ABSTRACT
BACKGROUND: Informational masking (IM) refers to the problem of understanding target speech in the simultaneous presence of a masking speech signal. The adverse effects are attributable to the additional information provided by the competing talker. Cochlear implant (CI) recipients are not able to understand speech satisfactorily in situations such as these. The aim of the present study is to examine the extent to which CI recipients are able to use differences between the speakers with respect to level (target-to-masker ratio, TMR) and fundamental frequency (f0) in order to improve speech intelligibility. SUBJECTS AND METHODS: Target (TS) and masker sentences (MS) were selected from the Oldenburg Sentence Test (OlSa) and were modified and superimposed. The TS were explicitly labelled using a keyword. The MS were changed with respect to TMR and f0. The TS intelligibility of different modifications was measured in six post-lingually deafened CI recipients and six normal listeners (NL). RESULTS: The NL revealed speech understanding close to 100% even for small differences in f0 of 40 Hz or level differences of 5 dB. In CI recipients, a significant change in intelligibility could only be demonstrated with increased TMR, but not with differences in f0. CONCLUSIONS: In CI recipients, IM cannot be reduced by means of differences in f0 between the competing talkers, but only by level differences. This might be explained by the restricted spectral resolution and the insufficient transmission of f0 in CIs. Adverse effects in NL can predominantly be attributed to IM, while additional masking effects take place in CI recipients.
Subject(s)
Cochlear Implants , Deafness/diagnosis , Deafness/rehabilitation , Hearing Tests/methods , Perceptual Masking , Adult , Female , Humans , Male , Middle Aged , Treatment OutcomeABSTRACT
BACKGROUND: Prosody reflects rhythmic and melodic aspects in speech and is one of the quality measures that shows--apart from pure speech understanding--a rising interest in the assessment of technical hearing aids, especially of cochlear implants. At present, there is no adequate test battery for the German speaking population. The test battery presented in this study aims to fill this gap. METHODS: The test battery consists of four different modules addressing different prosodic cues, namely duration, question vs statement, sentence stress, and speaker gender. One part of the test battery aims at reflecting everyday situations and was realized with six different speakers representing various pitch frequencies and speak rates. Another goal of the tests is to detect very small differences in prosody perception. Therefore, natural utterances were artificially modified by changing the underlying acoustic parameters (e.g. duration, fundamental frequency). Measurements were performed with 12 normal listeners in order to investigate the properties of the tests. RESULTS: As expected, the normal listeners revealed discrimination rates of close to 100% in that part of the test battery using natural utterances. For the part with the modified stimuli, all four modules revealed very steep discrimination functions indicating that the normal listeners were able to use very small changes in the acoustic parameters as cues for prosody perception. Moreover, the slope of the discrimination function revealed very low variability. CONCLUSIONS: Together with preliminary data from cochlear implant recipients who clearly showed different psychometric functions, the test battery seems suitable for a comprehensive evaluation of prosody perception. This facilitates future examinations of speech processing beyond the pure understanding of speech.
