@inproceedings {pub3203,
	title = {Balancing Gaussianity and sparseness in feature-space speaker adaptation for word prominence detection},
	author = {Andrea Schnall AND Martin Ernst Heckmann},
	year = {2016},
	month = {October},
	abstract = {Prosodic cues are an important tool of human communication. One of this cues is the word prominence, which we are using to express e.g. 
important information. Nevertheless in human-machine communication such cues are rarely used.
One problem for usage in speech processing is the large difference between different speakers. To overcome this problem, a common method 
is an adaptation of the new data to the trained model. 
Since for problem cases like the word prominence detection, we found that a classifier like SVM performs better than methods like GMM/HMM, 
we developed an adaptation method based on the radial basis function of the SVM. Through additional Gaussian and sparseness regularization 
terms, the method shows good performance, also with few available data. In this work we show how the weighting of the different terms 
influence the performance and can be used to improve the results. },
	publisher = {VDE },
	booktitle = {12. ITG Fachtagung Sprachkommunikation}
}