In this study we investigated the speaker discriminating potential of nasals and fricatives. In the past, vowel formants were often used for speaker discriminating purposes, but it seems that they are not the best choice for this task. Our study suggests that the spectral moments of nasals and fricatives may discriminate speakers better than vowel formants. Furthermore, we analysed whether the context of the nasals and fricatives and the speaking style had any effect on their speaker discriminating power. The results indicate that a restriction of the context does not improve the speaker specific features, since the unrestricted context provided the best F-ratios. In the comparison of read and spontaneous speaking style, the latter shows a slightly better performance.
Bibliographic reference. Schindler, Carola / Draxler, Christoph (2013): "Using spectral moments as a speaker specific feature in nasals and fricatives", In INTERSPEECH-2013, 2793-2796.