Automatic identification of persons by using the speaking style
|Number of Students||
In the literature many approaches exist that deal with the problem of detecting authors of written texts, whereby good results can be achieved. Nevertheless, very few research is done to find out, if similar methods can also be utilized to identify persons by their style of speaking.
The aim of this thesis is to adapt existing algorithms for author attribution in a way so that they can be applied to spoken text. They should be optimized in order to achieve the best possible accuracy for identifying persons by transcripts of their spoken text, i.e., without using the audio signal.