Ｒから出力される測定音声によって測定用マイク５６（ａ〜ｕ）までの伝達関数Ｈを求めて保持しておく。
すなわち、測定用スピーカ５２Ｌから測定用マイク５６（ａ〜ｕ）までの伝達関数はＨＬａ〜ＨＬｕ、そして、測定用スピーカ５２Ｒから測定用マイク５６（ａ〜ｕ）までの伝達関数はＨＲａ〜ＨＲｕとなる。
【００５９】
図２３は、例えば無響室などの再現環境６０において測定環境５３を再現する例を説明する模式図である。
図示されている再現環境６０に形成される第一閉曲面６１は、測定環境５３における第一閉曲面５５に対応しており、同等の立体を想定したものとされている。再現用スピーカ６２ａ〜６２ｕは第一閉曲面６１上に配置されるが、その配置位置は第一閉曲面５５における測定用マイク５６（ａ〜ｕ）に対応した位置とされる。なお、この図では再現用スピーカ６２ａ〜６２ｕに関して、その配置位置のみを示しているが、第一閉曲面６１の中心に対向するように配置される。そして、再現スピーカ６２（ａ〜ｕ）に、伝達関数ＨＬａ〜ＨＬｕ、または伝達関数ＨＲａ〜ＨＲｕに基づく演算処理を施された再現信号を供給することで、第一閉曲面６１内のいずれの聴取位置においてもその外側に、第一閉曲面５５の外側と同等の音場感を再現することができる。
【００６０】
第一閉曲面６１内には、自動車５０室内の再生用スピーカ５１（ａ〜ｊ）に対応した位置に、測定用マイク６５（ａ〜ｊ）を配置する。これらの測定用マイク６５（ａ〜ｊ）の配置位置を結ぶことによって形成される閉曲面が第二閉曲面７０となる。そして、これらの測定用マイク６５（ａ〜ｊ）の指向性の主軸は、図２２に示した測定用マイク５６（ａ〜ｕ）と同様に、第二閉曲面７０を形成する面における法線方向外向き発散する延長方向に対応させる。
このように、再現環境６０を構築して、再現用スピーカ６２（ａ〜ｕ）から伝達関数ＨＬａ〜ＨＬｕに基づく演算処理を施した再現信号を出力して測定用マイク６５（ａ〜ｊ）で測定することで、測定用スピーカ５２Ｌに対応した伝達関数ＧＬａ〜ＧＬｊを求めることができる。同様に、再現用スピーカ６０（ａ〜ｕ）から伝達関数ＨＲａ〜ＨＲｕに基づく演算処理を施した再現信号を出力して測定用マイク６５（ａ〜ｊ）で測定することで、測定用スピーカ５２Ｒに対応した伝達関数ＧＲａ〜ＧＲｊを求めることができる。
【００６１】
図２４は、測定環境５３、再現環境６０で求められた伝達関数に基づいて自動車５０室内に設置されている再生用スピーカ５１（ａ〜ｊ）に供給する再生信号を生成する音声信号処理装置８０の構成例を説明する図である。この音声信号処理装置８０は、カーオーディオシステムの一部として構成される。
音声再生部８１からは、例えば２チャンネルステレオとされる音声信号ＳＬ、ＳＲが出力される。これらの音声信号ＳＬ、ＳＲは、それぞれ測定スピーカ逆特性処理部８２Ｌ、８２Ｒを介して、演算部８３（８３Ｌ、８３Ｒ）に供給される。演算部８３Ｌでは、音声信号ＳＬに対して伝達関数ＨＬａ〜ＨＬｕに基づく演算処理、また演算部８３Ｒでは伝達関数ＨＲａ〜ＨＲｕに基づく演算処理が行われる。
【００６２】
演算部８３Ｌ、８３Ｒによって伝達関数ＨＬａ〜ＨＬｕ、ＨＲａ〜ＨＲｕに基づく演算処理を施された音声信号は、それぞれ演算部８５（８５Ｌ、８５Ｒ）に供給され、演算部８５Ｌでは、伝達関数ＧＬａ〜ＧＬｊに基づく演算処理、また演算部８５Ｒでは伝達関数ＧＲａ〜ＧＲｊに基づく演算処理が行われる。
そして、演算部８３Ｌ、８３Ｒによって伝達関数ＧＬａ〜ＧＬｊ、ＧＲａ〜ＧＲｊに基づく演算処理を施された音声信号は、加算部８７で加算されて、再生用スピーカ５１（ａ〜ｊ）に供給される再生信号ＳＣａ、ＳＣｂ、ＳＣｃ・・・、ＳＣｊが生成される。この加算部８７における加算処理は、同一の添え字（アルファベット小文字）が付された伝達関数に基づく演算処理を施された音声信号同士を加算するようにされる。
【００６３】
このような演算処理を行って得られた音声信号ＳＣ（ａ〜ｊ）を再生用スピーカ５１（ａ〜ｊ）に供給することで、図２１に示した自動車５０の室内において、この室内にはない仮想スピーカ５１Ｌ、５１Ｒの位置に仮想音像を知覚することができるようになる。
【００６４】
また、演算部８３において演算に用いられる伝達係数ＨＬａ〜ＨＬｕを、例えば著名なコンサートホールや映画館など複数の測定環境で測定して記憶しておくことで、仮想音像の音場を選択することも可能である。この場合、演算部８３に対して音場コントローラ８４を備え、必要に応じてメモリ８４ａに記憶されている伝達関数を選択して演算処理に用いるようにすればよい。
同様に、演算部８５において演算に用いられる伝達係数ＧＬａ〜ＧＬｕを、複数種類の再生用スピーカの配置状態（配置数、配置位置など）に対応させて測定して記憶しておくことで、再生用スピーカ５１（ａ〜ｊ）の配置状態に対応することも可能である。この場合、演算部８５に対して再生用スピーカセッティングコントローラ８６を備え、必要に応じてメモリ８６ａに記憶されている伝達関数を選択して演算処理に用いるようにすればよい。これにより、自動車５０の室内における再生用スピーカ５１（ａ〜ｊ）の複数種類の配置状態に対応させた再生信号ＳＣ（ａ〜ｊ）を生成することができるようになる。
このような、伝達関数の選択は、選択手段として操作部８８を備え、例えば聴視者が選択することができるようにしてもよい。
【００６５】
このように、測定環境の残響音場特性と再生用スピーカの配置状態を独立して設定することができるので、新車種による新しいスピーカ配置状態が提供された場合でも、無響室などの再現環境において、その配置状態に対応した伝達関数を求めればよい。
したがってメモリ８４ａ、８６ａに記憶される伝達関数は、更新することができるようされる。更新の方法としては、例えばディスクなどの記録媒体に記録されている伝達関数を読み込むことができるようにしてもよいし、或いは例えば所要の通信手段を備え、ネットワークを介してダウンロードすることができるようにしてもよい。そして、メモリ８４ａ、８６ａには、それぞれの伝達関数に対応するインパルス応答データが記憶するようにしてもよい。
【００６６】
なお、本発明は、例えば家庭の部屋などや、カーオーディオシステムにおいて音声の再生を行う以外にも、例えば臨場感、没入感を与えるアミューズメント、ゲームのようなバーチャルリアリティー産業機器に適用することもできる。
【００６７】
【発明の効果】
以上、説明したように本発明は、測定環境の音場に基づいて第一の伝達関数を求め、さらに再現環境において前記第一の伝達関数に基づく演算処理を施された音声信号を再生することで、前記再現環境において前記測定環境の音場（例えば残響、音像定位）を再現することができるようにしている。さらに、前記再現環境において、測定環境と同等の音場条件で再生環境に配置される再生用スピーカに対応した第二の伝達関数を求めることができる。
したがって、前記第二の伝達関数は再現環境で求めることができるので、前記測定環境における測定回数を削減することができる。さらに、前記再生用スピーカの新規の配置状態に対応する場合でも、前記再現環境において前記第二の伝達関数を求めることが可能になる。
【００６８】
また、前記第一の伝達関数を求めるための第一閉曲面の等方位性、対称性を利用して、前記第一の伝達関数を所定の規則で転置して利用することができるようにしている。つまり、前記第一閉曲面からみた或る位置を想定した仮想スピーカから求めた伝達関数を、他の位置を想定した仮想スピーカから求めた伝達関数として利用することができるようになる。したがって、前記第一の伝達関数を求めるための測定作業回数を削減することができるようになる。
【００６９】
また第一の伝達関数と第二の伝達関数を個々に管理することができるようにしているので、測定環境の音場と再生用スピーカの配置状態を、独立して選択することができる。したがって、測定環境の音場と再生用スピーカの配置状態を個々に更新、追加することも可能となる。
さらに、第一の伝達関数と第二の伝達関数を個々に管理する場合に、前記第一、第二の伝達関数を所定の規則で転置して利用することができるようにすることで、前記第一、第二の伝達関数を記憶しておく記憶手段の容量を小さくすることができるようになる。
【図面の簡単な説明】
【図１】本発明の本実施の形態の音声再生システムにおける測定環境について説明する図である。
【図２】本実施の形態の音声再生システムにおける再現環境における再現音声の再生系について説明する図である。
【図３】再現環境について説明する図である。
【図４】再現環境の第一閉曲面の内側においてインパルス応答を測定する例を説明する図である。
【図５】再現環境の第一閉曲面の内側においてインパルス応答を測定する例を説明する図である。
【図６】再現環境における仮想スピーカの伝達関数について説明する図である。
【図７】再生環境における５チャンネルステレオ音声の再生系について説明する図である。
【図８】再生環境における再生用スピーカの配置を説明する図である。
【図９】再生環境において第二閉曲面内を聴取位置とした場合の残響音場、及び音像定位を説明する図である。
【図１０】再生環境の第一閉曲面において伝達関数Ｇを求める例を説明する図である。
【図１１】再生環境において測定環境の音場を再現する場合の手順を説明するフローチャートを示す図である。
【図１２】伝達関数Ｈと伝達関数Ｇを個別に管理して行列演算を行う場合の手順を説明するフローチャートを示す図である。
【図１３】５チャンネルステレオの再生環境を説明する図である。
【図１４】５チャンネルステレオの信号Ｄ（Ａ〜Ｅ）に伝達関数Ｔ（Ａ〜Ｅ）に基づく演算処理を行う音声信号処理装置の構成例を説明する図である。
【図１５】測定環境としての映画館においてスピーカから測定用マイクまでの伝達関数を説明する図である。
【図１６】再生環境において再現される測定環境の残響、及び音像定位について説明する図である。
【図１７】仮想スピーカと再現用スピーカの位置関係を説明する図である。
【図１８】測定環境の壁からの反射情報の回転を説明する図である。
【図１９】第二閉曲面内において移動する仮想的な音像を作成する例を示す模式図である。
【図２０】仮想的な音像を移動させる音声信号を生成する音声信号処理装置の構成例を説明する図である。
【図２１】自動車の室内に配置される再生用スピーカを説明する図である。
【図２２】測定環境において自動車の例えば外観を囲むことができるような立体を想定した第一閉曲面を示す図である。
【図２３】再現環境において図２２に示した測定環境を再現する例を説明する図である。
【図２４】伝達関数に基づいて自動車室内に設置されている再生用スピーカに供給する再生信号を生成する音声信号処理装置の構成例を説明する図である。
【図２５】サンプリングリバーブ方式について説明する図である。
【図２６】測定用スピーカから出力される測定用音声の説明図である。
【図２７】測定環境で求めた伝達関数を用いて音声信号の生成を行う例を説明する図である。
【符号の説明】
１，５５ 測定環境、２ 測定信号再生手段、３，５２（Ｌ、Ｒ） 測定用スピーカ、４（ａ〜ｐ），１３（Ａ〜Ｅ），５６ 測定用マイク、５，１５，２２，４０，８０ 音声信号処理装置、６，１６，２３，４１，８１ 音声再生部、７（ａ〜ｐ），１７（ａ〜ｅ），２４，２６，２８，３０，３２，４３（ａ〜ｈ），８３（８３Ｌ，８３Ｒ），８５（８５Ｌ、８５Ｒ） 演算部、８（ａ〜ｐ），３６，６２ 再現用スピーカ、１０，１２，６１ 第一閉曲面、１１，６０ 再現環境、１３（Ａ〜Ｅ），６５ 測定用マイク、１４，１９，７０ 第二閉曲面、１８，５１ 再生用スピーカ（Ａ〜Ｅ）、２０ スピーカ、２１ 映画館、２５，２７，２９，３１，３３，８７ 加算部、３５ 仮想スピーカ、４２ ８２（Ｌ，Ｒ） 測定スピーカ逆特性処理部、４４ 音像定位コントローラ、５０自動車、８４ 音場コントローラ、８４ａ，８６ａ メモリ、再生用スピーカセッティングコントローラ[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an audio signal processing method and an audio signal processing device.
[0002]
[Prior art]
For example, when viewing contents such as movies and music, reverberation is added to give the reproduced sound a sense of reality.
