...............
Larger version of above photo.
Modern silencers do not touch the bullet, and are very effective at suppressing the muzzle report of the firearm. Some cogent explanations have been offered in this thread, including notably Riss'; however, there is more disinformation.
A standard mil-spec sound test measures the sound pressure one meter to the side of the muzzle (and a bunch of other specifics). Using this test, the best-in-class centerfire rifle suppressors reduce the report vs. a bare muzzle by approx 30-35 dB.
Some have said that while the sonic crack exists, suppression is pointless. This is far from the experience of most suppressor shooters. First of all, from the shooter's/spotter's point of view, the crack is not readily audible unless there is a physical structure (like a wall) to reflect it from downrange of the muzzle. For others in the area, there is no "BOOM" report, as many have mentioned. The sonic crack is typically only audible relatively close to the bullet path, and only in certain spatial relationships to the bullet path.
A suppressed AR-15 is still fairly loud for the shooter because so much noise "comes through" the action. On the other hand, a suppressed bolt gun (such as pictured) is pretty quiet from the shooter's perspective.
A setup like this literally goes "Pfft pfft pfft":
...............
Larger version of above photo.
Another way to think about what Riss said in terms of surface area and heat is that a suppressor buffers the pressure, releasing it at a lower peak and average pressure but over a longer period of time.