Danish Sentence Test Project (DAST)
Many research and development projects in the fields of speech and hearing sciences rely heavily on speech corpuses. For example, a signal processing engineer working on the development of a new compression algorithm will often use sentence testing to optimize the parameters of their algorithm, and a machine learning engineer working on the development of a new noise reduction algorithm often uses sentences from a speech corpus to train their classifiers. In such use cases, it is valuable to have access to a large number of sentences in order to test different algorithm parameters, to test with or train on different noise types or amounts of reverberation, and to investigate test/re-test reliability or generalizability.
Although speech corpuses large enough for such sentence testing exist in other languages, a large enough corpus in Danish does not. As a result, Denmark-based researchers and developers in both academic and industrial settings working within speech and hearing sciences have been somewhat handicapped in comparison to, for example, researchers and developers working in English-speaking countries. The main objective of this project is therefore to create a new sentence test together with a new, larger corpus of Danish sentences.
Supported by the William Demant Foundation, Widex, GN Hearing and DTU