As the reverberation adding process, a so-called digital reverb method is known. This digital reverb system generates a lot of delay information, which is a random delay time with respect to the original sound, and further reduces the volume as the delay time becomes longer, or applies feedback at places where the delay time is longer to reduce the reverberation time. Signal processing such as taking a long time is performed. As a result, a reverberation effect can be artificially generated for the original sound. However, since the parameters for generating the delay information are set based on the audibility of the operator who sets the parameters, this setting work is complicated. In addition, there is no concept of localizing the original sound because it artificially creates reverberation, and it is not excellent in reproducing the sound field.
[0003]
For example, a sampling reverb method is known as a method for actually measuring an impulse response in a sound field space and obtaining a reverberation effect based on spatial information such as localization of a sound source.
In this sampling reverb system, for example, as shown in FIG. 25, a measurement speaker 91 or the like is arranged as a sound source in a measurement environment (measurement sound field) 90 such as a hall. Then, an impulse response measurement audio signal TSP such as a time stretched pulse is supplied to the measurement speaker 91, and one or a plurality of measurement microphones 92a and 92b arranged at required positions in the same sound field. , 92c, and 92d are caused to input measurement sound output from the measurement speaker 91. In this case, as indicated by an arrow in FIG. 26, the measurement microphone 92a can detect the direct sound from the measurement speaker 91 and the reflected sound output from the measurement speaker 91 and reflected in the measurement environment. it can. Although not shown, the same applies to the measurement microphones 92b, 92c, and 92d.
Then, the impulse response measuring device 93 performs necessary signal processing on the sound signal waveform input from the measurement microphone 92 (ad), so that the impulse response including the reverberation of the measurement environment 90 (measured from the measurement speaker) is performed. To the transfer microphone). In this figure, the transfer functions corresponding to the respective measurement microphones 92 (a to d) are Hfl, Hfr, Hbl, and Hbr, respectively.
[0004]
Thus, when producing | generating an audio | voice signal using the transfer function calculated | required in the measurement environment 90, it carries out as FIG. 27 shows, for example.
As shown in the figure, reproduction speakers 101a, 101b, 101c, and 101d are arranged in a reproduction environment 100 such as a home room, and contents are reproduced by the audio reproduction means 102 for each of the reproduction speakers 101 (a to d). The playback signal is supplied. Then, by convolving impulse responses corresponding to the transfer functions Hfl, Hfr, Hbl, and Hbr to the audio signals reproduced by the audio signal reproducing unit 102, and supplying them to the reproduction speakers 101 (a to d), A reverberation effect similar to that of the measurement environment 90 can be obtained in the reproduction environment 100. As described above, the sampling reverb method uses an actually measured transfer function, so that it is excellent in the reproducibility of the sound field during reproduction, and the sound image localization in the reproduced sound field becomes clear.
[0005]
Further, when the arrangement positions of the measurement microphones 92 (a to d) in the measurement environment 90 and the reproduction speakers 101 (a to d) in the reproduction environment 100 are geometrically associated with each other, the reproduction speakers in the reproduction sound field can be obtained. In the enclosed area (within the closed curved surface), the sound source localization (sound image localization) of the measurement sound field is clearly reproduced. That is, in addition to obtaining a reverberation effect, it is also possible to reproduce the sound field of the measurement environment 90.
[0006]
In addition to the sampling reverb method described above, the method of actually using the measurement data of the sound field space includes, for example, a proximity four-point method that obtains a reverberation effect by analyzing a reflection component that looks like a virtual sound source. Are known.
The proximity four-point method is a method in which impulse response is measured at four points that are not on the same plane, and the spatial position information and the magnitude of the sound source are calculated. According to the proximity four-point method, the impulse response at an arbitrary position in the measured sound field space can be calculated from the spatial information of the sound source and the size and position of the sound source. Therefore, reverberation addition equivalent to that of the measured sound field space can be realized by convolving the impulse response thus obtained with the audio signal.
However, this proximity four-point method is originally a technique for analysis. For this reason, when artificially creating the spatial information of the sound source from which the impulse response is measured and the size and position of the sound source, it is necessary to limit the size and number of the sound sources with arbitrary threshold values. Therefore, the quality of reproducing the sound of the sound field space is not high compared to the sampling reverb method. However, since the impulse response at an arbitrary point in the measured sound field space can be obtained by arithmetic processing, there is an advantage that it is possible to flexibly cope with the arrangement and number of speakers in the reproduction environment. is doing.
[0007]
[Problems to be solved by the invention]
By the way, recently, there is an increase in content in which audio is constituted by audio signals of multiple channels (3 or more channels). Therefore, for example, in a reproduction environment such as a home, a system that reproduces sound using three or more reproduction speakers is widespread.
