EEG measurement of binaural sound immersion - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

EEG measurement of binaural sound immersion

Pascal Rueff
  • Fonction : Auteur
Nicolas Farrugia

Résumé

The purpose of this study is to examine whether brain imagery could provide new insights into the assessment of the listening experience of spatial audio content. Conventional methods of assessing sound reproduction are based on either quality or localization judgments. Their main disadvantage is that the subject is aware of his(her) rating task. In addition, dimensions such as emotions or cognitive load are generally not taken into account. Therefore, our objective is to explore cognitive processes involved when listening to spatial audio content by electroencephalography (EEG) measurement. Descriptors of the listening experience (e.g. immersion, realism) are sought in the electrical activity of the brain.The experiment is based on a reversed implementation of the oddball paradigm. The participant is presented a sequence of visual stimuli (320), which is composed of standard (265) and deviant (55) stimuli. These latter are introduced to cause a surprise effect, that will induce an event-related potential (ERP), known as the « P300 » in the brain. The participant is asked to click whenever he(she) detects a deviant stimulus. Audio stimuli are used to interfere with the detection task. The objective is to observe the effect of such distractors on the P300 response as a function of the type of sound spatialization. Binaural and stereophonic reproductions are compared. It is expected that the distracting effect of binaural reproduction is stronger than that of stereophony. Evidence of this will be sought in both behavioral data (i.e. the performances in detecting deviant stimuli) and EEG measurements. During the experiment, an EEG net of 256 electrodes was placed on the participant’s head to record his(her) brain activity. In addition, a headphone (Sennheiser HD650) was arranged over it to reproduce sound scenes that have been previously recorded in the experimental room. The pretext for justifying the use of a headphone was a listening session as a further step in the experiment. The participant was thus not aware that virtual sounds were going to be played by headphones, and that his(her) listening experience was under study, which was one novelty of this study. Besides, at no time was he(she) asked direct questions about the quality of sound reproduction.The sound scenes contained ambient noise in combination with isolated sounds potentially emanating from the experimenter (e.g. falling keys, running water from a tap, leafing through a book, …), but they were totally free from speech and music signals. The listening level was adjusted to 60 dBA. The oddball experiment was repeated twice: once with the binaural recording (by a dummy head Neumann KU 100) and the other time with the stereophonic recording (by a XY pair). The presentation order was randomized. Intensity stereophony was chosen to provide the highest contrast with binaural spatialization which uses both interaural differences of time and level, as well as spectral cues. The binaural and stereophonic scenes were similar but not identical to prevent any habituation effect.At the beginning of the experiment, the participant was undergoing a training phase for the visual detection task. At the end, he(she) had to complete a questionnaire about the task difficulty, the level of immersion. It was also checked whether he(she) realized that sounds were coming from the headphone. 19 subjects took part into the experiment. EEG measurements were first pre-processed (elimination of electrodes with anomalies, denoising). Only 124 electrodes were retained. The data set consisted of 6080 trials, 1045 of which were deviants. The cortical electrical activity was reconstructed from surface signals. Event-Related Potentials (ERP) were averaged per electrode and per subject for each condition. The differential response between standard and deviant stimuli was examined as a function of the audio distractor (binaural or stereophonic). Cluster of adjacent electrodes with similar differential responses were then clustered. Two Regions Of Interest (ROI) were identified in the frontal and the parietal areas. It was observed that the P300 amplitude was significantly higher in presence of the binaural distractor than for the stereophonic one, suggesting a stronger effect of surprise possibly caused by a more immersive reproduction. This was confirmed by behavioral results which showed a longer response time for binaural distractors (566 ms) than for stereophonic ones (550 ms). However, the detection error rate was not significantly different between the two conditions.
Fichier principal
Vignette du fichier
000021.pdf (2.92 Mo) Télécharger le fichier
Origine : Publication financée par une institution
Loading...

Dates et versions

hal-02275183 , version 1 (30-08-2019)

Identifiants

Citer

Rozenn Nicol, Olivier Dufor, Laetitia Gros, Pascal Rueff, Nicolas Farrugia. EEG measurement of binaural sound immersion. EAA Spatial Audio Signal Processing Symposium, Sep 2019, Paris, France. pp.73-78, ⟨10.25836/sasp.2019.21⟩. ⟨hal-02275183⟩
440 Consultations
205 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More