The performance of an automatic speech recognition (ASR) system strongly depends on the representation used for the front-end. If the extracted features do not include all relevant information, the...
Lic.-avh. (sammanfattning) Stockholm : Kungliga Tekniska hoĢgskolan, 2009.