However, in such a reproduction system, when the measurement environment is reproduced by, for example, the sampling reverb method, the transfer function is obtained in consideration of the number and arrangement (installation state) of the reproduction speakers 101 (ad) in the reproduction environment 100. There must be. Therefore, in the measurement environment 90, it is necessary to perform measurement assuming a plurality of reproduction systems used in the reproduction environment 100 while changing the number and arrangement of the measurement microphones 92 (ad). In addition, when a new reproduction system having a different installation state of the reproduction speakers 101 (a to d) is constructed, it is necessary to perform measurement corresponding to the reproduction system each time.
For this reason, costs and time required for measurement are wasted in a measurement environment (for example, a hall).
[0008]
Therefore, it is desired to be able to reproduce the reverberation characteristics of the measurement sound field flexibly with respect to the installation state of the reproduction speaker.
[0009]
[Means for Solving the Problems]
In order to solve such a problem, the present invention provides a first sound generation step of sounding a first sound from a first sound source arranged outside the first closed surface, and the first closed surface A first measurement step of measuring the voice signal of the first voice at a plurality of positions above, and the first sound source from the first sound source based on the voice signal measured by the first measurement step A transfer function generating step for generating a first transfer function corresponding to each of the plurality of positions on the closed surface, and a second geometrically equivalent position to the plurality of positions on the first closed surface. A second sound generation step of generating a second sound based on the first transfer function, and a plurality of sound generators on the second closed surface arranged inside the first closed surface A second measurement step of measuring the audio signal of the second audio at a position of A transfer function generation step for generating a second transfer function corresponding to each of the plurality of positions on the second closed surface from the second sound source based on the audio signal measured by the determination step; An audio signal obtained by performing arithmetic processing based on the second transfer function to obtain an audio signal, and audio that outputs the audio signal obtained based on the second transfer function by the arithmetic step An audio signal processing method is configured including a signal output step.
[0010]
Furthermore, as an audio signal processing method, a first sound generation step of generating a first sound from a first sound source arranged outside the first closed curved surface, and a plurality of positions on the first closed curved surface A first measurement step of measuring the voice signal of the first voice, and the voice signal measured by the first measurement step from the first sound source on the first closed curved surface. A transfer function generation step for generating a first transfer function corresponding to each of a plurality of positions, and a second sound source is disposed at a position geometrically equivalent to the plurality of positions on the first closed curved surface A second sound generation step for generating a second sound based on the first transfer function, and a plurality of positions on a second closed curved surface arranged inside the first closed curved surface. A second measuring step for measuring a second voice signal; and the second measuring step. A transfer function generating step of generating a second transfer function corresponding to each of the plurality of positions on the second closed curved surface from the second sound source based on the sound signal measured by Obtained by performing a first calculation step on the signal based on the first transfer function and a calculation process based on the first transfer function in the first calculation step Obtained by performing a second computation step on the audio signal based on the second transfer function and a signal processing based on the second transfer function in the second computation step. A voice signal output step of outputting the voice signal.
[0011]
In addition, as an audio signal processing device, a storage unit that stores a transfer function, an arithmetic processing unit that performs an arithmetic process based on the transfer function stored in the storage unit for the input audio signal, and the arithmetic process Output means for outputting the sound signal obtained by the means, wherein the transfer function is a first from a virtual sound source position where the sound signal is to be reproduced and localized to a first plurality of positions on the first closed curved surface. And a plurality of second positions on a second closed surface arranged on the inside of a closed surface geometrically equivalent to the first closed surface from each of the first plurality of positions. It is set as the transfer function calculated | required based on the 2nd transfer function until.
[0012]
Furthermore, as the audio signal processing device, the first storage unit that stores the first transfer function, the second storage unit that stores the second transfer function, and the input audio signal, First arithmetic processing means for performing arithmetic processing based on a first transfer function stored in one storage means, and the second storage for the audio signal obtained by the first arithmetic processing means. Second arithmetic processing means for performing arithmetic processing based on the second transfer function stored in the means, and output means for outputting the audio signal obtained by the second arithmetic processing means, The transfer function is a transfer function from a virtual sound source position where the sound signal is to be reproduced and localized to a first plurality of positions on the first closed curved surface, and the second transfer function is the first transfer function. Closed geometrically equivalent to the first closed curved surface from each of a plurality of positions. A second transfer function to the plurality of positions in the second closed surface on which is disposed on the inner surface.
[0013]
According to the present invention, the reproduction environment is obtained by obtaining the first transfer function based on the sound field of the measurement environment, and further reproducing the audio signal signal-processed based on the first transfer function in the reproduction environment. The sound field of the measurement environment (for example, reverberation, sound image localization) can be reproduced. Furthermore, in the reproduction environment, a second transfer function corresponding to a reproduction speaker arranged in the reproduction environment can be obtained under a sound field condition equivalent to that in the measurement environment.
Therefore, after obtaining the first transfer function in the measurement environment, the second transfer function corresponding to the reproduction speaker can be obtained in the reproduction environment, and the work efficiency in the measurement environment is improved. be able to.
[0014]
Also, by separately obtaining the first transfer function and the second transfer function and managing them separately, the sound field of the measurement environment and the arrangement of the measurement playback speakers in the reproduction environment can be selected independently. Become.
[0015]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in the following order.
1. Reproduction of one virtual speaker 2. Reproduction of multiple virtual speakers Example 4 using circular symmetry 4. Example of using circular symmetry in movement of virtual sound source Car audio system [0016]
In this specification, unless otherwise specified, “calculation processing based on a transfer function” for an audio signal means that an impulse response corresponding to the transfer function is subjected to convolution integration processing on the audio signal, It is assumed that the audio signal is converted into the frequency domain and multiplied by the transfer function. The former process can be realized as, for example, an FIR (finite impulse response) filter, and the latter can be realized by performing an FFT (fast Fourier transform) operation.
[0017]
1. Reproduction of one virtual speaker FIG. 1 is a schematic diagram for explaining a measurement environment in the sound reproduction system of the present embodiment.
For example, in the measurement environment 1 that is a hall or the like, for example, the measurement microphones 4a, 4b, 4c, 4d, 4e, 4f, 4g are arranged on the circumference having the radius R_bnd at a position that is not close to the wall of the measurement environment 1. 4h, 4i, 4j, 4k, 4l, 4m, 4n, 4o, 4p are arranged. In the present embodiment, a circumference having a radius R_bnd where the measurement microphones 4 (ap) are arranged is shown as a first closed curved surface 10 serving as a first sound field.
Each measurement microphone 4 (ap) has its directivity directed outward in the normal direction of the first closed curved surface 10. In the following description of the present specification, the arrow shown on the microphone indicates the directivity.
Further, the measurement speaker 3 is disposed as a sound source at a position having a radius R_sp from the center of the first closed curved surface 10. For example, the measurement signal TSP is supplied from the measurement signal reproducing means 2 to the measurement speaker 3. Since the measurement speaker 3 is provided to reproduce a virtual speaker in a reproduction environment, which will be described later, it is desirable that the directivity and the frequency characteristics assume an audibility to the listener in the reproduction environment.
[0018]
In the measurement environment 10, the measurement signal TSP is supplied to the measurement speaker 3 so that the measurement sound output from the measurement speaker 3 is input to each measurement microphone 4 (ap). FIG. 1 schematically shows a voice path from the measurement speaker 3 to the measurement microphone 4a.
The sound signal of the sound detected by each measurement microphone 4 (ap) is supplied to an impulse response measurement device (not shown), and the impulse corresponding to each measurement microphone 4 (ap) from the measurement speaker 3 Measure the response. The impulse response may be about 5 to 10 seconds for a large hole or the like, but may be shorter in a small hole or a hole with little sound. By this measurement, a transfer function based on this impulse response can be obtained. That is, FIG. 1 shows a voice path when the transfer function Ha corresponding to the measurement microphone 4a is obtained. Although not shown, transfer functions Hb to Hp corresponding to the measurement microphone 4b to the measurement microphone 4p can be obtained. The impulse response measurement may be performed for each measurement microphone, or may be performed simultaneously for all the measurement microphones 4 (ap). Further, the measurement signal is not limited to the TSP signal, and pseudo random noise, a music signal, or the like may be used. In the following description, the transfer function from the measurement speaker to the measurement microphone in the measurement environment is “H”.
[0019]
In this way, in the measurement environment 10, the transfer functions Ha, Hb, Hc, Hd..., Hp corresponding to the respective measurement microphones 4a, 4b, 4c, 4d. By using these transfer functions Ha to Hp, the measurement environment 1 can be reproduced in an environment (reproduction environment) other than the measurement environment 1.
First, referring to FIG. 2, a reproduction system of reproduced sound in a reproduction environment will be described.
