INTERSPEECH 2014
15th Annual Conference of the International Speech Communication Association

Singapore
September 14-18, 2014

Does Elderly Speech Recognition in Noise Benefit from Spectral and Visual Cues?

Yatin Mahajan, Jeesun Kim, Chris Davis

University of Western Sydney, Australia

Previous research with young adults has shown that temporal (amplitude modulated, AM) cues are sufficient for recognizing speech in quiet but not for speech in noise. Speech perception in noise is more robust when spectral (frequency modulated, FM) cues are provided in addition to AM ones; visual cues (AV) provide an additional benefit. The elderly typically have problems recognizing speech in noise and it has recently been found that FM discrimination is worse in this group. Given this, it may be that elderly participants will not show an FM speech in noise benefit. To test the relative effectiveness of adding FM cues to AM ones for elderly versus young participants we compared auditory only (AO) speech identification of sentences, vowels and consonants in noise with AM and AM+FM presentation conditions. We also evaluated the relative effectiveness of visual cues for the elderly group. Although the elderly had poorer speech recognition performance overall, they showed a comparable visual benefit to the young group. Moreover, contrary to the prediction of a reduced benefit for FM cues, the FM benefit for the elderly was similar to that of young adults. These results were discussed in relation to speech specific auditory processing.

Full Paper

Bibliographic reference.  Mahajan, Yatin / Kim, Jeesun / Davis, Chris (2014): "Does elderly speech recognition in noise benefit from spectral and visual cues?", In INTERSPEECH-2014, 2021-2025.