Word-by-word feedback
Every word in the phrase is marked: landed, close, or missed — with what the recogniser heard instead. Students see the specific sound to work on, not a vague score.
Speaking practice for every classroom
Some students know the answer. Writing it down is the part that gets in the way.
SayAble gives them another way to show it — out loud. Students hear exactly which words landed, practise as many times as they like, and never get handed an automatic grade.
No account needed to try it. Works in the browser on a laptop, tablet or phone.
🔥 3
↻ 7
Tongue twisters · 3 of 5
Say this out loud
Three
thin
thistles68
thawed
in
the
thicktick
fog.
80
accuracy
fluency
completeness
prosody
→ “thick” came through as “tick” — the th is the tricky part.
An assessment sometimes measures more than you intended
Ask a student why photosynthesis matters and you want to know whether they understand photosynthesis. What you often measure instead is how fast they can start writing, organise paragraphs, spell, type, hold attention, and turn an idea into written language.
Some of that is relevant. Not all of it is what you were trying to assess.
“Did the student not know the answer — or did the answer fail to show what the student knew?”
Features
What students and teachers actually get
Real pronunciation scoring
Accuracy, fluency, completeness and prosody, down to individual phonemes, powered by Azure AI Speech. Detects omissions and mispronunciations, not just whether a word was recognised.
Six languages
English, Spanish, French, Mandarin, Hindi and Russian — for world-language classes and for multilingual learners showing what they know in the language they think in.
No automatic grades
SayAble never produces a grade. It shows evidence of an attempt and leaves the academic judgment where it belongs — with the teacher.
Unlimited retries
No hearts, no lives, no penalty for trying again. Progress is measured against the student's own previous attempts, never a native-speaker baseline.
Teacher-built sets
Type your phrases, one per line — vocabulary for a unit, lines for a presentation, sentences a student is working on with their SLP. Ready in under a minute.
How it works
Three steps, about a minute each
-
1
A teacher makes a set
Give it a name and paste in phrases, one per line. Pick the language students will speak in.
-
2
A student says them out loud
Tap the mic, speak, tap again. A live level meter shows the mic is actually hearing them — no guessing.
-
3
They see what landed, and try again
Per-word marks, pace, and the sounds worth another go. At the end, a recap they can hand to a teacher or SLP.
Who it's for
One tool, three very different jobs
Students
- Show what you know without fighting a blank page
- Practise a word until it feels easy — no one is counting your mistakes
- Rehearse the sentences you need for a presentation
- Choose not to speak; nothing here is compulsory
Teachers
- Build a speaking set in under a minute
- See attempts and evidence — you make the call, always
- Works for world languages, vocabulary, and oral rehearsal
- Nothing to install; it runs in the browser
Schools & districts
- Canadian data residency available — audio stays in Canada
- Clear answers for a privacy impact assessment
- Retention and deletion you configure, not us
- No grades produced, so no automated academic decisions
Pricing
Start free, pay when a school adopts it
Illustrative pricing during the beta — nothing here is a commitment yet.
Classroom
Free
One teacher trying it out
- Up to 30 students
- Starter sets, plus your own
- All six languages
- Word-match feedback
- No account required
School
$5/student/year
A department or a whole school
- Everything in Classroom
- Full pronunciation scoring — accuracy, fluency, completeness, prosody, phonemes
- Unlimited sets and classes
- Progress records and export
- Email support
District
Custom
Multi-school rollouts
- Everything in School
- Single sign-on
- Canadian data residency
- DPA and privacy-assessment support
- Onboarding and PD sessions
Evidence
We'll tell you exactly how much we've proven
Plenty of education tools claim results they haven't measured. Here is the ladder we hold ourselves to, and where we honestly are on it today.
- Personal experience“This helped me.” Valuable, not proof.
- Research rationaleExisting research says the problem is real and worth investigating.
- Prototype evidence You are hereThe technology works end to end. That's all we claim so far.
- Pilot evidenceEarly users find it useful in real classrooms.
- Controlled researchMeasurable effects under defined conditions.
- Independent replicationOther researchers find the same thing.
So: no case studies yet, because we haven't run pilots yet. If you'd like to be one, that's the most useful thing you could offer us.
About
Where SayAble came from
There's a moment that happens in classrooms everywhere. A teacher asks a question. The student understands the idea — you can see it. They might explain it perfectly to a friend at lunch. Then the assignment comes in short, disorganised, or barely there.
SayAble began with that gap, and with first-hand experience of it: knowing exactly what you want to say, and finding that getting it from your head onto the page is the hard part.
It isn't built on the belief that speaking is better than writing. When writing is the skill being assessed, writing should stay. It's built on a narrower question: when writing isn't the skill you're measuring, could some students benefit from another way to show what they know?
What we won't do
- Ask teachers to lower their standards
- Claim every student benefits from the same accommodation
- Claim speaking is always better than writing
- Ask you to trust an AI score you can't inspect
- Pretend a personal story is scientific proof
- Hide the limits of our technology
FAQ
Questions teachers ask first
Does the AI grade my students?
No. SayAble does not produce a grade at all — not a hidden one, not a suggested one. It shows what a student attempted and how it came through. Every academic judgment stays with you.
Will this create more work for me?
It shouldn't. Building a set is typing phrases one per line. Students practise on their own, and the recap is a short summary rather than a stack of recordings to sit through. If it adds workload in a pilot, we'd want to know — that's a product failure, not your problem.
Can I still require writing?
Absolutely. If writing is the learning objective, writing should remain part of the assessment. SayAble is for the cases where the format is hiding understanding rather than measuring it.
What about students with an accent, or a stutter?
An accent is never treated as an error. Attempts are compared to the phrase and to the student's own previous attempts — never to a native-speaker baseline, which would penalise exactly the students this is meant to help. For students working on disfluency, the useful output is practice attempts and a record to share with a speech-language pathologist, not a fluency score.
What happens if the transcript is wrong?
Recognition gets things wrong, and the interface assumes it. Students see what was heard, can tell the difference between "I said it differently" and "it misheard me", and simply try again.
Does this replace an IEP or speech therapy?
No. SayAble is a tool, not a plan and not a clinician. It doesn't diagnose anything and it isn't a substitute for professional educational planning or speech-language pathology.
Where does student audio go, and how long do you keep it?
Speech is sent to a speech-recognition service to be assessed, and can be processed in a Canadian region so it stays in Canada. Retention is a setting your school controls, not a default we impose. We'll give you the data-flow document your privacy officer needs, in writing.
What if a student doesn't want to speak out loud?
Then they shouldn't have to. Choice matters, and a format that creates a bigger barrier than the original task isn't an accommodation.
Let students become SayAble.
Try a set yourself — it takes about a minute, and you don't need an account.