In the reproduction signal generation device 5, the sound reproduction unit 6 can output an arbitrary sound signal S, for example. The audio signal S output from the audio reproduction unit 6 is supplied to the calculation units 7a, 7b, 7c, 7d..., 7n, 7o, 7p. The calculation unit 7 (ap) performs a required calculation process on the supplied audio signal S and performs a calculation process based on the transfer functions Ha to Hp obtained in the measurement environment 1. As a result, the reproduction unit SHa, SHb, SHc, SHd..., SHn, SHo, SHp in which the impulse response corresponding to the transfer function is convoluted with the audio signal S from the arithmetic unit 7 (ap). Is output. Each reproduction signal SH (ap) is supplied to reproduction speakers 8a, 8b, 8c, 8d..., 8n, 8o, 8p arranged in the reproduction environment. Therefore, each reproduction speaker 8 (ap) outputs sound based on the reproduction signal SH (ap) based on the transfer function H (ap) in the measurement environment 1.
[0020]
FIG. 3 is a schematic diagram for explaining the reproduction environment.
The reproduction environment 11 is, for example, an anechoic room or a studio with little reverberation.
Then, the reproduction speaker 8 (ap) shown in FIG. 2 is arranged in the reproduction environment 11. In this case, the reproduction speaker 8 (ap) corresponds to the arrangement position of the measurement microphone 4 (ap) shown in FIG. 1 and is on the outer periphery of the first closed curved surface 12 formed with the radius R_bnd. Arranged toward the inside. That is, the reproduction speakers 8 (ap) and measurement microphones 4 (ap) with the same subscript (alphabet) correspond to each other.
The first closed curved surface 10 in the measurement environment 1 and the first closed curved surface 12 in the reproduction environment 11 are equivalent closed curved surfaces formed with the same radius, but are given different reference numerals for convenience.
Then, by supplying the reproduction signals SH (ap) to the reproduction speakers 8 (ap), the listener inside the first closed curved surface 12 can use the measurement speakers 3 shown in FIG. It can be felt that the sound field when the audio signal S is reproduced is reproduced on the outside of the first closed curved surface.
[0021]
Here, when there is no sound source in a certain closed curved surface, in order to accurately reproduce the sound field in another place, the sound pressure and normal direction of the outer periphery of the closed curved surface are reproduced between the original sound field and the reproduced sound field. It is known that the particle velocities need to be matched (refer to “Literature: Electronic systems and digital processing” edited by the Institute of Electronics, Information and Communication Engineers (Corona)). Specifically, a large number of omnidirectional microphones that are paired on a closed curved surface are installed, and calculation processing is performed to measure sound pressure and particle velocity at each point. For this reason, a large number of measurement microphones are installed outward in the normal direction on the first closed curved surface 10, and a large number of reproduction speakers are arranged at positions corresponding to these measurement microphones on the first closed curved surface 12. Thus, when the inside of the first closed curved surface 12 is set as the viewing position, the listener can obtain the same localization feeling and reverberation feeling as in the measurement environment 1, and the measurement is not in the reproduction environment 11. A virtual sound image can be perceived at the position of the speaker 3 for use. In other words, at any listening position inside the first closed curved surface 12, a sound field feeling equivalent to that of the measurement environment 1 can be obtained outside.
However, as described above, it is difficult to actually realize that a very large number of microphones and a large number of reproduction speakers are required. Therefore, the inventor of this application pays attention to the fact that sound pressure and particle velocity components are included in the output of a directional microphone, for example, a unidirectional microphone, and a limited number of directional microphones and a corresponding number of them. Experiments have confirmed that almost the same acoustic effect can be obtained with the reproduction speaker.
[0022]
In this manner, the sound field in the measurement environment 1 such as a hall can be reproduced in the reproduction environment 11 that is an anechoic chamber or the like. Accordingly, as shown in FIG. 1, the measurement of the impulse response in the measurement environment 1 is performed once, for example, and then the measurement environment 1 is set as the reproduction environment 11 by using the measurement data (transfer function) thereafter. In other environments, the sound field of the measurement environment 11 can be simulated at any time.
Thereby, the measurement of the impulse response assuming the measurement environment 1 can be performed in the first closed curved surface 12, and the measurement work in the measurement environment 1 needs to be performed many times. In other words, it is possible to efficiently perform the measurement work in the concert hall that is the measurement environment 1.
[0023]
FIGS. 4 and 5 are schematic diagrams for explaining an example of measuring the impulse response inside the first closed curved surface 12 inside the first closed curved surface 12 of the reproduction environment 11.
In this case, measurement microphones 13A, 13B, 13C, 13D, and 13E are arranged inside the first closed curved surface 12. These measurement microphones 13 (A to E) are arranged so as to correspond to reproduction speakers arranged in a reproduction environment such as a home room. That is, in the illustrated example, a playback system including, for example, a 5-channel playback speaker is assumed. In the present embodiment, the closed curved surface formed with the measurement microphones 13 (A to E) as the outer periphery is referred to as a second closed curved surface 14. Therefore, the inside of the second closed curved surface 14 is the listening position.
Since the second closed curved surface 14 corresponds to the arrangement of the reproduction speakers in the reproduction environment, the second closed curved surface 14 does not necessarily have a circular shape like the first closed curved surfaces 10 and 12. Furthermore, since the second closed curved surface 14 needs to be formed inside the first closed curved surfaces 10 and 12, when performing measurement in the measurement environment 1, the width of the second closed curved surface 14 is taken into consideration. It is desirable to form the closed curved surfaces 10 and 12.
Also, the audio signal output from the measurement microphone 13 (A to E) is supplied to, for example, an impulse response measurement unit (not shown), and the transfer function is obtained by an arithmetic process such as a required phase shift process.
[0024]
For example, as shown in FIG. 4, the transfer function of each measurement speaker 8 (ap) to the measurement microphone 13A is G1A, G2A, G3A... The transfer functions to the measurement microphone 13B are G1B, G2B..., The transfer functions to the measurement microphone 13C are G1C, G2C..., The transfer functions to the measurement microphone 13D are G1D, G2D. The transfer functions to the microphone 13E are G1E, G2E,. In the following description, the transfer function from the reproduction speaker to the measurement microphone in the reproduction environment is “G”.
Further, as shown in FIG. 5, the transfer functions from the measurement speaker 8a to the measurement microphones 13A, 13B, 13C, 13D, and 13E are G1A, G1B, G1C, G1D, and G1E. The transfer functions from 8b to the measurement microphones 13A, 13B, 13C, 13D, and 13E are G2A, G2B, G2C, G2D, and G2E.
Furthermore, the transfer function from the measurement speaker 3 in the measurement environment 1 to the measurement microphone 13 (A to E), which is a virtual sound image, including the transfer function H (a to p) in the measurement environment 1 described above is TA. , TB, TC, TD, TE. In the following description, the transfer function from the measurement speaker in the measurement environment, which is a virtual sound image in the reproduction environment, to the measurement microphone in the reproduction environment is “T”.
[0025]
In this case, the transfer functions TA, TB, TC, TD, and TE can be expressed as shown in Equation 1.
[Expression 1]

That is, as described above, it is considered that the sound field of the measurement environment 1 is reproduced outside the first closed curved surface for the listener inside the first closed curved surface 12 by the transfer function H (ap). The measurement microphone 13 (A to E) of the second closed curved surface 14 shown in FIG. 6 is approximately the same as the situation where the measurement sound is output from the measurement speaker 3 that is a virtual sound image in the reproduction environment 11. become.
Therefore, inside the second closed curved surface 14, the transfer coefficients TA, TB, TC, TD, and TE are equivalent to the transfer function when the measurement sound is output from the measurement speaker 3 in the measurement environment 1.
[0026]
By the way, for example, when N reverberation sound field characteristics in the measurement environment 1 are to correspond to the arrangement state of M kinds of reproduction speakers, the measurement speakers 3 are arranged in N places, and the reproduction speakers are arranged. It is necessary to perform N measurements corresponding to the state. That is, it was necessary to perform N × M measurements. However, in the present invention, the measurement corresponding to the arrangement state of the reproduction speakers is performed in the reproduction environment 11, and therefore, the measurement environment 1 may be performed N times.
In addition, when the types of reproduction speaker arrangement states are increased, it is necessary to perform measurement in the N measurement environments 1 when reproducing the N measurement environments 1 corresponding to the arrangement states. . Therefore, when the number of measurement positions (N locations) in the measurement environment 1 is large, it is a difficult task. However, in the present invention, if transfer functions are obtained in the measurement environments 1 at N locations, a reproduction speaker is used. The measurement may be performed in the reproduction environment 11 each time the number of arrangement states increases. Therefore, even when dealing with the arrangement state of the new reproduction speaker, it is not necessary to go to N measurement environments, and an efficient measurement operation can be performed.
[0027]
Next, based on the transfer functions TA, TB, TC, TD, and TE obtained in the reproduction environment 11, an example of reproducing, for example, 5-channel stereo sound in a reproduction environment such as a home room will be described. .
First, referring to FIG. 7, a 5-channel stereo sound playback system in a playback environment will be described.
