In general… no, vvvv can’t really do the same kind of signal processing that can be done in Max/MSP or PureData (or Reaktor). VVVV’s support for audio is limited to directshow audio graph construction (including vst plugins) and some basic analysis , the base.net plugin stuff (which can do… some stuff, i’m not really familiar with it).
The structure of the VVVV engine isn’t really suited to signal-rate audio processing. VVVV evaluates everything in a sequence of frames, in which every node effectively does its stuff at the same time. Each frame takes a bit of time to process, meaning that you generally end up with a rate around 30-100 frames/second. Every node in a VVVV patch must be processed during each frame (except for some optimizations where some nodes only do work if their inputs change).
MSP does a similar kind of frame-based processing with its signal processing stuff. However, only the signal-rate nodes are processed in each of these frames. The signal stuff runs on a separate thread. Other nodes that aren’t signal nodes send messages to each other and respond to messages from other nodes. So, their activity is separated from the signal stuff. The signal thread can run at >44100 samples/second.
In case you can’t tell, i like Max/MSP…
For rendering though, VVVV can’t be beat.