15th International Congress of Phonetic Sciences (ICPhS-15)Barcelona, Spain |
We introduce a model for speaker adaptation that assumes no knowledge about speaker identity. The model is found to reproduce the difference in human vowel recognition performance for stimuli presented in blocked and mixed speaker context.
Bibliographic reference. Weenink, David J. M. (2003): "Adaptive vowel normalization: simulating human recognition in mixed and blocked speaker context", In ICPhS-15, 1429-1432.