Aided speech recognition in noise, perceived effort and explicit cognitive capacity
2008 (English)In: International Hearing Aid Research Conference IHCON 2008, Lake Tahoe, California, 13-17 August 2008.,2008, 2008Conference paper (Refereed)
Speech recognition in noise is an effortful process requiring explicit cognitive processing. It may be influenced by level and type of noise and by the signal processing algorithms employed when hearing is aided. These complex relationships may be understood in terms of the working memory model for Ease of language Understanding (ELU, Rönnberg et al., in press). This model predicts that under challenging listening conditions, explicit cognitive processing demands will be high and that persons with good explicit cognitive capacity will be better listeners. Previous work has suggested that they may also find listening less effortful (Behrens et al., 2004; Larsby et al., 2005; in press). We studied this issue by including subjective effort ratings in a larger study designed to investigate aided speech recognition in noise and cognition. 32 experienced hearing aid users participated. Effort was rated using a visual analogue scale and the speech material was the Hagerman sentences presented in three fixed speech to noise ratios of +10 dB, +4 dB and -2dB. Effort was rated in modulated and unmodulated noise with fast and slow compression release settings, after each of two nine week training sessions with the same settings. Speech recognition performance was tested objectively under the same conditions using an adaptive procedure. Order of testing was balanced. Explicit cognitive capacity was measured using the reading span test.
ANOVAs and correlations were computed. Preliminary results showed that decreasing SNR led to greater perceived effort and that the difference in perceived effort between the highest and the lowest SNR was greater in unmodulated noise than in modulated noise. Speech recognition performance in unmodulated noise generally correlated with effort ratings under similar conditions but in modulated noise generally it did not. Effort ratings correlated with reading span performance at the lowest SNR (-2dB) but only in unmodulated noise after the first training session.
These preliminary findings show that subjective ratings of the effort involved in aided speech recognition covary with noise level and performance but that these effects are reduced by noise modulation. Further, the perceived effort of aided speech recognition at low SNR may be related to explicit cognitive capacity as measured by the reading span test. However, we only find evidence of this in unmodulated noise after the first training session. These findings extend previous work on perceived effort and cognitive capacity and provide further evidence that type of noise is an important factor in this relationship.
Place, publisher, year, edition, pages
IdentifiersURN: urn:nbn:se:liu:diva-44359Local ID: 76392OAI: oai:DiVA.org:liu-44359DiVA: diva2:265221