Effect of automatic scene classification on speech perception of cochlea implant users in complex noises

Speech perception in everyday life with multiple noise sources is challenging for users of cochlea implants (CIs). The aim of this study was the evaluation of speech perception of CI users in dynamically changing listening situations with complex noises and reverberation. Additionally, the effect of automatic scene classification (SCAN) on speech perception was examined. 16 uni- and 16 bilateral CI users (Cochlear Nucleus 6 processors) as well as 15 normal hearing (NH) participated in this study. The speech reception threshold (SRT) in noise was determined under free field conditions and with reverberation. In order to test the effect of the classificator, the test conditions were integrated in an interleaved version of the Oldenburg Sentence Test (OLSA). The mean SRT across all conditions was -10.7 dB SNR in the NH group. The unilateral CI users achieved 2.9 dB SNR (without classificator) and 1 dB SNR (with classificator). The mean SRT of bilateral CI users was 0 dB SNR and -1.5 dB SNR (without and with classificator respectively). Both CI groups had their maximum improvement by applying the classificator in the free field condition with only one noise source in the rear hemisphere (Improvement unilateral: 4.2 dB; bilateral: 4.1 dB). Using the automatic scene classificator can improve speech perception of CI-Users in many listening situations. The effect depends on number and type of noise sources as well as the acoustic properties of the room.

Cochlear Deutschland GmbH & Co. KG

Poster-PDF A-1573.PDF

Publication History

Article published online:
10 June 2020