I’ve been working on setting up a DSP pipeline on my Macbook Air for headphone listening and needed to find a crossfeed component. I initially looked for an implementation of the B2SB algorithm but struck out on finding a Mac implementation. Instead, I ran across the Canz3D Spatial Simulation plugin and boy am I glad I did!
I use crossfeed extensively when listening on my phone, and I’ve also tried more sophisticated speaker simulation algorithms like Out of Your Head and Toneboosters Isone. I’ve found all of these speaker simulation DSPs to sound pretty artificial and gimmicky and never stick with them.
Canz3D is in a different league. I haven’t even tweaked any of the numerous settings and it already sounds amazing. Whereas plain crossfeed does narrow the stereo image somewhat, it’s really pretty subtle and doesn’t fundamentally change the image that much. The other speaker simulation algorithms I’ve tried seem very aggressive, messing with frequency response and sounding somewhat artificial. Canz3D really does sound more speaker-like, with an improved front image and surprisingly much improved instrument separation and localization.
So, what are the downsides?
- It’s not clear to me whether or not this thing is currently maintained and supported. It fades to silence after 3 minutes unless you enter a registration code, and I’m not sure how to get such a code. I’ve contacted the author, we’ll see.
- It seems to be Mac only
- It works best with simpler recordings that don’t use mono close-miking. It still sounds fine with other stuff, but it’s definitely more impressive with hifi stereo recording like Nozze de Figaro.