DVA04_声音基础解说.ppt

* 频响曲线：输出增益与输入信号频率变化的函数关系 相对频响 * * 其中post-masking的时间会比较长，约是pre-masking的10倍。 * * 时域和频域掩蔽效应的折衷 稳态信号，滤波器组要有好的频谱分辨力，这意味着需要一个长的窗函数或长的 FIR 滤波器。 瞬态信号，滤波器组要有好的时间分辨力，这就要求短的窗函数和 FIR 长度。 * * When digitizing a signal, one inevitably introduces quantization noise into the representation. (Click) The difference between the signal and the quantization representation. (Click) The traditional solution to quantization noise was to digitize the signal with a large number of bits. (Click) With each bit increase, the quantization noise decreases by 6 dB. Each bit increase means a doubling in the number of levels, and thus halving the quantization noise. (Click) The CD format typically using a 16-bit sample resolution, resulting in a data rate of 4.32 Mb/s for stereo sounds. * When digitizing a signal, one inevitably introduces quantization noise into the representation. (Click) The difference between the signal and the quantization representation. (Click) The traditional solution to quantization noise was to digitize the signal with a large number of bits. (Click) With each bit increase, the quantization noise decreases by 6 dB. Each bit increase means a doubling in the number of levels, and thus halving the quantization noise. (Click) The CD format typically using a 16-bit sample resolution, resulting in a data rate of 4.32 Mb/s for stereo sounds. * This is the frequency domain representation of what we discussed it the previous slide. The blue is the original signal. The red is the digitized signal. (Click) The green is the difference of the two or the quantization noise. As you can see, the quantization noise is not limited to the frequency region of the signal. (Click) Just as in the time domain representation, you can see with a larger number of bits the quantization noise is reduced. * From the previous slides, we know that quantization noise is wideband. The quantization noise of one frequency spreads to other frequencies. To combat this problem, MPEG utilizes sub-ba