After a repair the system does not usually get uniformly louder. It gets different in one place: a note that was there is gone, a new one sits on top at a particular engine speed, or the tone at idle has moved. That selectivity is a clue. A hole in a pipe makes noise across the range, while a change in a resonant structure shows up where that structure controls the sound.
So the first question is whether the new sound follows engine speed, follows load, or appears at one road speed only. Each of those points to a different mechanism, and answering it before touching the part saves a lot of guesswork.
A weld adds material. On a thin-wall shell, a bead along a seam or a patch over a crack changes local stiffness and adds mass at one point on a panel designed to vibrate in a particular way. If the weld sits near a point of maximum movement, the panel behaves differently by more than the size of the weld suggests.
The same weld pulls as it cools, and on a thin shell that pull shows as a slight flattening or a shallow dish around the repair. A curved surface pulled flat radiates sound differently. Neither change is a fault in the weld; both are consequences of putting heat into a resonant panel.
An incomplete repair changes the sound as well, and does so in a way that mimics resonance. A pinhole or a crack that was bridged but not sealed produces a thin high hiss, loudest when the gas is cold and the flow is fast, and it can fade as the metal expands and closes the gap.
That is the opposite of most mechanical noises, which get worse as things heat and loosen. A sound that is present cold and fades hot, especially a thin one with a hiss or a tick in it, should be treated as a leak until it is proven otherwise.
With the engine idling and the car cold, pass a hand carefully around the repair without touching it, feeling for a pulse of gas. A strip of tissue held near the joint will flutter at a leak. On a system that is not too hot, a little soapy water at the seam shows bubbles. Neither check disturbs anything.
For resonance, tap the shell cold with a knuckle along the repaired area and listen. A free, even panel gives a consistent note; a panel with a patch gives a dull thud at the patch and a different note either side. The boundary of that change tells you how far the weld has altered the panel.
A repair on a muffler shell often disturbs what is inside. Packing that was compressed against the wall can be pushed away from the weld, leaving a path where gas passes without being absorbed. A baffle reattached at a slightly different angle changes the volume of the chamber it bounds, and chamber volume is what sets the note.
This is the case where the sound change is not a defect at all. A muffler that has been opened and closed will rarely sound identical to one that was never cut, and a small shift in tone is expected. The titanium exhaust mufflers section describes how shell construction and packing affect the result.
A changed note with no leak, no rattle and no crack is a characteristic, not a failure. What matters is whether the new sound comes with something else: a buzz at one engine speed, a ticking that grows, discolouration spreading from the repair, or a smell of exhaust in the cabin. Those point at a mechanical problem behind the noise.
If the part has to be opened again, the weld is the place to start, and the same rules apply as for any repair on thin-wall material: sound metal, adequate shielding, and a profile that does not concentrate strain. The stainless steel exhaust mufflers range covers the sections where these repairs are common, and the note you are chasing is usually a chamber rather than a weld.
