The Role of Harmonicity in the Masking of Speech: Behavioural and Modelling results - Forum Acusticum 2020 Accéder directement au contenu
Communication Dans Un Congrès Année : 2020

The Role of Harmonicity in the Masking of Speech: Behavioural and Modelling results

Résumé

Harmonic tone complexes are much less effective maskers of speech than a noise with the same overall spectral envelope. Three main factors are thought to contribute to this effect: Firstly, harmonic complexes allow ?glimpsing? into frequency regions between harmonics where there is little masker energy. Secondly, the periodicity of the complexes allows them to be more effectively segregated from the target speech. Thirdly, tone complexes contain far fewer envelope modulations than noise, resulting in reduced modulation masking. To estimate their respective contributions, complexes which varied regarding all three factors were devised. Data obtained from normal-hearing listeners revealed that speech reception thresholds (SRTs) were generally worse for complexes with dynamic as opposed to static F0 contours and improved with increasing masker F0, providing evidence for spectral glimpsing. In contrast, inharmonic complexes eliciting a weaker pitch percept led to similar SRTs than harmonic complexes, suggesting that periodicity as such does not aid the segregation of speech and masker. Modulation masking, on the other hand, was found to be an important factor, as SRTs were consistently worse for complexes that contained a broader range of envelope modulations. The predictions of a modulation-based speech intelligibility model, the sEPSM<sup>corr</sup>2, were able to capture the effects of spectral glimpsing and modulation masking. However, the model generally overestimated the masking effectiveness of the tone complexes, as it does not take pitch-related effects that contribute to increased stream segregation relative to noise maskers into account.
Fichier principal
Vignette du fichier
000080.pdf (439.14 Ko) Télécharger le fichier
Origine : Accord explicite pour ce dépôt

Dates et versions

hal-03234174 , version 1 (26-05-2021)

Identifiants

Citer

Kurt Steinmetzger, Stuart Rosen. The Role of Harmonicity in the Masking of Speech: Behavioural and Modelling results. e-Forum Acusticum 2020, Dec 2020, Lyon, France. pp.3487-3488, ⟨10.48465/fa.2020.0080⟩. ⟨hal-03234174⟩

Collections

FA2020
34 Consultations
19 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More