Improved speaker, verification over the cellular phone network using phoneme-balanced and digit-sequence-preserving connected digit patterns
Tomoyuki Kato, Tohru Shimizu · 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03). · 2003
In order to achieve high accuracy text-prompted speaker verification over the cellular phone network, a phoneme-balanced connected digit pattern for enrollment and digit-sequence-preserving connected digit patterns for verification (i.e. patterns preserving partial digit sequences of the enrollment pattern) are proposed. In addition to those, a decision procedure using multiple patterns is designed to overcome the low quality of cellular phone speech. Experimental results showed the phoneme-balanced and digit-sequence-preserving patterns reduced more than 50% of equal error rate compared to the conventional randomly-chosen and randomly-reordered digit patterns. The decision procedure reduced 60% of the error rate. Overall, the error rate obtained by the proposed method was 1% for 99% of clients and 95% of imposters.