Agree re transients. I think Miller Puckette's "bonk" object did transient detection using filter bank. Can't find the paper right now. In general I think detecting rapid changes in phase performs better than changes in amplitude for transient detection.
I believe that the good sounding phase vocoder algorithms do transient detection and instantaneously reset the phase of the spectral peaks associated with the transient (effectively passing through the original transient without smearing it).
Agree re transients. I think Miller Puckette's "bonk" object did transient detection using filter bank. Can't find the paper right now. In general I think detecting rapid changes in phase performs better than changes in amplitude for transient detection.
I believe that the good sounding phase vocoder algorithms do transient detection and instantaneously reset the phase of the spectral peaks associated with the transient (effectively passing through the original transient without smearing it).