In the playback device 15, the audio playback unit 16 can play back an audio signal in content recorded on a recording medium such as a disc. The audio signal S output from the audio reproduction unit 16 is supplied to the calculation units 17a, 17b, 17c, 17d, and 17e. The calculation unit 17 (a to e) performs a required calculation process on the supplied audio signal S, and performs a calculation process based on the transfer functions TA, TB, TC, TD, and TE obtained in the reproduction environment 11. Do. As a result, the reproduction signals STA, STB, STC, STD, and STE obtained by performing arithmetic processing on the audio signal S based on the transfer function are output from the arithmetic unit 17 (a to e). Each reproduction signal ST (A to E) is supplied to reproduction speakers 18A, 18B, 18C, 18D, and 18E. Therefore, each reproduction speaker 18 (A to E) outputs sound based on the reproduction signals STA, STB, STC, STD, and STE based on the transfer function T (A to E).
[0028]
The reproduction speakers 18 (A to E) in the reproduction environment are arranged at required positions on the second closed curved surface 19 as shown in FIG. The first closed curved surface 19 and the first closed curved surface 12 in the reproduction environment 11 are equivalent closed curved surfaces, but different reference numerals are attached for convenience.
As described above, the reproduction speakers 18 (A to E) are arranged on the second closed curved surface 19 so as to face the inside thereof, and the sound based on the reproduction signals STA, STB, STC, STD, and STE is output. For a listener who is inside the two closed curved surface 19, a sound field equivalent to the sound field by the reproduction speaker 8 (ap) arranged on the first closed curved surface 12 indicated by a broken line can be reproduced. Therefore, when the inside of the second closed curved surface 19 is set as the listening position, a reverberant sound field and sound image localization in the measurement environment 1 can be obtained as shown in FIG. Accordingly, for example, the user can listen to the sound of the content by using a reverberant sound field such as a hall and sound image localization while staying in a home room.
Further, since the present invention obtains the transfer function based on the impulse response actually measured in the measurement environment and the reproduction environment, for example, with a higher quality voice than the method of referring to the analysis result such as the 4-point proximity method. It is possible to reproduce reverberation and sound image localization.
[0029]
Note that the number of playback speakers in the playback environment may vary depending on the configuration of the playback system, and it is not easy to make the arrangement position the same depending on home circumstances. In such a case, conventionally, it was necessary to perform measurement in accordance with the arrangement of the reproduction speakers in the measurement environment. In this embodiment, the transfer function in the measurement environment is used as master data, and thereafter the anechoic chamber is used. Thus, measurement according to the reproduction environment can be performed. That is, it is possible to reduce the fee for use of the hall for performing the measurement and to easily obtain a transfer function corresponding to the reproduction environment.
In addition, even if the measurement environment that has a characteristic reverberation sound field is, for example, in a remote place, it is not necessary to go to the measurement environment if the master data is obtained as described above. Even in this respect, it is possible to efficiently perform measurement in the measurement environment.
[0030]
Furthermore, in the present embodiment, an audio signal processing device is shown as the configuration of the playback device 15, but it is applied to, for example, an AV (Audio Visual) amplifier, a surround processor, or a CD or DVD player used at home or the like. Can do.
[0031]
By the way, as shown in Equation 1, the matrix G and the vector Hv can be separated. Therefore, in the reproduction environment 11, for example, the transfer function T (A to E) can be obtained without taking the measurement method described with reference to FIGS.
For example, data corresponding to the vector Hv (impulse response for obtaining the transfer function H) in the measurement environment is measured as reverberation characteristic data such as a hall. On the other hand, as data corresponding to the arrangement state of the reproduction speakers in the reproduction environment, for example, the transfer function G to the measurement microphone 13 (A to E) in the first closed curved surface 12 as shown in FIG. Separately measured based on the measurement signal TSP. Then, the transfer function T (A to E) can be obtained by Expression 1 using the vector Hv and the matrix G obtained individually.
Therefore, the vector Hv and the matrix G are individually managed in the audio signal processing apparatus so that matrix calculation can be performed as necessary, so that the number of transfer function measurements can be reduced. Furthermore, the reverberation characteristics and the arrangement of the reproduction speakers can be selected independently.
[0032]
In the present embodiment, the measurement speaker 3 is described as a virtual speaker in the reproduction environment in the measurement environment 1, but the source reproduced in the reproduction environment is, for example, a natural sound that is easily perceived by humans. For example, when the sound is composed of one kind of sound, or when the characteristic of the measurement speaker 3 is a characteristic close to flat, the source itself is recognized as a virtual sound image. However, in order to accurately represent the virtual sound image of the source itself, it is more desirable to perform correction including the inverse characteristics of the speaker.
[0033]
FIG. 11 is a flowchart illustrating a procedure for obtaining a transfer function as described above and reproducing the sound field of the measurement environment in the reproduction environment.
First, the measurement sound is output from the measurement speaker 3 on the first closed curved surface 10 of the measurement environment 1 to measure the impulse response (S001), and the transfer function H is obtained based on the measured impulse response (S002). Next, the reproduction sound is output by the sound signal subjected to the calculation process based on the transfer function H in the first closed curved surface 12 of the reproduction environment 11 (S003), and the impulse response of the reproduction sound is output on the second closed curved surface 14. Is measured (S004), and a transfer function G is obtained based on the impulse response measured on the second closed curved surface 14 (S005). Then, when sound is reproduced in the reproduction environment, the reproduced sound reproduction signal is subjected to arithmetic processing based on the transfer function G and output (S006).
As a result, the reverberation effect and sound image localization equivalent to those in the measurement environment can be reproduced in the reproduction environment as shown in FIG.
[0034]
Further, the procedure in the case of performing matrix calculation by separately managing the transfer function H and the transfer function G is as shown in the flowchart of FIG.
Similar to the flowchart shown in FIG. 11, the measurement sound is output from the measurement speaker 3 on the first closed curved surface 10 of the measurement environment 1 to measure the impulse response (S101), and the transfer function is based on the measured impulse response. H is obtained (S102). Then, the measurement speaker 3 is arranged in the reproduction environment 11, the measurement sound is output from the measurement speaker 3 on the first closed curved surface 12, and the impulse response is measured (S 103). A transfer function G is obtained based on the measured impulse response (S104).
When sound is reproduced in the reproduction environment, arithmetic processing based on the transfer function H is performed on the reproduced audio reproduction signal (S105), and further, the audio reproduction signal subjected to arithmetic processing based on the transfer function H Is subjected to arithmetic processing based on the transfer function G (S106).
By such a procedure, the transfer function H and the transfer function G are individually managed and matrix calculation is performed as necessary, so that the reverberation characteristics and the arrangement of the reproduction speakers can be independently selected.
[0035]
2. Reproduction of a plurality of virtual speakers Next, an example of generating a transfer function corresponding to a reproduction system capable of realizing, for example, 5-channel stereo composed of discrete audio signals will be described.
FIG. 13 is a diagram for explaining a reproduction environment of 5-channel stereo.
The reproduction environment shown in this figure is, for example, a wide space such as a movie theater 21, and five speakers 20A, 20B, 20C, 20D, and 20E indicated by solid lines are arranged in order to realize discrete five channels. A center front signal DA, a front right signal DB, a rear right signal DC, a rear left signal DD, and a front left signal DE are supplied to each speaker 20 (A to E), and sound corresponding to each signal is reproduced. . That is, in the movie theater 21 as the reproduction environment shown in this figure, a reverberation effect in a wide space can be obtained.
[0036]
However, in a narrow space such as a home room, for example, it is difficult to arrange the speakers 20 (A to E) at the same interval as the movie theater 21 shown in FIG. It may be arranged as indicated by the broken line. Therefore, in order to construct a sound field in which sound image localization and reverberation effect can be obtained like a movie theater 21 by the reproduction speakers 18 (A to E) on the second closed curved surface 19, the signals D (A to E) are used. May be subjected to arithmetic processing based on the transfer function T (A to E).
[0037]
FIG. 14 is a diagram illustrating a configuration example of an audio signal processing device that performs arithmetic processing based on the transfer function T (A to E) for each signal D (A to E).
The audio signal processing device 22 shown in FIG. 14 outputs a center front signal DA, a front right signal DB, a rear right signal DC, a rear left signal DD, and a front left signal DE as reproduced audio signals from the audio reproducing unit 23. Have been able to.
The signals D (A to E) output from the sound reproduction unit 23 are respectively calculated by calculation units 24 (a to e), 26 (a to e), 28 (a to e), 30 (a to e), and 32. (A to e). For example, in the arithmetic unit 24 (a to e), the transmission coefficient TAA for the center front signal DA, the transmission coefficient TBA for the front right signal DB, the transmission coefficient TCA for the rear right signal DC, and the rear left signal DD. An arithmetic process based on the transmission coefficient TEA is performed on the transmission coefficient TDA and the front left signal DE. Then, the addition signals 25, 27, 29, 31, and 33 add the signals subjected to the arithmetic processing to generate the reproduction signal VA.
