Decomposition of high order Ambisonic recordings for reproduction using irregular loudspeaker arrays

摘要

多チャネルスピーカを用いた高次アンビソニックに基づく音場再現法はスピーカアレイの中心付近において高精細な音場再現を実現でき，注目を集めている．しかし，アンビソニックエンコーディングおよびデコーディングは球面調和解析に基づくため，スピーカ配置が3次元的に均一でない場合は十分なデコーディング性能が得られないという問題がある．そこで本報告では，不均一なアレイにおいてもデコーディング行列の悪条件を回避し，高精細なデコーディングが可能な新たなデコーディング方式を提案する．我々の構築した包囲型157chスピーカアレイの配置を用いた計算機シミュレーションにより，提案法の有効性を示す．%Ambisonics is a technique for sound field synthesis and reproduction that has shown promising results for the presentation of three dimensional sound spaces. Sound fields are naturally represented by their expansions in terms of spherical harmonic functions; such descriptions are known as Ambisonic encodings. Precise regeneration of the sound field from a set of spherical harmonic expansion coefficients requires a sufficient number of loudspeakers arranged in a regular layout. Decoding of Ambisonic recordings for their reproduction using a spherical, uniform loudspeaker array is not difficult. Nevertheless, evenly distributing a large number of loudspeakers is unfeasible in most scenarios. Irregular configurations are known to lead to ill-conditioned and singular re-encoding matrices. We propose a method to decompose high order Ambisonic re cordings, based on the geometry of the target array, for accurate reproduction over irregular loudspeaker arrays.