The relationship between the Speech Transmission Index and measures of cognitive performance

The presence of intelligible ambient speech is known to impair cognitive performance at silent, concentrated work. This experimental result is relevant to openplan offices, where individual task performance is assumed to be impaired by conversations of colleagues. This is a severe matter for the efficiency of companies. However, it has to be acknowledged that measures of cognitive performance in laboratory studies do not directly match real life office tasks and effects are expected to be task specific. The Speech Transmission Index (STI) is assumed to predict how much performance is reduced due to ambient speech in dependency of its intelligibility. A model in terms of an STIperformance-curve was proposed by Hongisto (2005) and target values for the acoustic quality of open-plan offices were defined. However, the model as well as the classification scheme still need to be elaborated. The reported studies systematically evaluate the proposed STI-performance-curve for STI values from 0.20 to 0.70, where the greatest performance effects are predicted. As a consequence of the results, new target values have to be discussed. Since the decline of performance due to ambient speech of varying intelligibility is expected to be task-sensitive, it needs to be investigated whether the model applies to different tasks. Therefore, experimental results are presented which address the issue of task sensitivity