论文部分内容阅读
为提高子带清浊音(unvoiced/voiced,U/V)解码端恢复算法在不同能量电平下的鲁棒性,提出了一种改进型能量自适应U/V参数解码端恢复算法。通过跟踪长时能量的变化轨迹,在Gauss混合模型(Gaussian mixed model,GMM)下,用归一化的能量参数和线谱频率参数(line spec-tral frequency,LSF)对U/V参数的分布特性进行估计。测试结果表明:在较低的能量电平下,与用绝对能量对U/V参数进行恢复的算法相比,该能量自适应U/V参数恢复算法能够将清浊音误判率降低10%～25%,并将合成语音的平均意见得分(mean opinion score,MOS)提高0.03～0.09,改善了算法的性能。
In order to improve the robustness of the unvoiced / voiced (U / V) decoding end recovery algorithm at different energy levels, an improved energy adaptive U / V decoding terminal restoration algorithm is proposed. By tracking the long-term energy trajectories, the distribution of U / V parameters was normalized with normalized energy parameters and line-spec-tral frequency (LSF) under a Gaussian mixed model (GMM) Characteristics are estimated. The test results show that the energy adaptive U / V parameter recovery algorithm can reduce the misjudgment rate of voiced / voiced sound by 10% ~ 50% at lower energy level compared with the algorithm of recovering U / V parameters by absolute energy. 25%, and increase the mean opinion score (MOS) of synthesized speech by 0.03 ~ 0.09, which improves the performance of the algorithm.