sayit vs ELSA Speak (2026): diagnosis or drills?
sayit vs ELSA Speak, compared honestly: feedback depth, your own content vs 8,000 fixed lessons, accent philosophy, exam prep, classrooms, price and privacy.
ELSA Speak and sayit are both serious pronunciation tools — both score individual sounds, both cover IELTS/TOEFL, both sell to schools. But they are built around opposite ideas of what feedback is for. ELSA is a gamified lesson library that grades you; sayit is a diagnostic coach that tells you what you said, what you should have said, and how to close the gap — on any text you choose.
We build sayit, so read this as an informed but interested comparison. Every claim about sayit is testable in a free session, and we hedge every claim about ELSA to well-known public facts.
The head-to-head at a glance
| sayit | ELSA Speak | |
|---|---|---|
| Feedback | Per-phoneme heatmap + target IPA vs your actual sound + articulation tip | Per-phoneme percentage scores |
| Content | Your own paragraphs and PDFs, AI-generated passages, 500+ leveled texts | 8,000+ fixed lessons |
| Accent goal | Clarity in your own accent | Drill toward a target (US-leaning) accent |
| Exam prep | Full IELTS/TOEFL-style speaking exams | IELTS/TOEFL lesson content, CEFR prediction |
| Schools | Teacher CMS, class assignments, gradebook, parent reports | ELSA for Schools with a teacher dashboard |
| Platform | Browser — no install | Mobile-first |
| Free tier | Yes, no card | Limited daily allowance |
| Price | Free tier; paid plans on pricing | Pro ~$12/mo (as of 2026) |
| Privacy | Recordings processed for feedback only; on-premise option | Standard cloud processing |
Feedback depth: a diagnosis vs a score
This is the core difference, so it deserves the most space.
ELSA's scoring is real — it evaluates individual phonemes and colors them, which already beats most apps. But the unit of feedback is a percentage. Your /ɪ/ scored 71%. What now? A number tells you that something is off, not what happened or what to change.
sayit's unit of feedback is a correction. You record a sentence and get a color-coded per-word, per-phoneme heatmap; open any weak sound and you see the target IPA next to the sound you actually produced, plus a concrete articulation tip — "your /iː/ in sheep came out as /ɪ/; spread your lips and lengthen the vowel." That is the difference between a thermometer and a doctor.
The gap is architectural, not cosmetic. sayit's recognizer is a language-model-free phoneme recognizer: it scores raw sounds instead of auto-correcting them into dictionary words the way transcription-oriented speech recognition does. The full explanation is in how AI pronunciation scoring works and the autocorrect problem.
On top of the per-sound view, sayit also scores fluency, completeness, stress and intonation per recording — the prosody layer covered in word stress and intonation.
Content: your material vs ELSA's library
ELSA's 8,000+ lessons are professionally built and cover all 44 phonemes, ordered into courses. If you want a curriculum handed to you, that library is a genuine strength — it is bigger than anything sayit ships.
The limitation is that they are ELSA's sentences. You cannot drill your conference talk, your standup update, your dissertation defense. sayit inverts this: paste any paragraph, import a PDF of your own material, generate AI practice passages on any topic, or use the 500+ leveled built-in texts. Four modes — read aloud, shadowing, dictation, freeform — run over whatever content you loaded. And the Magic Wand coach watches your history, auto-diagnoses your weakest sounds, and builds a personalized drill plan, so you get curriculum-like structure without a fixed curriculum.
Accent philosophy: clarity vs target accent
ELSA descends from the accent-training tradition: there is a reference pronunciation (US-leaning), and drills move you toward it. Nothing wrong with that goal if it is your goal.
sayit's position is different: your accent is not an error. The scorer is deliberately accent-tolerant — it ignores regional variation (an Indian retroflex /t/, a Spanish tapped /r/) and enforces only the contrasts that change meaning for listeners: ship/sheep (/ɪ/ vs /iː/), rice/lice (/r/ vs /l/), sink/think (/s/ vs /θ/). You keep your identity; you lose the misunderstandings. The reasoning and the linguistics are in accent-tolerant pronunciation AI.
Which philosophy is "right" depends on you. If your job interview is in Ohio and you want to sound local, ELSA's framing (or BoldVoice's, even more so) fits. If you want to be effortlessly understood without erasing where you are from, sayit is built for exactly that.
Exam prep: both real, differently shaped
Both apps take IELTS and TOEFL seriously. ELSA offers exam-oriented lesson content and CEFR level prediction, which is useful for placing yourself. sayit ships full exam simulations — complete IELTS-style and TOEFL-style speaking tests, scored per-phoneme like everything else, so your exam practice feeds the same weakness data as your daily drills.
Classrooms: both have school offerings
Honest note: ELSA for Schools is a real product with a teacher dashboard and the weight of a 25M-user brand behind it — schools should evaluate it.
sayit's Teams side is a full CMS rather than a dashboard: roster import and join codes, classes and batches, teachers writing custom paragraphs and assigning to a whole class in one click (or differentiating per batch), an automated gradebook with completion rates and average scores plus CSV export, automated guardian email reports, and strict row-level isolation so teachers only see their own students. It is FERPA-ready and can run cloud or fully on-premise. Details in pronunciation software for language schools — or contact us for a pilot.
Price, free tier, privacy
ELSA Pro runs ~$12/mo as of 2026; the free tier is a thin daily sampler. sayit's free tier requires no card and is enough to practice on, with paid plans listed on pricing.
On privacy: sayit processes recordings only to produce your feedback — they are never sold or harvested — and institutions that cannot send student audio to a third-party cloud can deploy it fully offline / on-premise. If you practise sensitive work material aloud, that difference is not academic.
Choose ELSA Speak if…
- Gamified mobile habit-building works on you. Streaks, points and a polished daily-lesson loop are ELSA's superpower, and consistency beats everything else in pronunciation.
- You want the biggest fixed lesson library — 8,000+ professionally sequenced lessons is more ready-made curriculum than anyone else offers.
- You are happy drilling toward a US-reference accent inside content ELSA chose for you.
Choose sayit if…
- You want a diagnosis — which phoneme, what you said instead, how to fix it — not a percentage.
- You want to practise your own material: PDFs, scripts, slides, plus AI-generated passages.
- You want clarity in your accent, not accent replacement.
- You care about privacy or need an on-premise deployment for a school.
- You want a real free tier in the browser, no install, no card.
Run the two-minute test
Say "I sink the answer is right" into both apps. One will hand you a number. The other will show you the exact spot where /θ/ became /s/ and tell you where to put your tongue. For more alternatives beyond these two, see ELSA Speak alternatives.
Start free on sayit — it runs in your browser, no card required.
Hear exactly which sounds to fix.
Say one sentence and get sound-by-sound feedback in seconds. No install, no card.