What "lossless trim" actually means, and why most trimmers aren't

There are two genuinely different ways to cut a clip out of an audio or video file, and most tools don't tell you which one they're using. It matters more than it sounds like it should.

Two ways to make a cut

The first way: decode the whole file into raw audio or video, cut the exact point you asked for, and re-encode it back into a file. This lands on the exact millisecond every time. It's also slow, relative to how simple a “trim” sounds, and if the original was in a lossy format like MP3 or MP4 — which most files are — re-encoding it throws away a little more quality on top of what the original encoding already lost. Do it a few times on the same file and it's noticeable.

The second way: don't decode anything. Compressed audio and video is already stored as a sequence of small chunks, and a lot of formats let you just find the chunks nearest your start and end points and repackage that slice as a new file, byte for byte, untouched. This is what “stream copy” means. It's close to instant, and there's no quality loss at all, because nothing was ever decoded in the first place.

The catch

Stream copy can only cut where a chunk boundary already exists — it can't manufacture a new one at the exact point you asked for, because that would require decoding. For audio, those boundaries are tiny, usually a few dozen milliseconds apart, so the cut lands close enough to instant that you won't hear the difference. For video it's more noticeable: keyframes — the frames a video can actually be cut on — are often a few seconds apart, so a fast trim can land a fraction of a second off from exactly where you asked.

For trimming silence off a voice memo, or roughly cutting a clip down to size, that's not something you'll notice. For matching a video cut to an exact beat in an edit, it might be — and in that specific case, an instant trimmer is the wrong tool, full stop, no matter how it's built.

Audio Trimmer and Video Trimmer both use stream copy, on purpose — fast and lossless, with cuts that land on the nearest frame rather than the exact millisecond. For most reasons people trim a clip, that's the right tradeoff.