東京藝術大学 大学院 音楽音響創造

要旨

人間は耳から聞いた音によって、音の方向や距離などの情報を得ることができる。本研究では音源の移動によると、知覚された音像の動きと実際の動きとの偏差について、2つの実験を行い調査した。

実験1は非個人頭部伝達関数(HRTF)を使用して、ホワイトノイズを用いて、前方180度範囲内、同じ速度で8種類の移動パターンを提示し、ヘッドホンによる再生で、移動音源の定位を判断した。

実験2-1は22.2chシステムを使用して、3Dパンニングでホワイトノイズの移動音源を作成し、水平面の360度内に上下二層のレイヤー(6×2=12種類)の移動音源の定位について調べた。

実験2-2aでは22.2chシステムで再生した移動音源をIn-Earマイクロホンで個人別に収録して、ヘッドホンから再生して、移動音源の定位を回答した。

実験2-2bは個人ごとに録音した上下の移動も含む音源を用いて、上下の移動定位を判断した。

以上の実験から以下の結果が得られた。

非個人のHRTFを用いた場合は、個人のHRTFを用いた場合と比べて、全体的に移動の終点の偏差が大きくなる。特に前方0度経由する場合、偏差が一番大きい。移動範囲の幅に関して、前方0度経由する場合、提示した音源より幅広く回答された。左右90度は移動の終点の場合、提示した音源より幅狭く回答された。

個人のHRTFを用いた場合、中層の移動は上層より、偏差が小さい。そして、移動範囲の幅の偏差について、前方0度を起点として移動するパターンの偏差はほかの移動パターンと比べて、偏差が小さい。それ以外のパターンは、中層、上層ともに偏差が大きくなる。

上下の移動について、起点、終点の偏差とも、移動ありのパターンは移動方向にずれている傾向があって、幅の偏差に関して、上下移動ありの移動パターンは幅狭く回答されて、上下固定の移動パターンは広く回答される傾向があった。

これで、上下移動の判断は困難であると言える。それは、上下移動と水平移動両方を含んでいるため、感知が混乱していたと考えられる。


We can obtain auditory information such as directions and distances of the sounds source we heard. In this study, two experiments according to the movement of the sound source were conducted to investigate the deviation between the movement of the perceived sound image and the actual motion.

In Experiment 1, using the non-individual head related transfer function (IRTF), white noise signal were presented in eight kinds of movement patterns in the range of 180 degree of frontal direction of horizontal plane at the same speed. Localizations of moving sound sources, which were reproduced by a headphone, were judged by participants.

In Experiment 2-1, moving sound sources of white noise were created by 3D panning on a 22. 2 channel system, and the localizations of the moving sound sources in the upper and middle layers within 360 degrees range of the horizontal plane (6 patterns X 2 layers = 12 types) were investigated.

In Experiment 2-2а, the moving sound sources reproduced by the 22.2 channel system were individually recorded by an In-Ear microphone, then the recorded signals were reproduced by headphone, and the localization of the moving sound source were judged.

In Experiment 2-2b, using individually recorded sound sources in which movements across upper and middle layers were also included, localizations of movements in the vertical direction were judged.

From the above experiments the following results were obtained. When non-individual HRTF was used, the error of localization of the end point of the movement was larger comparing to the case using the individual HRTF. Especially for the movement passing through 0 degree azimuth, the deviation was the largest. Regarding of the moving range, for the movement passing through 0 degree azimuth, the range was evaluated wider than the actual range of the presented sound source. When endpoints of movements were 90 or -90 degrees azimuth, the range were answered narrower. When using an individual’s HRTF, the deviation of the middle layer movement was smaller than that of the upper layer. In addition, for the deviation of movement range, the deviation of the pattern in which the movement started from the 0 degree azimuth was smaller than deviations of other movement patterns.

For movements across upper and middle layers, both the error of localization of the start point and the end point tended to shift to the movement direction. For the deviation of the width, the movement patterns with the vertical movement were evaluated narrower and the patterns without the vertical movement tended to be evaluated wider.

It indicated that it was difficult to judge the vertical movement. Since it included both vertical and horizontal movement, the perception could be confused.