Objective Evaluation Method of Broadcasting Vocal Timbre Based on Feature Selection

Table 5

Results of GMM model order.

GMM order	4	8	16	32	64
Classification accuracy	60.21%	71.35%	85.63%	89.48%	89.98%
Training duration	568.7 s	663.9 s	1035.1 s	1335.6 s	2147.3 s