So I've been reading Microsound to get a better understanding of granular synthesis. While the general idea of granular synthesis is straightforward there are a lot of interesting theories about what granular synthesis can do.
Inspired by what I've been reading thus far, I decided to see how I might use Live to make grains. Here's the idea: take one instance of Sampler, drop in a sample and set the amplitude envelope so that it operates as an AD envelope and set its shape to a smooth triangle that lasts anywhere from 10-60ms. Drop an instance of the arpeggiator running at top speed (10hz) and a velocity plugin set to full random in front of Sampler. Group the arpeggiator and the velocity plug-ins and then create 9 chains of the devices. Offset the speed of each arpeggiator so that there is a 1hz gap between each one (e.g. first = 10hz, second = 11hz). This will give a nice smearing effect when you trigger the arpeggiators. From here on there are endless possibilities to get interesting sounds. Some things worth trying include: mapping velocity to the sample start time and the time parameters; using LFOs to generate panorama effects; varying the time and modes of the arpeggiators; changing the time and shape of the amplitude envelope.
Here are a few examples of what I came up with:
Granular synthesis using Sampler by matt.patey