摘要
Most information in speech is carried in spectral changes over time, rather than in static spectral shape per se. A form of signal processing aimed at enhancing spectral changes over time was developed and evaluated using hearing-impaired listeners. The signal processing was based on the overlap-add method, and the degree and type of enhancement could be manipulated via four parameters. Two experiments were conducted to assess speech intelligibility and clarity preferences. Three sets of parameter values (one corresponding to a control condition), two types of masker (steady speech-spectrum noise and two-talker speech) and two signal-to-masker ratios (SMRs) were used for each masker type. Generally, the effects of the processing were small, although intelligibility was improved by about 8 percentage points relative to the control condition for one set of parameter values using the steady noise masker at -6 dB SMR. The processed signals were not preferred over those for the control condition, except for the steady noise masker at -6 dB SMR. Further work is needed to determine whether tailoring the processing to the characteristics of the individual hearing-impaired listener is beneficial.
源语言 | 英语 |
---|---|
页(从-至) | 2987-2998 |
页数 | 12 |
期刊 | Journal of the Acoustical Society of America |
卷 | 131 |
期 | 4 |
DOI | |
出版状态 | 已出版 - 4月 2012 |
已对外发布 | 是 |