Optimisation and Modelling of speech intelligibility

Speech intelligibility is a way to express how well one can understand a particular speech signal. For many audio/speech applications it is important that the speech intelligibility is good. For example, think about hearing aids, mobile telephones, public address systems, etc.

It is possible to model the speech intelligibility. Within the CAS group several speech intelligibility models have been developed that are among the state-of-the-art in intelligibility modelling. These models are very useful both optimise the speech intelligibility in noisy conditions and, to be able to objectively evaluate the speech intelligibility in an instrumental way. Within the topic of speech intelligibility there are several possible assignments in a diversity of directions:

• Development of speech intelligibility models taking hearing impaired people into account.
• Optimisation of speech intelligibility models based on existing speech intelligibility models.
• Machine learning based optimisation of speech intelligibility.

Contact

Last modified: 2019-10-04