The Danish sentence test (DAST) corpus of audio and audio-visual recordings of sentences and monologues

A new, larger corpus of Danish sentences has been developed to create a new Danish Sentence Test (DAST). The corpus is made up of audio and audio-visual recordings of 1200 linguistically balanced sentences, all of which are spoken by both two male and two female professional talkers. The sentences were constructed using a template-based method that facilitated control over both word frequency and sentence structure. The resulting written sentences were evaluated linguistically in terms of phonetic distributions and naturalness. Due to a relatively low number of sentences being rated as problematic, all 1200 sentences were included in the recording stage. Besides the sentences for each talker, 30+ minutes of monologues were also recorded from each of the same four talkers. For one selected talker, the sentences were presented to 60 normal-hearing listeners at a fixed set of signal-noise ratios to estimate a psychometric function for each sentence. These psychometric functions, together with the naturalness results, phonetic distributions, and quality assessments, were employed to build equivalent lists of 20 sentences for speech-in-noise testing. Development of an adaptive version of the test to measure speech reception thresholds is under way, as is the validation of the equivalency of the resulting lists.