摘要
为实现音频信号的低速率编码,提出一种正弦参数量化的新方案.该方案利用人耳掩蔽效应和实时参数统计特性,基于高速率原理建立并显式求解率失真优化问题,得到具有解析形式的参数量化器,实现了在正弦分量间和正弦参数间动态分配编码比特.本量化方案无迭代运算过程,适合实时低速率音频编码应用.与经典方法比较,平均比特速率每正弦减少约17%.当速率大于15bit/正弦,重建信号的感觉失真小于球形量化方案.
In order to achieve low bit rate coding of audio signals ,this paper presents a new scheme for quantization of sinu-soidal parameters .By using the masking effect of the human auditory system and real-time parameter statistical properties ,the scheme based on high-rate theory to establish a rate-distortion optimization problem that is solved explicitly ,derives the parameter quantizers which have analytical expressions ,and dynamically implements the coding rate distribution between sinusoidal components and sinusoidal parameters .The quantization scheme without iterative calculation processes ,is suitable for real-time low-rate audio coding applications .Compared with the classical method ,the average bit rate of the proposed scheme is reduced 17% per sinusoid . When the rate is larger than 15 bit per sinusoid ,the perceptual distortion of the reconstructed signals is smaller than the spherical quantization scheme’s perceptual distortion of the reconstructed signals .
出处
《电子学报》
EI
CAS
CSCD
北大核心
2014年第4期798-803,共6页
Acta Electronica Sinica
基金
北京市教育委员会科技计划(No.KM201310017009)
关键词
高速率原理
正弦编码
匹配追踪
自适应联合量化
high-rate theory
sinusoidal coding
matching pursuit
adaptive joint quantization
