论文部分内容阅读
为了实现比ATRAC1高一倍的压缩比,ATRAC3在MDCT频谱化时的采样单元由原来的512个样本(由低到高三个字频带分别是128个、128个、256个)提高到了1024个(四个子频带各256个),每个声组(212字节)对应的样本数提高了一倍,这是ATRAC3降低位速率的根源所在。剩下的工作只是在高压缩的情况下如何提高音质的了。Sony在ATRAC3里引入了结果调整(Gain Adjustment)和旋律分割(Tonal Component Separation)两项新技术来做到这点。
In order to achieve twice the compression ratio of ATRAC1, the sampling unit of ATRAC3 during MDCT spectrum conversion is increased from 512 samples (128, 128, 256, respectively) from the original 512 samples to 1024 256 for each of the four subbands), doubling the number of samples per sound group (212 bytes), which is where ATRAC3 reduces the bit rate. The rest of the work is just how to improve the sound quality in the case of high compression. Sony did two new techniques called Gain Adjustment and Tonal Component Separation in ATRAC3.