[0038]
Incidentally, transfer functions TA (A to E), TB (A to E), TC (A to E), TD (A to E), which are used in the arithmetic units 24, 26, 28, 30, and 32 shown in FIG. TE (A to E) can be obtained by the method described above with reference to FIG.
That is, the speaker 20A shown in FIG. 13 can be made to correspond to the measurement speaker 3 that is a virtual sound image in the reproduction environment shown in FIG. That is, the transfer functions TA, TB, TC shown in FIG. 6 correspond to the transfer functions TAA, TAB, TAC... Shown in FIG.
In this case, the transfer functions TA (A to E), TB (A to E), TC (A to E), TD (A to E), and TE (A to E) are as shown in Equation 2. Can be represented.
[Expression 2]

[0039]
Further, the virtual measurement speaker 3 shown in FIG. 6 is made to correspond to the speaker 20B, speakers 20C, 20D, and 20E shown in FIG. 13, thereby measuring the second closed curved surface 19 from each speaker 20 (B to E). Transfer functions TB (A to E), TC (A to E), TD (A to E), and TE (A to E) between the microphones 13 (A to E) may be obtained. Here, as shown in FIG. 15, in a movie theater 21 as a measurement environment, the transfer function from the speaker 20A to the measurement microphone 4 (ap) is represented by Ha (1-16), and the measurement function from the speaker B to the measurement microphone. 4 (ap) to 4 (ap), the transfer function from the speaker 20C to the measurement microphone 4 (ap) to Hc (1 to 16), the transfer function from the speaker 20D to the measurement microphone 4 ( When the transfer function from a to p) is Hd (1 to 16), the transfer function from the speaker 20E to the measurement microphone 4 (ap) is He (1 to 16), and these transfer functions are matrix H. The matrix T shown in Equation 2 can be expressed as shown in Equation 3.
[Equation 3]

[0040]
By obtaining the transfer function H as shown in Equation 3 and further obtaining the transfer function G as described in FIGS. 4 and 5, for example, the transfer function TA (A to E) is finally obtained. TB (A to E), TC (A to E), TD (A to E), and TE (A to E) are obtained.
Then, the transfer function TA (A to E), TB (A to E), TC (A to E), TD (A to E), TE (A to E) is performed by the audio signal processing device 22 shown in FIG. The audio signals V (A to E) subjected to the arithmetic processing based on the above are supplied to the reproduction speakers 18 (A to E) in the reproduction environment shown in FIG. In the second closed curved surface 19 composed of the speakers 18 (A to E), a sound field that assumes the speakers 20 (A to E) as virtual speakers is formed. That is, for example, reverberation and sound image localization in a wide space such as a movie theater can be reproduced in a room used as a reproduction environment. As long as the listener is within the second closed curved surface 19, the listener can obtain a sound field feeling equivalent to the sound field in the original measurement environment, and can be immersed in a richer presence. it can.
A 5.1 channel system applied to, for example, a DVD-Video (Digital Versatile Disc Video) is known as an audio playback system for discrete audio signals. When the present invention is applied to such a reproduction system, it is possible to realize an appreciation that reproduces a room larger than a reproduction environment in which a listener is present and a room with good sound. As a result, it is possible to relax an unnatural sound that makes the user aware of the presence of the reproduction speaker into a natural sound.
[0041]
3. In the above embodiment, for example, the measurement speakers are arranged at positions assuming all of the plurality of virtual speakers (for example, the speakers 20 (A to E)) shown in FIG. 16 in the measurement environment. An example was given in which the transfer function was measured. Therefore, it is necessary to measure the transfer function according to the number of virtual speakers, and the work efficiency may not be good.
[0042]
For example, when the position of the virtual speaker, the position of the reproduction speaker, and the three relative geometric positions of the center of the first closed curved surface are common, a known transfer function can be used.
For example, as shown in FIG. 17, the virtual speakers 35A and 35B share the above-described three positional relationships. That is, the arrangement relationship of the reproduction speakers 8p, 8a, and 8b with respect to the virtual speaker 35A is the same as the arrangement relationship of the reproduction speakers 8b, 8c, and 8d with respect to the virtual speaker 35B. Therefore, when the direction of reflected sound is not strong in the measurement environment, these transfer functions are almost the same. Thereby, regarding the matrix H shown in the above equation 3, the second column is regarded as a vector,
[Expression 4]

It is possible to transpose a matrix of known transfer coefficients.
[0043]
In FIG. 17, the virtual speaker 35C does not share the above three positional relationships with the virtual speakers 35A and 35B. Assuming that such a situation occurs, in the measurement environment 1, the number of measurement microphones 4 arranged on the first closed curved surface 10 is increased, and the number of reproduction speakers 8 on the first closed curved surface 12 is increased. Thus, it becomes possible to take correspondence of the positional relationship. Of course, it is possible to cope with the virtual speaker 35C by a method such as an interpolation process from the arrangement relationship of the reproduction speakers arranged adjacent to each other, but the above arrangement is preferable because of the simplicity of the process.
[0044]
Such a method can be made to correspond to the example shown in FIG. That is, for example, a transfer function corresponding to the output from the speaker 20A can be obtained, and this transfer function can be used as a transfer function assuming the speaker 20B, for example.
Further, in the reproduction environment 11, the position of the reproduction speaker 8 is also set to the three relative geometric positions of the position of the reproduction speaker 8 on the first closed curved surface 12, the position of the reproduction speaker 18, and the center of the first closed curved surface 12. Is common, the sound field can be approximated by appropriately transposing the values of specific rows of the matrix G in vector units. Therefore, the number of measurements in the reproduction environment 11 can be reduced.
Also in this case, the reverberation characteristics and the reproduction speakers can be independently selected by separately managing the matrix G and the matrix H so that the matrix operation can be performed as necessary. Then, with respect to the matrix G and the matrix H, a plurality of transfer functions are transposed using the position symmetry, for example, to reduce the capacity of the storage means for managing the transfer functions in the audio signal processing device. There is an advantage that can be.
[0045]
Further, as shown in FIG. 18, the reflection information from the wall of the measurement environment (for example, a hall) 1 also rotates in the same manner as the virtual speaker. Therefore, the first closed curved surface 10 on which the measurement microphones 4 (ap) are arranged is set at a position close to the central portion of the measurement environment 1 where the influence of the wall is not great and there are no obstacles nearby. It is desirable to make measurements. In this example, since the target is an audio signal composed of a plurality of channels as an input and a plurality of speakers corresponding to the audio signal, panning, in which the sound image is narrowed by using a transfer function in common. The original reverberation effect, which is supposed to be an effect of the above, will not fade.
[0046]
4). Example of Utilizing Circular Symmetry in Movement of Virtual Sound Source FIG. 19 is a schematic diagram showing an example of creating a virtual sound image that moves while rotating around a distance R_hm from the listener in the second closed curved surface 19. It is.
19, the second closed curved surface 19 is formed by, for example, eight reproduction speakers 36A, 36B, 36C, 36D, 36E, 36F, 36G, and 36H, and the inside of these reproduction speakers 36 (A to H) It becomes the listening position.
By changing the transfer function to be convoluted with respect to the audio signals supplied to the reproduction speakers 36 (A to H) arranged in this way, for example, a sound image such as a footstep can be rotated.
In the example shown in this figure, the virtual sound image is rotated to a virtual position VHm1, VHm2, VHm3,... Divided into, for example, 32 circumferences, for example, a radius R_hm that is concentric with the first closed curved surface 10. Each position. That is, in the measurement environment, the measurement microphone 4 is arranged at a position corresponding to these virtual positions VHm (1 to 32), and the transfer function H is measured. Then, by switching and using the transfer coefficient H corresponding to these 32 locations, the sound field in which the sound image is rotated by the sound output from the reproduction speakers 36 (A to H) is reproduced.
Note that, for example, a virtual position VHm1, a position with an odd number at the end is a position having the same radius as the measurement microphone 4 in the measurement environment 1 and the reproduction speaker 8 in the reproduction environment 11, for example, the virtual position A position with an even number at the end, such as VHm2, is a radial position corresponding to between the measurement microphone 4 and the reproduction speaker 8.
[0047]
The audio signal supplied to the reproduction speakers 36 (A to H) is generated by, for example, the audio signal processing device 40 shown in FIG.
The audio signal SV reproduced by the audio reproduction unit 41 is supplied to the calculation unit 43 via the measurement speaker inverse characteristic processing unit 42. The calculation units 43 (a to h) perform necessary calculation processing based on the transfer functions TA to TH for the supplied audio signal SV, and output reproduction signals SVa, SVb, SVc,. The output reproduction signals SV (a to h) are supplied to the reproduction speakers 36 (A to H).
[0048]
The transfer function T (A to H) used in the calculation unit 43 is obtained based on the transfer function H in the measurement environment 1 and the transfer function G in the reproduction environment 11. However, for the transfer function H, the symmetry of the circle of the first closed curved surface 10 can be used in the same manner as the example described in FIG. That is, the measurement speaker 3 can perform the measurement twice, for example, corresponding to the virtual positions VHm1 and VHm2, and can replace the transfer function H obtained by the measurement at these two locations. Thereby, the matrix H shown in Expression 3 can be obtained.
