Well IMHO in my experience the only real advantage of analog summing over ITB channel+bus saturation is the absence of aliasing.
Really you can't "economically" (meaning intern of CPU usage) create saturation ITB that does not produce aliasing, the same is true for any compression faster than 1s for example.
That's because you theoretically have to oversample to account for any harmonics generated over over the Nyquist frequency (half the sample rate) so if your saturation algorithm produces 2nd,3rd, 4th and 5th harmonics (pretty standard in analog emulation) you are going to get aliasing for any content over 4000Hz (unless you oversample) if you do 2x oversampling you get aliasing over 8Khz and so on. The same goes for compressors because you are multiplying the envelope followers with the audio signal when you do the gain reduction so if you have 0.2s attack and 0.2s release for example you are multiplying a 2.5Hz envelope follower with your audio signal, this means that over 8KHz you get aliasing.
Oversampling is crucial for non linear digital audio processing, you have to roughly grant at least 192Khz of bandwidth to saturator and compressors to get an acceptable level of aliasing (but you'll never be alias free)
IMHO this is the only reason why analog processing sound smoother and more defined, it doesn't has any aliasing in the high end!
Saverio
Really you can't "economically" (meaning intern of CPU usage) create saturation ITB that does not produce aliasing, the same is true for any compression faster than 1s for example.
That's because you theoretically have to oversample to account for any harmonics generated over over the Nyquist frequency (half the sample rate) so if your saturation algorithm produces 2nd,3rd, 4th and 5th harmonics (pretty standard in analog emulation) you are going to get aliasing for any content over 4000Hz (unless you oversample) if you do 2x oversampling you get aliasing over 8Khz and so on. The same goes for compressors because you are multiplying the envelope followers with the audio signal when you do the gain reduction so if you have 0.2s attack and 0.2s release for example you are multiplying a 2.5Hz envelope follower with your audio signal, this means that over 8KHz you get aliasing.
Oversampling is crucial for non linear digital audio processing, you have to roughly grant at least 192Khz of bandwidth to saturator and compressors to get an acceptable level of aliasing (but you'll never be alias free)
IMHO this is the only reason why analog processing sound smoother and more defined, it doesn't has any aliasing in the high end!
Saverio
Statistics: Posted by HoRNet — Fri Jan 19, 2024 10:05 am
