One of the elements of a structured meeting is to have a meeting report. In Flanders for instance there is a new legislation that requires all city council meetings to be available, either in print-out form or in audiovisual format.
Also in other meetings, a transcription of the meeting is required.
These speech recognition systems use the recorded audio streams as an input. From initial tests, we can conclude that the quality of the captured audio signal (e.g. the speech intelligibility) has a influence on the quality of speech recognition systems.
The goal of this work is to qualify the required audio quality in order to perform a good transcription.