The PerspectiveLiberator An Upmixing 6DoF Rendering Plugin for Single-Perspective Ambisonic Room Impulse Responses Kaspar M uller Franz Zotter

2025-05-06 0 0 496.95KB 4 页 10玖币

侵权投诉

The PerspectiveLiberator – An Upmixing 6DoF Rendering Plugin

for Single-Perspective Ambisonic Room Impulse Responses

Kaspar M¨uller, Franz Zotter

Institute of Electronic Music and Acoustics, University of Music and Performing Arts Graz, Austria

Email: kaspar.mueller@outlook.de, zotter@iem.at

Introduction

Virtual reality interfaces allow the user to interactively

change the position and look direction in six degrees of

freedom (6DoF), and consistently with the visual part, the

acoustic perspective needs to be updated. For audio ren-

dering based on directional room impulse responses, the

spatial impulse response rendering (SIRR) [

] and (Am-

bisonic) spatial decomposition method (SDM) [

], as

well as higher-order SIRR (HO-SIRR) [

] were helpful to

obtain rendering of suﬃcient resolution when using single-

perspective room impulse responses. We can assume that

these methods accomplish a reliable single-perspective

rendering with dynamic head rotation. Existing 6DoF

audio rendering approaches using Ambisonic room im-

pulse responses (ARIRs) or similar are, e.g., interpola-

tion of B-format signals [

], Vienna Multi Impulse

Response (MIR)

, time warping [

], the 4-pi sampling

reverberator [

], and higher-order parametric perspective

extrapolation [10].

Based on the ARIR extrapolation in [

], our contribution

presents a plug-in meant to free the virtual perspective

from the measured one: The PerspectiveLiberator. Fig-

ure 1 shows the two processing stages:

Oﬄine pre-processing detects and localizes sound events

corresponding to the most distinct early ARIR peaks, and

it removes according segments from the residual ARIR.

The spatial resolution of such peak segments is increased

by 4DE upmixing [

] to higher order Ambisonics (HOA).

Real-time processing extrapolates the ARIR by time-,

level- and directional-aligned translation of the sound-

event segments according to the desired listener perspec-

tive and delays the residual ARIR to arrive after the

direct sound peak (ﬁrst sound event). Finally, the HOA

output signal is obtained by convolution-based rendering.

Input

ARIR

ARIR Parameter

Estimation

4DE Directional

(for ﬁrst-order input)

Sound-Event

Localization

Detection and

Enhancement

Translation of

Residual ARIR

Input

Signal

Real-Time Processing Oﬄine Pre-Processing

h(t)

χ(t)

Listener

Perspec-

tive

x(t)

θ(t)

hr(t)

h(n)

s(t)

x(t)

h(n)

s(t)˜

hr(t)

{ˆ

x, T, θ}(n)

¯a0.5(t)

h(n)

s(t)Directional

Segmentation

Sound-Event

Translation of

Sound Events

Directional

Real-Time

Rendering

χ(t)

Figure 1: Signal ﬂow graph of the PerspectiveLiberator.

Figure 2:

User interface of the PerspectiveLiberator plug-in.

First-Order ARIR Parameter Estimation

As shown in Figure 1, the estimation of ARIR parameters

is the initial step of the pre-processing stage. Since most

recorded ARIRs are ﬁrst-order, we perform the parameter

estimation based on ﬁrst-order ARIR data only, even

though any higher-order input is possible. The two basic

analysis features needed for the subsequent processing are

the direction of arrival and the short-time amplitude.

Direction of Arrival

We compute a Cartesian direction of arrival (DOA) vector

for each ﬁrst-order input ARIR sample from the time aver-

aged pseudo intensity vector (PIV) after band limitation

θ(t) = ˜

I(t)

k˜

I(t)k,with ˜

I(t) = Fav(WBP(t)"XBP (t)

YBP(t)

ZBP(t)#),(1)

where

WBP

(

) is the omnidirectional, and

{X, Y, Z}BP

(

)

are the ﬁrst-order directional signals of

(

), zero-phase

bandpass ﬁltered between 200 Hz and 3 kHz.

Fav {·}

is a

zero-phase averaging ﬁlter over 0.25 ms.

Short-Time Amplitude

To obtain a meaningful ARIR amplitude estimate, which

is especially needed for a robust peak detection, we pro-

pose to incorporate the directional ARIR content repre-

sented by the broadband PIV magnitude, instead of solely

relating to the omnidirectional amplitude signal. Time

averaging of the PIV with a Hamming-windowed moving-

average ﬁlter over 0.5 ms, denoted by FH{·}, yields

¯a0.5(t) = qFHkI(t)k,with I(t) = W(t)"X(t)

Y(t)

Z(t)#.(2)

1https://www.vsl.co.at/de/Vienna_Software_Package/

Vienna_MIR_PRO

arXiv:2210.03360v2 [cs.SD] 10 Oct 2022

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

ThePerspectiveLiberator{AnUpmixing6DoFRenderingPluginforSingle-PerspectiveAmbisonicRoomImpulseResponsesKasparMuller,FranzZotterInstituteofElectronicMusicandAcoustics,UniversityofMusicandPerformingArtsGraz,AustriaEmail:kaspar.mueller@outlook.de,zotter@iem.atIntroductionVirtualrealityinterfacesallowt...

展开>> 收起<<

The PerspectiveLiberator An Upmixing 6DoF Rendering Plugin for Single-Perspective Ambisonic Room Impulse Responses Kaspar M uller Franz Zotter.pdf

共4页,预览1页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

The PerspectiveLiberator An Upmixing 6DoF Rendering Plugin for Single-Perspective Ambisonic Room Impulse Responses Kaspar M uller Franz Zotter

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: