论文部分内容阅读
为了高效率量化线谱频率(linear spectrumfrequency,LSF)参数,提出了基于G auss ian混合模型(G auss ian m ix ture m ode l,GMM)的LSF量化算法。假设LSF矢量属于GMM中的某一个G auss ian分布,用G auss ian分布随机矢量的量化方法对LSF矢量进行了量化。利用准确的G auss ian分布变量量化误差,得到了G auss ian分布矢量的比特分配方法。应用G auss ian分布随机变量的非均匀量化方法量化每一维LSF参数。最后给出了分裂矢量量化、基于概率密度函数(probab ility dens ityfunction,PDF)量化方法和该算法的性能对比。该无记忆LSF量化算法在21 b/帧可以达到透明量化,比传统Sp litVQ节省3 b。
In order to efficiently quantize linear spectrum frequency (LSF) parameters, a LSF quantization algorithm based on the G auss ian hybrid model (GMM) is proposed. Assuming that the LSF vector belongs to a certain G auss ian distribution in the GMM, the LSF vector is quantized with a G auss ian randomized vector quantization method. Using the accurate G auss ian distribution variable quantization error, a bit allocation method of G auss ian distribution vector is obtained. The LSF parameters of each dimension were quantified by non-uniform quantization of G auss ian distribution random variables. Finally, we give the vector quantization of split vector, the quantization method based on probability density function (PDF) and the performance comparison of the algorithm. This memoryless LSF quantization algorithm achieves transparent quantization at 21 b / frame, saving 3 b from the traditional Sp litVQ.