For the transfer function G, in a reproduction environment (not shown), a measurement microphone is arranged at a position corresponding to the reproduction speakers 36 (A to H), and arranged at a position corresponding to the virtual positions VHm (1 to 32). What is necessary is just to obtain | require based on the reproduction signal output from the speaker for reproduction made.
[0049]
The arithmetic unit 43 performs arithmetic processing on the reproduction signal SV based on the transfer function T (A to H) obtained in this way, but the sound image position controller 44 assumes virtual positions VHm1, VHm2,. By controlling the transfer function T (A to H), for example, the virtual position of the sound image can be rotated in the second closed curved surface 19 as shown in FIG.
[0050]
In the embodiment described above, an example in which the first closed curved surface 10 is circular has been described, but it is not necessarily circular. That is, it suffices if measurement can be performed with the plurality of measurement microphones 4 facing outward on the outer periphery of the closed curved surface. However, when the number of measurements is reduced using symmetry, the first closed curved surface needs to be, for example, a circle or a regular polygon.
[0051]
The number of measurement microphones 4 arranged on the first closed curved surface 10 is not limited to the number described in the present embodiment. Therefore, in order to increase the amount of measurement data (transfer function) in the measurement environment 1 and make the sound pressure distribution inside the first closed curved surface 10 as accurate as possible, arrange as many measurement microphones 4 as possible. That's fine.
[0052]
Furthermore, as for the area of the first closed curved surface 10 (radius in the present embodiment), for example, the outer periphery of the first closed curved surfaces 10 and 12 may be outside the second closed curved surfaces 14 and 19. .
[0053]
Moreover, although the measurement microphone 4 (ap) and the measurement microphone 13 (AE) have been described as directional microphones, for example, a plurality of omnidirectional microphones are used even when they are not directional microphones. Therefore, it is sufficient that a corresponding effect can be obtained by delaying or subtracting the output.
[0054]
Further, with respect to the matrix G2, amplitude and delay characteristics may be used based on a geometric composition without performing measurement in the reproduction environment 11.
In addition, when the first closed curved surface 10 is configured in a circular shape as described in the present embodiment, a transfer function on one side that is symmetric about a certain axis in the first closed curved surface 10 may be diverted to the other side. Is possible. For example, as shown in FIG. 1, when the transfer function for the measurement microphone 4b is Hb, the transfer function Hb is arranged in a symmetrical position with the measurement microphone 4a as an axis. It becomes possible to make common.
Similarly, in the second closed curved surface 14, for example, as shown in FIG. 4, when the transfer function G2A from the reproduction speaker 8b to the measurement microphone 13A is used, the transfer function G2A is transferred from the reproduction speaker 8p to the measurement microphone. It can be shared with 13A.
[0055]
In the above embodiment, the first closed curved surface and the second closed curved surface are both assumed to be on a flat surface, for example, but by placing a measurement microphone on a three-dimensional closed curved surface, a three-dimensional sound field can be obtained. Can also be reproduced.
[0056]
Furthermore, when measuring the impulse response and when determining the transfer function, the sound pressure distribution approximation in the first closed curved surface and the second closed curved surface is approximated by the sound pressure distribution using the virtual speaker as the sound source in the target sound field. In order to make it closer, compensation processing indicated by a square matrix may be further applied to the matrix shown in the above equation 1. This compensation processing is processing for compensating for the directivity of the speakers and microphones when using an insufficient number of speakers and microphones to approximate the sound pressure distribution.
[0057]
5). Next, an example in which the present invention is applied to a car audio system will be described.
In this example, as shown in FIG. 21, speakers 52 </ b> L and 52 </ b> R that are assumed to be arranged in front of the automobile 50, for example, by reproduction speakers 51 (a to j) arranged in the room of the automobile 50. An object of the present invention is to provide a listener in a room with an environment in which stereo sound is reproduced (corresponding to a measurement speaker described later).
For this reason, in the measurement environment 53 shown in FIG. 22, a solid (in this embodiment, a rectangular parallelepiped) that can surround the appearance of the automobile 50 is assumed, and this solid is defined as the first closed curved surface 55. For example, 21 measurement microphones 56a to 56u are arranged on the outer periphery. The main axis of directivity of the measurement microphone 56 (a to u) diverges outward in the normal direction from the center of the first closed curved surface 55 as indicated by an arrow on the surface forming the first closed curved surface 55. It corresponds to the extending direction. Note that if the measurement microphone 56 (a to u) is arranged on a line that touches two or more faces, or a point if the placement position of the measurement microphones 56 (a to u) is a point, The direction of the main axis of directivity will be determined.
[0058]
The measurement speakers 52L and 52R are arranged at positions assuming the front of the automobile 50 with respect to the first closed curved surface 55, the measurement speaker 52L for measurement sound for left sound, and the measurement speaker 52R for right sound. Can be used to output measurement audio. Then, the transfer function H to the measurement microphone 56 (a to u) is obtained and held by the measurement sound output from the measurement speakers 52L and 52R.
That is, the transfer function from the measurement speaker 52L to the measurement microphone 56 (au) is HLa to HLu, and the transfer function from the measurement speaker 52R to the measurement microphone 56 (au) is HRa to HRu. Become.
[0059]
FIG. 23 is a schematic diagram for explaining an example of reproducing the measurement environment 53 in a reproduction environment 60 such as an anechoic room.
The first closed curved surface 61 formed in the illustrated reproduction environment 60 corresponds to the first closed curved surface 55 in the measurement environment 53, and is assumed to be an equivalent solid. The reproduction speakers 62a to 62u are arranged on the first closed curved surface 61, and the arrangement positions thereof are positions corresponding to the measurement microphones 56 (a to u) on the first closed curved surface 55. In this figure, only the positions of the reproduction speakers 62 a to 62 u are shown, but they are arranged so as to face the center of the first closed curved surface 61. Then, by supplying the reproduction speaker 62 (a to u) with a reproduction signal subjected to arithmetic processing based on the transfer functions HLa to HLu or the transfer functions HRa to HRu, any listening in the first closed curved surface 61 is heard. A sense of sound field equivalent to the outside of the first closed curved surface 55 can be reproduced outside the position.
[0060]
In the first closed curved surface 61, measurement microphones 65 (a to j) are arranged at positions corresponding to the reproduction speakers 51 (a to j) in the automobile 50. A closed curved surface formed by connecting the arrangement positions of these measurement microphones 65 (a to j) becomes the second closed curved surface 70. The main axes of directivity of these measurement microphones 65 (a to j) are the normal lines on the surface forming the second closed curved surface 70, similarly to the measurement microphones 56 (a to u) shown in FIG. Corresponds to the extension direction that diverges outward.
In this manner, the reproduction environment 60 is constructed, and the reproduction microphones 65 (a to j) output the reproduction signals subjected to the arithmetic processing based on the transfer functions HLa to HLu from the reproduction speakers 62 (a to u). By measuring, transfer functions GLa to GLj corresponding to the measurement speaker 52L can be obtained. Similarly, the reproduction speaker 60 (a to u) outputs a reproduction signal that has been subjected to arithmetic processing based on the transfer functions HRa to HRu, and measures it with the measurement microphone 65 (a to j), thereby measuring the measurement speaker 52R. The transfer functions GRa to GRj corresponding to can be obtained.
[0061]
FIG. 24 shows an audio signal processing device 80 that generates a reproduction signal to be supplied to the reproduction speakers 51 (a to j) installed in the room of the automobile 50 based on the transfer functions obtained in the measurement environment 53 and the reproduction environment 60. It is a figure explaining the example of a structure. The audio signal processing device 80 is configured as a part of a car audio system.
From the audio reproduction unit 81, for example, audio signals SL and SR, which are 2-channel stereo, are output. These audio signals SL and SR are supplied to the calculation unit 83 (83L and 83R) via the measurement speaker inverse characteristic processing units 82L and 82R, respectively. The arithmetic unit 83L performs arithmetic processing based on the transfer functions HLa to HLu on the audio signal SL, and the arithmetic unit 83R performs arithmetic processing based on the transfer functions HRa to HRu.
[0062]
The audio signals that have been subjected to the arithmetic processing based on the transfer functions HLa to HLu and HRa to HRu by the arithmetic units 83L and 83R are supplied to the arithmetic units 85 (85L and 85R), respectively. In the arithmetic unit 85L, the transfer functions GLa to GLj In addition, the calculation process based on the transfer functions GRa to GRj is performed in the calculation unit 85R.
Then, the audio signals subjected to the arithmetic processing based on the transfer functions GLa to GLj and GRa to GRj by the arithmetic units 83L and 83R are added by the adding unit 87 and supplied to the reproduction speakers 51 (a to j). Reproduction signals SCa, SCb, SCc,..., SCj are generated. In the adding process in the adding unit 87, the audio signals subjected to the arithmetic process based on the transfer function with the same subscript (lower case alphabet) are added.
