ABSTRACT
This chapter provides an overview of objective quality and intelligibility measures which have been used to assess the quality and intelligibility of speech processed by noise-reduction algorithms. Subjective listening tests provide perhaps the most reliable method for assessing speech quality or speech intelligibility. These tests, however, can be time-consuming requiring in most cases access to trained listeners. For these reasons, several researchers have investigated the possibility of devising objective, rather than subjective, measures of speech quality and intelligibility. The ideal measure should predict with high accuracy the results obtained from subjective listening tests with normal-hearing listeners. Objective measures of speech quality are implemented by first segmenting the speech signal into 10–30 ms frames, and then computing a distortion measure between the original and processed signals. A single, global measure of speech distortion is computed by averaging the distortion measures of each speech frame.
