How does an app tell a natural pause from a problem pause?
Not every gap in your speech is a mistake. sayit judges a pause against what the punctuation expects, so the same length gap can score two different ways.
By checking where the pause happened, not just how long it lasted — the same half-second of silence is completely normal after a full stop and reads as hesitation in the middle of a connected phrase, so location matters as much as duration.
30-second version: sayit compares the actual gap between two words against an ideal gap that depends on the punctuation immediately before it: roughly 0.06 seconds mid-phrase, 0.25 seconds after a comma, 0.45 seconds after a sentence-ending period or question mark. A gap that runs longer than that ideal by more than a small tolerance gets marked "dragging"; a gap that's about right gets marked "good." The goal was never zero pauses — a take with no pauses anywhere, including after full stops, usually scores worse, not better.
Why "fewer pauses" is the wrong instinct
A common but backwards assumption is that pausing is a weakness to eliminate. In fact, the absence of any pause at all — running straight through commas and periods without a beat — is its own problem, covered from the opposite direction in why my pace is 'a bit fast'. The two problems are mirror images: rushing means your gaps come in shorter than the punctuation expects; dragging means they come in longer. Both get measured against the same ideal, just on opposite sides of it.
Where dragging tends to show up isn't usually after punctuation — an extra-long pause after a period is often unremarkable, even sounding deliberate and confident. It's the mid-phrase pauses, the ones with no punctuation cue at all, that read as hesitation when they run long: a gap in the middle of "I went to the... store" where nothing in the sentence structure calls for a break there.
How sayit tells the two apart
Every adjacent word pair gets its own ideal gap based on the punctuation boundary that precedes it, and the actual gap is compared against that specific ideal rather than a single flat threshold applied everywhere. A 0.4-second gap after a period sits comfortably within the expected range and scores as good; that same 0.4-second gap in the middle of an unpunctuated phrase is nearly seven times the mid-phrase ideal and gets flagged as dragging. Same duration, opposite verdict, because the position tells a completely different story about what's happening.
What causes mid-phrase dragging
The most common cause is word retrieval — reaching for a word or a grammatical structure mid-sentence and stalling before it arrives. This is different from a nervous, filler-heavy hesitation (covered in how many filler words is too many) because there's often no "um" at all, just silence, which can be harder to notice in yourself than a filler sound would be.
The fix isn't to force a faster pace generally — it's to plan further ahead. Before starting a sentence, have the whole clause roughly formed rather than starting to speak and hoping the ending arrives in time. Reading aloud from a prepared passage removes this kind of pause almost entirely, which is part of why reading aloud and speaking freely find different mistakes — freeform speech is where dragging pauses actually show up.
Drill it
Record a short unscripted answer — describe your weekend, explain a simple process — rather than reading a fixed sentence, since dragging pauses are a freeform-speech pattern that a memorized sentence won't reliably reproduce. Compare the tempo takeaway across two takes of the same prompt: one where you plan the first sentence before starting, one where you start speaking immediately.
How do I know it's improving?
Look for dragging flags to cluster less around mid-phrase positions and shift, if anywhere, toward the punctuation boundaries where longer pauses are actually expected and scored fine. A take with zero dragging flags anywhere, including after full stops, is worth double-checking — it can mean you've overcorrected into rushing through the pauses that were supposed to be there.
Try it
Open sayit free — no install, no card — and record a short unscripted answer to a simple question. Check whether any dragging flags land mid-phrase rather than after a period or comma; that position is the actual signal.
Hear exactly which sounds to fix.
Say one sentence and get sound-by-sound feedback in seconds. No install, no card.