[0063]
By supplying the audio signals SC (a to j) obtained by performing such arithmetic processing to the reproduction speakers 51 (a to j), the interior of the automobile 50 shown in FIG. A virtual sound image can be perceived at the positions of the virtual speakers 51L and 51R that are not present.
[0064]
The sound field of the virtual sound image can be selected by measuring and storing the transfer coefficients HLa to HLu used for calculation in the calculation unit 83 in a plurality of measurement environments such as famous concert halls and movie theaters. Is also possible. In this case, a sound field controller 84 may be provided for the calculation unit 83, and a transfer function stored in the memory 84a may be selected and used for calculation processing as necessary.
Similarly, the transfer coefficient GLa to GLu used for calculation in the calculation unit 85 is measured and stored in correspondence with the arrangement state (number of arrangement, arrangement position, etc.) of a plurality of types of reproduction speakers, thereby reproducing It is also possible to correspond to the arrangement state of the loudspeakers 51 (a to j). In this case, a playback speaker setting controller 86 may be provided for the calculation unit 85, and a transfer function stored in the memory 86a may be selected and used for calculation processing as necessary. As a result, it is possible to generate the reproduction signals SC (a to j) corresponding to the plurality of types of arrangement states of the reproduction speakers 51 (a to j) in the room of the automobile 50.
Such a transfer function selection may be provided with an operation unit 88 as selection means, and may be selected by a viewer, for example.
[0065]
In this way, the reverberant sound field characteristics of the measurement environment and the playback speaker placement state can be set independently, so even if a new speaker placement state is provided by a new vehicle model, the reproduction environment such as an anechoic room In this case, a transfer function corresponding to the arrangement state may be obtained.
Accordingly, the transfer function stored in the memories 84a and 86a can be updated. As an updating method, for example, a transfer function recorded on a recording medium such as a disc may be read, or for example, a necessary communication means may be provided and downloaded via a network. It may be. The memories 84a and 86a may store impulse response data corresponding to each transfer function.
[0066]
The present invention can be applied to virtual reality industrial equipment such as an amusement or game that gives a sense of realism or immersion, for example, in addition to reproducing sound in a home room or a car audio system, for example. .
[0067]
【The invention's effect】
As described above, the present invention obtains the first transfer function based on the sound field of the measurement environment, and further reproduces the audio signal subjected to the arithmetic processing based on the first transfer function in the reproduction environment. Thus, the sound field (for example, reverberation, sound image localization) of the measurement environment can be reproduced in the reproduction environment. Furthermore, in the reproduction environment, a second transfer function corresponding to a reproduction speaker arranged in the reproduction environment under a sound field condition equivalent to that of the measurement environment can be obtained.
Accordingly, since the second transfer function can be obtained in a reproduction environment, the number of measurements in the measurement environment can be reduced. Furthermore, the second transfer function can be obtained in the reproduction environment even when dealing with a new arrangement state of the reproduction speaker.
[0068]
In addition, the first transfer function can be transposed according to a predetermined rule using the isotropic orientation and symmetry of the first closed curved surface for obtaining the first transfer function. Yes. In other words, a transfer function obtained from a virtual speaker assuming a certain position viewed from the first closed curved surface can be used as a transfer function obtained from a virtual speaker assuming another position. Accordingly, the number of measurement operations for obtaining the first transfer function can be reduced.
[0069]
In addition, since the first transfer function and the second transfer function can be managed individually, the sound field of the measurement environment and the arrangement state of the reproduction speakers can be selected independently. Therefore, it is possible to individually update and add the sound field of the measurement environment and the arrangement state of the reproduction speakers.
Furthermore, when the first transfer function and the second transfer function are individually managed, the first and second transfer functions can be used by transposing them according to a predetermined rule. The capacity of the storage means for storing the first and second transfer functions can be reduced.
[Brief description of the drawings]
FIG. 1 is a diagram for explaining a measurement environment in an audio reproduction system according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating a reproduction sound reproduction system in a reproduction environment in the sound reproduction system according to the present embodiment.
FIG. 3 is a diagram illustrating a reproduction environment.
FIG. 4 is a diagram illustrating an example in which an impulse response is measured inside a first closed curved surface of a reproduction environment.
FIG. 5 is a diagram illustrating an example in which an impulse response is measured inside a first closed curved surface of a reproduction environment.
FIG. 6 is a diagram for explaining a transfer function of a virtual speaker in a reproduction environment.
[Fig. 7] Fig. 7 is a diagram for describing a 5-channel stereo sound playback system in a playback environment.
FIG. 8 is a diagram illustrating the arrangement of playback speakers in a playback environment.
FIG. 9 is a diagram illustrating a reverberant sound field and sound image localization when the listening position is in the second closed curved surface in a reproduction environment.
FIG. 10 is a diagram for explaining an example of obtaining a transfer function G in a first closed curved surface of a reproduction environment.
FIG. 11 is a flowchart illustrating a procedure for reproducing a sound field of a measurement environment in a reproduction environment.
FIG. 12 is a flowchart illustrating a procedure for performing matrix calculation by separately managing a transfer function H and a transfer function G.
FIG. 13 is a diagram for explaining a reproduction environment of a 5-channel stereo.
FIG. 14 is a diagram illustrating a configuration example of an audio signal processing device that performs arithmetic processing based on a transfer function T (A to E) on a 5-channel stereo signal D (A to E).
FIG. 15 is a diagram illustrating a transfer function from a speaker to a measurement microphone in a movie theater as a measurement environment.
FIG. 16 is a diagram for explaining reverberation and sound image localization in a measurement environment reproduced in a reproduction environment.
FIG. 17 is a diagram illustrating the positional relationship between a virtual speaker and a reproduction speaker.
FIG. 18 is a diagram illustrating rotation of reflection information from a measurement environment wall.
FIG. 19 is a schematic diagram illustrating an example of creating a virtual sound image that moves within a second closed curved surface.
FIG. 20 is a diagram illustrating a configuration example of an audio signal processing device that generates an audio signal that moves a virtual sound image;
FIG. 21 is a diagram for explaining a reproduction speaker arranged in a room of an automobile.
FIG. 22 is a diagram showing a first closed curved surface assuming a solid that can surround, for example, the appearance of an automobile in a measurement environment.
FIG. 23 is a diagram for explaining an example of reproducing the measurement environment shown in FIG. 22 in a reproduction environment.
FIG. 24 is a diagram illustrating a configuration example of an audio signal processing device that generates a reproduction signal to be supplied to a reproduction speaker installed in an automobile room based on a transfer function.
FIG. 25 is a diagram for explaining a sampling reverb system.
FIG. 26 is an explanatory diagram of measurement sound output from a measurement speaker.
FIG. 27 is a diagram illustrating an example in which an audio signal is generated using a transfer function obtained in a measurement environment.
[Explanation of symbols]
1, 55 measurement environment, 2 measurement signal reproduction means, 3, 52 (L, R) measurement speaker, 4 (ap), 13 (AE), 56 measurement microphone, 5, 15, 22, 40 , 80 Audio signal processing device, 6, 16, 23, 41, 81 Audio reproduction unit, 7 (ap), 17 (ae), 24, 26, 28, 30, 32, 43 (ah) , 83 (83L, 83R), 85 (85L, 85R) arithmetic unit, 8 (ap), 36, 62 reproduction speaker, 10, 12, 61 first closed surface, 11, 60 reproduction environment, 13 (A To E), 65 Measuring microphone, 14, 19, 70 Second closed surface, 18, 51 Playback speakers (AE), 20 Speakers, 21 Movie theater, 25, 27, 29, 31, 33, 87 Addition , 35 Virtual speaker, 42 82 (L, R) Measurement speaker inverse characteristic processing unit, 44 Image localization controller, 50 motor vehicles, 84 sound field controller, 84a, 86a memory, speaker setting controller for playback

Claims (2)

Outputting measurement sound from a speaker arranged at a virtual sound source position, detecting the measurement sound by each of a plurality of first directional microphones arranged outward on a first closed curved surface, First storage means for storing a first transfer function which is a transfer function obtained by the detected measurement voice;
Each of a plurality of speakers arranged inward on the first closed curved surface outputs measurement sound and faces outward on a second closed curved surface disposed inside the first closed curved surface. A second transfer function which is a transfer function obtained by detecting the measurement sound by each of a plurality of second directional microphones arranged in the first and second measurement microphones and storing the second transfer function. Two storage means;
Arithmetic processing means for multiplying the input transfer signal by the first transfer function and the second transfer function;
A plurality of speakers which are each arranged at the same position as the position on the second closed curved surface where each of the plurality of second directional microphones is arranged and to which an audio signal obtained by the arithmetic processing means is supplied When,
An audio signal processing apparatus comprising:   The audio signal processing device according to claim 1, wherein the first closed curved surface is circular.

Images