|
发表于 2012-2-18 18:14:06
|
显示全部楼层
找到了这个动态卷积大概的介绍
Dynamic convolution is a name of a mathematical method that is aimed to model a nonlinear device, using linear techniques. The idea is simple and can be extended to multiple dimensions of parameters, but basically, the simplest one dimensional implementation is using the idea of measuring multiple impulse responses at different states of the non-linear device, e.g. at different volume levels. In the reproduction, the input signal is convolved with the appropriate impulse response that is dynamically selected according to the input signal's parameter, such as its level.
简单翻译就是说采样在不同情况下(比如不同的音量大小)的脉冲相应(音箱音质动态等特点的数学模型),之后输出信号就是输入信号卷积一个合适的模型(就是说输入信号把这个模型的特点都吸收了), 被吸收的模型是根据输入信号的参数(比如音量)动态选择的。
不知道说清楚了没。。。
说白了就是算法更复杂些,考虑的参数更多点,假如说line6只采样一个音量下的数学模型,其他音量下都用同一个模型,而这个东西就采样了不同音量下的数学模型,通过测量音量的大小来选择用哪个模型,这样出来的结果更精确点 |
|