Beiträge der auditiven Gruppierung zur Sprachwahrnehmung
Zusammenfassung der Projektergebnisse
This project investigated the contribution of spatial, pitch, and timbre cues to our ability to understand speech. A specific focus was on obtaining a better understanding of the nature by which spatial cues contribute to speech intelligibility both in anechoic and echoic conditions. A well established approach to modelling the spatial contribution to speech intelligibility is based on Equalization-Cancellation modelling which allows target speech that would be masked without the presence of spatial cues, to become audible and useful for understanding speech. It was found that the Equalization Cancellation approach alone is not sufficient for explaining the full benefit of spatial processing. In addition, spatial cues also seem to support our ability to better select those audible speech glimpses that belong to the target speech. Using speech stimuli that were partially spatially diffuse, it was nevertheless found that the observed suppression of interfering sources was compatible with Equalization Cancellation processing. As part of this study, spatial diffuseness was generated with a cascade of all-pass filters. In a control condition, it was found that even applying these filters on speech signals without altering spatial properties, an impairment in intelligibility resulted. This was an unexpected result, which we attribute to the all-pass filters causing a reduction in the salience of pitch cues encoded in the temporal envelope of speech stimuli. A further study investigated the role of late reverberation in spatial speech intelligibility using a scenario with one reverberated target speaker and two anechoic interfering speakers. When all speakers were separated, a reduction in intelligibility was found when the length of the reverberant tail was increased, in line with expectations. Interestingly, for co-located speakers, intelligibility improved. This was an unexpected and interesting result that was attributed to the increasing spatial diffuseness of the target speech that increased with longer reverberation, which may have helped to better segregate the target from the anechoic interfering speakers. Further studies explored metrics to predict speech intelligibility based on energetic masking and the distribution of pitch cues in the target versus interfering speaker. Some success was achieved in combining these metrics to make per-speech token ‘microscopic’ predictions of speech intelligibility. One collaborative study investigated the ability to segregate musical voices in Bach's The Art of the Fugue. It was found that timbre differences between voices helped improve segregation in normal hearing subjects but not so in hearing impaired listeners. Differences were strongest when more voices were present.
Projektbezogene Publikationen (Auswahl)
-
“Improved binaural speech intelligibility by adding reverberation to the target speaker,” 23rd International Congress on Acoustics, Aachen, pp. 5773-5778
J. Grosse & S. van de Par
-
The role of reliable interaural time difference cues in ambiguous binaural signals for the intelligibility of multitalker speech. The Journal of the Acoustical Society of America, 147(6), 4041-4054.
Schoenmaker, Esther & van, de Par Steven
-
“Hörstrom Segregation von Vokalen basierend auf konkurrierenden F0 und Timbre Merkmalen,“ Fortschritte der Akustik - DAGA 2020, Hannover, Germany, 810-813.
V. Kies, M. David & S. van de Par
-
Tracking Musical Voices in Bach's The Art of the Fugue: Timbral Heterogeneity Differentially Affects Younger Normal-Hearing Listeners and Older Hearing-Aid Users. Frontiers in Psychology, 12.
Siedenburg, Kai; Goldmann, Kirsten & van, de Par Steven
-
“Einfluss von Tonhöhenmerkmalen auf die Sprechertrennung,“ Fortschritte der Akustik - DAGA 2022, Stuttgart, Germany, 1210-1213
A. Oetjen & S. van de Par
-
“Einfluss von Tonhöhenmerkmalen bei räumlicher Sprechertrennung,“ Fortschritte der Akustik - DAGA 2022, Stuttgart, Germany, 1200-1201
M. Meierott, A. Oetjen & S. van de Par
