Spectral properties of fricatives: a forensic approach

Abstract

This paper reports on the acoustic-phonetic analysis of the voiceless fricatives/s, ∫, f, θ/taken from high-quality recordings of six native British English speakers reading phonetically controlled stimuli under various face disguise conditions. Speech samples were extracted from an audio-visual 'face cover' corpus that was collected for the purpose of investigating multimodal speech and speaker recognition in a forensic context. Findings are discussed with regard to constraints in speech production and acoustic damping effects caused by certain mask materials.