Validation of a loudspeaker-based room auralization system using speech intelligibility measures

Sylvain Favrot*, Jörg M. Buchholz

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference proceeding contributionpeer-review

5 Citations (Scopus)

Abstract

A novel loudspeaker-based room auralization (LoRA) system has been proposed to generate versatile and realistic virtual auditory environments (VAEs) for investigating human auditory perception. This system efficiently combines modern room acoustic models with loudspeaker auralization using either single loudspeaker or high-order Ambisonics (HOA) auralization. The LoRA signal processing of the direct sound and the early reflections was investigated by measuring the speech intelligibility enhancement by early reflections in diffuse background noise. Danish sentences were simulated in a classroom and the direct sound and each early reflection were either auralized with a single loudspeaker, HOA or first-order Ambisonics. Results indicated that (i) absolute intelligibility scores are significantly dependent on the reproduced technique and that (ii) early reflections reproduced with HOA provide a similar benefit on intelligibility as when reproduced with a single loudspeaker. It is concluded that speech intelligibility experiments can be carried out with the LoRA system either with the single loudspeaker or HOA technique.

Original languageEnglish
Title of host publicationAudio Engineering Society Convention Papers
Place of PublicationNew York, NY
PublisherPRAESENS VERLAGSgesmbH
Pages579-587
Number of pages9
Volume2
ISBN (Print)9781615671663
Publication statusPublished - 2009
Externally publishedYes
Event126th Audio Engineering Society Convention - 2009 - Munich, Germany
Duration: 7 May 200910 May 2009

Other

Other126th Audio Engineering Society Convention - 2009
Country/TerritoryGermany
CityMunich
Period7/05/0910/05/09

Fingerprint

Dive into the research topics of 'Validation of a loudspeaker-based room auralization system using speech intelligibility measures'. Together they form a unique fingerprint.

Cite this