Word error and confusion patterns in an audiovisual German matrix sentence test (OLSA)
* Presenting author
One of the established tests for speech intelligibility in quiet and in noise is the Oldenburg sentence test OLSA. This test evaluates the speech reception threshold (SRT), i.e., the speech level or signal-to-noise ratio that leads to a specified word intelligibility, using phonetically balanced 5-word sentences, where each word is equally intelligible. The standard matrix sentence test is audio-only, although it is known that adding visual information to a speech intelligibility task can change the outcome, e.g., increasing the speech intelligibility, or inducing the McGurk effect. Adding visual cues, i.e., being able to see the speaker, could alter the homogeneity of the words with respect to their intelligibility and the expected outcome of the matrix sentence test. In this work we compare the word error and confusion patterns of the audio-only trials versus the audio-visual ones. Possible reasons of these effects are discussed.