Who Are You Listening to? Towards a Dynamic Measure of Auditory Attention to Speech-on-speech - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2018

Who Are You Listening to? Towards a Dynamic Measure of Auditory Attention to Speech-on-speech

Résumé

When studying speech-on-speech perception, even when participants are explicitly instructed to focus selectively on a single voice, they can spuriously find themselves listening to the wrong voice. These paradigms generally do not allow to infer, retrospectively, which of the speakers was listened to, at different times during presentation. The present study sought to develop a psychophysical test paradigm, and a set of speech stimuli to that purpose. In this paradigm, after listening to two simultaneous stories, the participant had to identify, among a set of words, those that were present in the target story. Target and masker stories were presented dichotically or diotically. F0 and vocal-tract length were manipulated in order to parametrically vary the distance between the target and masker voices. Consistent with the hypothesis that correct-identification performance for target words depends on selective attention, performance decreases with the distance between the target and masker voices. These results indicate that the paradigm and stimuli described here can be used to infer which voice a participant is listening to in concurrent-speech listening experiments.
Fichier principal
Vignette du fichier
Huet_etal_2018_Interspeech.pdf (355.07 Ko) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte
Loading...

Dates et versions

hal-02385549 , version 1 (28-11-2019)

Identifiants

Citer

Moïra-Phoebé Huet, Christophe Micheyl, Etienne Gaudrain, Etienne Parizet. Who Are You Listening to? Towards a Dynamic Measure of Auditory Attention to Speech-on-speech. Interspeech 2018, Sep 2018, Hyderabad, India. pp.2272-2275, ⟨10.21437/Interspeech.2018-2053⟩. ⟨hal-02385549⟩
97 Consultations
149 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More