English speaking assessment
Measure your English.
About twelve minutes, start to finish. Get a clear report that names what makes listeners work harder — and what is already fine.
- We measure. Six areas, fourteen items, each one calculated the same way every time.
- We name it. Not “work on your pronunciation” — which sounds, in which words.
- We do not teach. No lessons, no drills. You take the result wherever you like.
No account needed to record. Scoring costs 1 credit — packs from $20.
Most people stop improving because nobody tells them what is wrong
“Your English is great”
Kind, but it does not help. There is nothing in it you can use.
You feel something is wrong
People ask you to repeat yourself. Meetings take more effort than they should.
So you practise everything
Years of work, most of it on things that were never the problem.
What we measure
Six areas, and fourteen smaller items inside them. Each one is calculated in a fixed, written-down way. Being this specific is the whole point: a result you can act on has to name something.
Pronunciation
Phonemic Accuracy · Word Stress
Intonation
Pitch Variation · Terminal Contours · Uptalk
Sentence Stress
Focus Words · Contrastive Emphasis
Rhythm
Stress-Timing · Connected Speech (both observed)
Fluency
Speech Rate · Disfluency Rate · Pause Placement
Clarity
Articulation · Vocal Stability (observed)
Some things are described but not given a level: your rhythm, how loudly you project, and how you change your style between formal and informal speech. Tested against native speakers, our rhythm measures varied as much for them as for learners; your microphone affects projection, and the situation affects style. A level would not mean much, so we say that rather than pretend.
How it works
Five short parts. Each one is there because something can only be measured that way.
1. The Mirror 55s
Every English sound appears in this paragraph. That is not an accident — it is the only way to check all of them in under a minute.
2. The Exchange 40s
Questions go up, statements go down. Getting this backwards is the most common reason capable speakers are misread as uncertain — and it cannot be seen in a paragraph with no questions in it.
3. The Emphasis Drill 60s
Same words, four meanings. Because the sounds are identical across all four readings, any difference we measure is purely emphasis — the cleanest possible test.
4. Off the Page 90s
Reading hides things. It suppresses hesitation and hands you someone else's rhythm. When you choose your own words, your real fluency shows up — and that is what listeners actually hear.
5. Two Rooms 60s
Sounding right at work and sounding right with friends are different skills. Register can only be measured as a difference, so we hold the news constant and change who is listening.
Nine more test sets, after your first one
The five parts above are the Core Diagnostic. It is the one we calibrated, and the right place to begin. When you sign in, nine more are available:
- A second version of the core test. It measures the same things using different words, so when you test again in three months we measure your speech — not how well you remember the paragraph.
- Focused sets that go deeper on one thing than twelve minutes allows: vowel sounds, word endings, intonation, rhythm, speaking under pressure, formal versus informal speech, and interview speech.
- A four-minute short test, if you want to see what this is like first.
Every set says what it measures and what it leaves out. A set with no conversation in it cannot measure your intonation, and it says so instead of quietly skipping it. Because the sets are different, we only ever compare a result with a result from the same set.
What this is not
This is not a course. We do not teach you, give you exercises, or coach you. We measure carefully, once, and give you a map. What you do next is up to you — but if you hire a teacher afterwards, you will know exactly what to ask them for.
How we avoid making things up
Your scores are calculated from your recording by software that measures sound, plus a pronunciation model. They are not guessed by an AI. The same recording always gives the same result.
An AI writes the sentences that explain your result — and our code stops it from mentioning any number we did not actually measure.
When we cannot measure something, the report says so and gives the reason. Levels are marked provisional until our study of how reliable they are is finished. See how we measure →