A pipe section that is slightly oval will pass every check that compares it with a nominal size. Its average diameter is right, its circumference is right, and a tape around it gives the number you expect. What is wrong is the relationship between one axis and the other.
A slip joint does not care about averages. It cares about the worst point of contact, and an oval end gives you two lines of contact with two gaps between them. Those gaps are leak paths, and they run the full length of the engagement, so the clamp has nothing to seal against.
The usual cause is a clamp from a previous life. A band tightened hard onto a thin wall pulls the tube into a slightly flattened shape, and the shape stays after the clamp is removed because the metal has yielded. Used sections carry this history invisibly.
The other causes are handling and bending. A pipe dropped on its end, or cut with a blade that pinched as it went through, ends up out of round at exactly the face you need to join. A bend also pulls the section oval, so an end cut close to a bend is more likely to be oval than one cut from a straight run.
Use a caliper, not a tape, and take readings at two axes at right angles, then repeat a little further in from the end. The difference between the two axes is the ovality, and the difference between the two depths tells you whether the fault is at the cut face or runs into the pipe.
Do the same on the mating part and compare the axes, not the averages. Rotating one part against the other can line up the two large axes and turn a joint that will not assemble into one that slides together, which is a legitimate trick for assembly but not a repair.
Ovality is corrected by pushing the small axis back out, in small increments, with a tool that supports the wall around the whole circumference rather than a single point. A proper expander or a collet does that; a hammer and a socket does not, and it work-hardens the metal where it hits.
Work slowly and check with the caliper between passes. Each pass should move the metal a little and leave the surface smooth. If the wall starts to show a crease, a bright line or a change in sound when tapped, stop: the metal has been pushed past what it will take, and that end now needs to be cut off and replaced.
Titanium springs back further than stainless, so a tool that visibly oversizes the pipe may leave it almost where it started once released. It also work-hardens quickly, so the number of attempts you get is small and the third attempt is much harder than the first.
That combination makes local heat tempting and unwise, because a torch on a thin titanium wall changes the structure in a way you cannot see and cannot undo. On a titanium exhaust system the better answer for an oval end is usually to cut back past the affected length and join with a fresh, round section.
If the end is round at the face and oval deeper in, resizing will not hold, because the clamp will simply flatten it again at the same place. That pipe needs either a longer engagement so the clamp sits on round metal, or a joint type that does not depend on round-on-round contact.
A sleeve that bridges the affected length is the pragmatic option, and it also gives the clamp a uniform surface to work against. Where the pipe is a replacement section that has to be cut anyway, cutting back to sound round metal and adding the difference in length is cleaner than rescuing an end that will keep going oval.
