FPGA Implementation for GMM-Based Speaker Identification

Table 1

Hardware and software results (percentage recognition accuracy) for testing with 5 s test utterance.

Utterance; length	Software; (5 seconds)	Hardware; (5 seconds)
Test 1	80.77	78.30
Test 2	56.40	55.20
Test 3	68.54	64.90
Test 4	72.75	69.40
Mean	69.62	66.95
S.D	10.17	9.61
95% confidence	9.96	9.42