IELTS speaking practice with AI, out loud.
SpeakWisely lets you practise the IELTS speaking test by speaking, not typing. Clara, an AI examiner, asks the questions out loud, listens to your answers and runs all three parts in about fifteen minutes. At the end you get an estimated band on the four IELTS speaking criteria and a clear list of what to fix.
AI-estimated assessment inspired by IELTS speaking criteria. Not an official IELTS or IDP result.
What happens in each part of the speaking test?
The mock follows the three parts of the IELTS speaking test, in order.
Part 1: Introduction
Four short questions about you: home, work or study, everyday life. Clara asks each one out loud and you answer out loud.
Part 2: Long turn
A task card. You get one minute to prepare, with space for notes, then speak for up to two minutes. Afterwards the report checks how well you covered the card.
Part 3: Discussion
Four deeper questions that grow out of your Part 2 topic: opinions, comparisons, the future.
Practise with a tutor, or sit it like the real thing
Choose how Clara behaves before you start.
- Help me as I go
- Tips and corrections after each answer, and a Band 7+ version of any answer when you ask for it.
- Exam conditions
- Clara stays neutral, like a real examiner. Notes and corrections wait until the end.
In either mode you can ask her to repeat a question or explain a word. Asking never counts as an answer and is never scored.
What is in the band report?
Every mock and practice answer ends in a report built from what you actually said.
- Four bands and an overall band
- Fluency and Coherence, Lexical Resource, Grammatical Range and Accuracy, and Pronunciation, plus an overall band and a CEFR level.
- The evidence behind each band
- Every band lists what was measured, the band that measurement points to and how much it counts.
- Your mistakes, quoted
- Grammar and vocabulary issues are quoted from your own transcript. An issue that cannot be found in what you said is dropped.
- A plan and a model answer
- Focus areas, an improvement plan and a band-8 model answer you can read and listen to.
- A retry, compared
- Answer again and see what changed between the two attempts.
- Your transcript
- The words you said, as they were recognised. The audio itself is not kept.
How is the band worked out?
The AI model is not asked for a score. Your speech is measured, the model finds grammar, vocabulary and organisation issues as quoted evidence, and a fixed rubric turns that evidence into bands, so the same answer always gets the same band.
Measured directly from your speech: pace (words per minute), pauses, filler words, repetition, range of vocabulary, sentence complexity, linking words, how confidently your speech was recognised. Pronunciation is the hardest thing to judge from audio, so its band carries a confidence label. A session that covers only some parts of the test is capped lower than a full mock, because there is less evidence.
Which topics can I practise?
Choose a part, a difficulty and a topic. Questions are written fresh for each session.
Technology, Education, Business, Travel, Environment, Health, Lifestyle, Society, Entertainment, Science, Work, Relationships, Daily Life, Current Affairs.
Questions about IELTS speaking practice
- Is the band an official IELTS score?
- No. AI-estimated assessment inspired by IELTS speaking criteria. Not an official IELTS or IDP result. Use it to see where you stand and what to work on before the real test.
- Will my accent lower my band?
- No. Accent is never treated as an error. Pronunciation feedback looks only at how clearly your words came through.
- Are my recordings stored?
- Your recording is processed and then deleted. It is never stored — only your transcript and results are kept.
- Can I practise just one part?
- Yes. Pick Part 1, 2 or 3, or the full test, along with a difficulty and a topic. A single part takes a few minutes.
Want to work on fluency or grammar between mocks? See everyday English speaking practice and the English tenses guide, or compare plans and prices.