8–12 Sept 2026
Europe/Vienna timezone

The Influence of Speech Directivity on Speech Recognition and Scene Analysis in Multi-talker Scenes

FA2026/760
9 Sept 2026, 11:40
20m
Saal 11A (Messe Congress Graz)

Saal 11A

Messe Congress Graz

Speaker

Allison Trine (University of Illinois Urbana-Champaign)

Description

Sound source directivity plays a role in perception, including for scene analysis and speech recognition in complex acoustic environments. For example, speech directivity cues enhance target speech recognition when the target faces the listener and masker talkers face away from the listener. We examined this effect of speech directivity on target speech recognition as a function of masker head orientation angle. We also tested listeners’ ability to use speech directivity cues to discriminate facing vs. non-facing talkers in a multi-talker scene and to identify the spatial location of the target talker facing the listener. Speech recognition showed significant improvement when maskers faced >45 degrees away from the listener, with additional incremental improvements as masker head angle increased to 180 degrees. Scene analysis (i.e., target talker location identification) accuracy and speed likewise improved as the head angles of non-facing talkers increased. Performance in all tasks was poorer when extended high frequencies (>8 kHz) were filtered out, likely due to the loss of directivity cues associated with these highly directional frequencies. Directivity cues may play an important role for speech perception in complex acoustic environments.

Authors

Brian Monson (University of Illinois Urbana-Champaign) Allison Trine (University of Illinois Urbana-Champaign)

Presentation materials

There are no materials yet.