What function in lossy codecs soften transients, like framesize in opus? I am wondering because can this kind of impact (exaggerated) be used to make a dsp (VST, LADSPA, LV2, etc) that can tame transients in material like overheads, aggressive vocals, etc? Maybe it could be a little bit like a tape effect?
It's block size of time to frequency domain transform (often MDCT). Short blocks result in higher time resolution and lower frequency resolution, long blocks result in lower time resolution and higher frequency resolution. Softening artifacts happen when a transient is encoded with a long block (due to bandwidth deficiency or mistake of psychoacoustic model).
I don't know if it is used in transient softening DSPs, I guess it's possible.
rutra80: Thank you so much for the reply. It would be interesting to see what it can do to audio recordings that can be problematic because they are have too much transients like in overhead or close miced with condensors.