2021/03/12 by Winkler, Alexandra, Schlüter, Anne, Gebauer, Tina +4
#610 Medical sciences #CI #GÖSA #HSM #Medicine #Sprachverstehen #Sprechtempo #Störgeräusch #Zeitkompression #noise #speech rate #speech recognition #time compression
paper · doi:10.3205/zaud00014
Speech recognition for cochlear implant (CI) users can be derived using the Hochmair-Schulz-Moser-(HSM) sentence test or the Göttingen sentence test (GÖSA). Both tests use everday sentences, but they differ in articulation, speech tempo, interfering noise and calibration. The GÖSA is regarded in clinical practice as being more difficult to understand for CI users than HSM. Speech rate as a measure of speech tempo of GÖSA is 279 syllables/min and higher than the speech rate of HSM (222 syllables/min). The typical interfering noises, which are used (CCITT noise for HSM and GÖnoise for GÖSA), cause different masking patterns of speech material. Speech recognition was determined using different noises and speech rates for HSM and GÖSA for CI users and people without hearing impairment (NH). CCITT noise and speech simulating noises (GÖnoise for GÖSA and a noise generated from the speech material of HSM) were used for GÖSA and HSM. Speech rate of GÖSA and HSM was changed to present both speech materials at 222 and 279 syllables/min. Therefore, speech recognition thresholds (SRT50) for a score of 50% correct were measured for four different conditions per speech test (two different noises and two different speech rates). All signals (noise and speech) were presented in a sound attenuated booth using one loudspeaker, which was placed in front of the listeners. There was no significant difference in SRT50-values for all GÖSA conditions for both groups of participants. SRT50-values in HSM significantly differed depending on speech rate and noise. SRT50-values increased with increasing speech rate. CI users showed significantly increased SRT50-values in CCITT whereas for group NH, decreasing SRT50 values in CCITT were observed. When using the same calibration, SRT50 values for HSM with higher speech rate were comparable to those for GÖSA for group NH in speech simulating noise and for group CI in CCITT.