A study on improving decisions in closed set speaker identification

In this study, closed-set, text-independent speaker identi cation is considered and the problem of improving the reliability of the decisions made by available algorithms is addressed. The work presented here is based on the idea of combining the evidences from di erent algorithms or decision strategies to improve the recognition performance and the reliability. For this purpose, the models generatedby a single algorithmfor 17 speakers from the SPIDRE database are considered and a matrix of speaker-to-model tness values is processed by two di erent decision strategies. Ideas from the Mathematical Theory of Evidence are applied to combine the decisions produced by these two strategies to generate a better decision on the speaker identity. The combined decision show an improved degree of corectness hence suggesting a promisingway of combining the decisions from partially successful algorithms.