Figure 8From: Context-dependent acoustic modeling based on hidden maximum entropy model for statistical parametric speech synthesisAveraged CMOS scores for the HMEM1, HMEM2, and HSMM. 95% confidence intervals are also indicated.Back to article page