I didn't tested it, but it should work:

the main difference is sf* objects in Max use 0/1 for start/stop. The rest is more or less the same.

I just rearranged boxes visually, so the patchers should do the same thing. Those sf* objects are quite complex wrt exact timing (in other ways also), I/O times are hard to predict. I must admit I hardly understand what's happening under the hood – this patch should record longer and longer sound files (as there is 1/5 second delay added each time). But it doesn't. I'm curious why this is the case.