What does sayit not do? The honest limits list

sayit scores pronunciation and fluency in real depth — but it doesn't score grammar, doesn't grade vocabulary, doesn't issue an official exam result, and isn't a substitute for a human teacher. Being precise about what it doesn't do is as important as explaining what it does, so here's the honest list, not a marketing page's careful silence around the gaps.

30-second version: sayit measures how clearly you produced sounds and how fluently you spoke — not whether your grammar was correct, not how wide your vocabulary is, not whether your IELTS or TOEFL content and structure were strong. Its exam band estimates are explicitly unofficial. It's not a real-time conversation partner with genuinely open-ended replies, not a human tutor, and there's no offline or on-premise deployment, no Android app, and no live-published date for the iPhone app.

Does sayit score my grammar?

No. Nothing in sayit's scoring pipeline evaluates whether your sentence structure, verb tense, or word order was grammatically correct — accuracy, fluency, and every other score are all built from how you produced sounds and how you paced your speech, never from the grammatical correctness of what you said.

Does it grade my vocabulary?

No. sayit doesn't assess whether your word choice was appropriate, sophisticated, or varied. A word pronounced clearly scores well on pronunciation regardless of whether it was the ideal word for the context — vocabulary range simply isn't a dimension this measures.

Is the IELTS or TOEFL score official?

No, and it's never presented as one. Estimated bands are built only from fluency and pronunciation — the criteria audio alone can genuinely assess — and every estimate carries an explicit label that it's unofficial. The real exams also grade vocabulary and grammar, which sayit doesn't touch; the band estimate's specific limits go into this directly.

Is sayit a replacement for a human tutor or teacher?

No. It's precise and consistent about the specific things it measures, and it never gets tired or impatient — but it can't hold a genuinely open conversation, can't explain a grammar rule in context, can't read your motivation or adjust its whole approach the way a good teacher does. It's a tool for a specific, narrow job: telling you exactly which sounds slipped and how fluently you spoke, repeatably and at any hour.

Is the conversation practice a real, open-ended AI chat?

No — it's a scripted roleplay with suggested lines you speak and get scored on, not a free-form conversation where you can say anything and get a genuinely dynamic reply. What conversation practice actually is covers the distinction directly.

Can our institute run sayit offline or on our own servers?

No — it's a hosted cloud service; audio is processed on sayit's servers to be scored, always. There's no offline mode and no on-premise install, covered honestly in the deployment model.

Is there an Android app?

Not currently. The web app runs fully in a mobile browser on Android today, which is the honest recommendation until a dedicated app exists.

Does my chosen accent (British, Australian, etc.) fully change what I'm scored against?

Not yet, in one specific sense: the canonical target your pronunciation is compared to is still American English by default today, even if you've switched your listening voice to British. How the TTS voice works explains this gap directly rather than leaving it implied.

The honest limits, at a glance

Try it

None of this is a reason to skip trying it — open sayit and see what it does well: read a line and watch exactly which sounds get flagged, in real IPA, in seconds.