Calm, precise, and entirely focused on the format you are actually going to sit.
Oscar is an AI tutor personality in Enverson AI — a teaching style you select in the app. He is not a human teacher and he is certainly not an examiner; sessions with him are preparation, not assessment, and no result from a practice session has any official standing.
What he does have is the format. Speaking tests are not conversations, and learners lose marks for reasons that have nothing to do with their English: answering a two-minute task in forty seconds, drifting off the prompt card, treating a Part 3 question as a Part 1 question. Oscar keeps the shape of the test exactly, so the format stops being a variable on the day.
He runs the real format, in order. A session follows the structure of the test you name: the three parts of the IELTS speaking interview, with the one-minute preparation and the two-minute long turn, or the independent and integrated TOEFL tasks with their short response windows. Nothing is compressed for convenience, because the discomfort of the timing is part of what you are practising.
He keeps time visibly and says so. Oscar tells you how long you spoke, where you stopped early, and where you were still going when the window closed. Most learners have no idea what two minutes of continuous speech feels like until they have produced it a dozen times; after a week with Oscar, you do.
He teaches answer shapes. Every task type gets a reusable frame — position, reason, example, consequence — so you are never staring at an unfamiliar prompt with nothing to say. He also drills the small mechanical habits that hold an answer together: signposting, paraphrasing the question, giving yourself a legitimate second to think without dead silence.
He gives feedback against the criteria. Comments are grouped the way the marking is: fluency and coherence, lexical resource, grammatical range and accuracy, pronunciation. Rather than a general verdict you get the specific thing that pulled the answer down — a repeated linker, a missing example, a tense that collapsed under pressure — and what to do about it next time.
An illustrative example of Oscar's style — not a transcript of a real session, and not an official assessment.
Oscar: Part 2. Here is your card: describe a skill you would like to learn, why you want to learn it, how you would learn it, and how it would change your daily life. One minute to prepare, then speak for two minutes. Starting now.
You: (speaks for 1 minute 10 seconds, then stops)
Oscar: You stopped at one minute ten, so you left roughly half the task unused — and that alone limits what the fluency and coherence descriptor can credit. Looking at the card again: you covered why and how, but you never touched the fourth bullet, how it would change your daily life. That bullet was your extra fifty seconds.
You: I ran out of things to say.
Oscar: That is nearly always a structure problem rather than an ideas problem. Use the bullets as your sections and give each one a reason plus one concrete example — a specific morning, a specific person. Two other things: you used "and also" four times, which caps your range of linking, and "I would be able to" appeared three times where "it would let me" or "I could" would vary it. Same card, second attempt, full two minutes.
You: (speaks for 2 minutes 2 seconds, covering all four bullets)
Note the absence of a score. The feedback names the criterion, the cause and the fix — which is the part you can act on before the test.
Exam practice is demanding and it is rarely the whole of anyone's week. Change tutor whenever you want, and your level, lesson history, vocabulary and mistake memory follow you across. That is genuinely useful in the run-up to a test: if Oscar keeps flagging the same conditional structure in your Part 3 answers, a session with Cleo can take that structure apart properly, and Oscar will see the improvement in your next mock. After the exam, most learners drift towards Atlas or Mira — and the record of everything you fixed comes with you.
No. Oscar is an AI tutor personality inside Enverson AI — a teaching style you choose in the app, not a human teacher and not an examiner of any kind. Practice with him is preparation, and the feedback he gives is guidance rather than an official result. Only the test provider can score your exam.
He works in the IELTS and TOEFL speaking formats, including the three-part IELTS interview with the one-minute preparation and two-minute long turn, and the independent and integrated TOEFL speaking tasks with their short response windows. Tell him which test you are sitting at the start and he holds that format for the whole session.
He will not hand you a number, because a number from a practice session would be misleading. What he gives you instead is more useful: which of the assessment areas your answer was weakest in, the specific feature that pulled it down, and what to change on the next attempt. Bands come from examiners; the fixable detail comes from practice.
Run the real format as many times as you need, at whatever hour suits you, until the timing stops being the thing you are worried about.
Try Enverson AI Free