Thesis
The factors affecting the psychometric function for speech intelligibility
- Creator
- Awarding institution
- University of Strathclyde
- Date of award
- 2013
- Thesis identifier
- T13361
- Qualification Level
- Qualification Name
- Department, School or Faculty
- Abstract
- Older listeners often report difficulties understanding speech in noisy environments. Increasing the level of the speech relative to the background - e.g. by way of a hearing aid - usually leads to an increase in intelligibility. The amount of perceptual benefit that can be gained from a given improvement in signal-to-noise ratio (SNR), however, is not fixed: it instead depends entirely on the slope of the psychometric function. The shallower the slope, the less benefit the listener will receive. The aim of the research presented in this thesis was to better understand the factors which lead to shallow slopes. A systematic survey of published psychometric functions considered the factors which affect slope. Speech maskers, modulated-noise maskers, and target/masker confusability were all found to contribute to shallow slopes. Experiment 1 examined the role of target/masker confusion by manipulating masker intelligibility. Intelligible maskers were found to give shallower slopes than unintelligible ones but subsequent acoustic analysis demonstrated that modulation differences between the maskers were responsible for this effect. This was supported by the fact that the effect was seen at low SNRs. Experiment 2 confirmed that the effects of modulation and target/masker confusion occur at different SNRs. Experiments 3 and 4 demonstrated that directing attention to the target speech could "undo" the effects of target/masker confusion. In Experiments 5 and 6 a new method was developed to study whether slope effects are relevant to "real-world" situations. The results suggested that using continuous speech targets gave shallower slopes than standard speech-in-noise tests. There was little evidence found to suggest that shallow slopes are exacerbated for older or hearing-impaired listeners. It is concluded that in the complex demands of everyday listening environments the perceptual benefit received from a given gain in SNR may be considerably less than would be predicted by standard speech-in-noise paradigms.
- Resource Type
- DOI
- Date Created
- 2013
- Former identifier
- 987900
关系
项目
| 缩略图 | 标题 | 上传日期 | 公开度 | 行动 |
|---|---|---|---|---|
|
|
PDF of thesis T13361 | 2021-07-02 | 公开 | 下载 |