Figure 7From: Context-dependent acoustic modeling based on hidden maximum entropy model for statistical parametric speech synthesis Comparison of RMS error of phoneme durations as objective measure between the proposed method and HSMM one. Back to article page