Speaking Mock Test 1
Start here: advise your cousin on her first laptop, describe a busy farmers market, choose a bike for your family, and describe a very strange floor lamp.
Free CELPIP speaking practice for the 2026 format. Six complete speaking tests, each with all 8 tasks on the official prep and recording clock. You speak into your microphone, we transcribe every answer, and you get expert feedback on all four criteria with a CLB estimate.
Six full-length CELPIP speaking tests, all free. Every test runs the same 8 tasks on the same per-task clock, and they are graded Easy to Hard so you can start where you are and work up.
Start here: advise your cousin on her first laptop, describe a busy farmers market, choose a bike for your family, and describe a very strange floor lamp.
An easy sitting: advise a friend on a winter coat, describe a busy community pool, argue for a community centre class, and explain a fishbowl clock.
A busy family restaurant, a used car your partner disagrees about, a fire hall open house, and a job the neighbours built out of whatever they had.
A crowded hardware store, a venue for your team celebration that your co-worker disputes, a deck blocked by a tree, and an unnameable piece in a gallery.
A crowded airport check-in hall, a volunteer project another supervisor disputes, a print shop order gone wrong, and a broken hand-cranked machine at a flea market.
A hotel lobby during a conference, a weekend route your friend wants to paddle instead, an evening shift two people both need, and a nameless machine at a flea market.
The 2026 CELPIP Speaking section keeps its 8 tasks, its per-task clock and its CLB 1 to 12 scoring. Here is the format at a glance so nothing on test day feels new.
Because the clock resets at every task, a strong Task 1 buys you nothing on Task 8. Every one of our speaking tests runs on that same per-task timer, which is the single most useful thing to rehearse.
CELPIP Speaking in the 2026 format is 8 tasks and roughly 15 minutes, and it is the last section you meet on test day. The first thing to understand is that there is no interviewer. You sit at a computer wearing a headset, a prompt appears on screen, a preparation clock runs down, and then a recording clock starts and you talk into a microphone. Nobody nods along, nobody asks a follow-up question, and nobody rescues a sentence that has gone wrong. That silence is the part candidates most often say they were unprepared for, and it is the best reason to rehearse on a real recording clock rather than in front of a patient friend.
The clock is set per task and resets on every advance. Preparation is 30 seconds on most tasks and 60 on the two hardest, and it is not skippable: you cannot bank it, and you cannot borrow it from a task you finished early. When the preparation window closes the recording begins whether or not you have a plan. Speaking to the end of the recording window matters just as much, because length is part of the marking. Trailing off after 20 seconds of a 60 second task leaves marks on the table even when everything you said was correct.
The 8 tasks are deliberately different jobs. Some are personal, like advising someone or telling a story from your own life. Some are visual: Tasks 3 and 4 share one picture, first asking you to describe it and then asking what will probably happen next in that same frozen scene. Some are social and test your tone as much as your grammar, like persuading someone to accept your choice or handling an awkward situation without causing offence. The last one hands you something with no ready-made name and asks you to describe it anyway, which is really a test of how you build language when the obvious word is missing.
Scoring is where speaking differs most from reading. There is nothing to add up. Each of the 8 tasks is rated on the same four criteria, and your reported level from CLB 1 to 12 comes out of all 8 together, so one bad task cannot sink you and one brilliant task cannot carry you. It is worth reading those four criteria before you practise, because they explain why two candidates who noticed exactly the same things in a picture can land three levels apart. All 6 of our CELPIP speaking practice tests are free, and every answer you record is transcribed and returned with expert feedback on each criterion and a CLB estimate.
Every sitting runs these 8 tasks in this order, with the preparation and recording times shown. The clock resets at each one.
| Task | What you do | Prepare | Speak |
|---|---|---|---|
| Task 1 | Giving Advice | 30 sec | 90 sec |
| Task 2 | Talking about a Personal Experience | 30 sec | 60 sec |
| Task 3 | Describing a Scene | 30 sec | 60 sec |
| Task 4 | Making Predictions, on the same picture as Task 3 | 30 sec | 60 sec |
| Task 5 | Comparing and Persuading, plus 60 sec to choose | 60 sec | 60 sec |
| Task 6 | Dealing with a Difficult Situation | 60 sec | 60 sec |
| Task 7 | Expressing Opinions | 30 sec | 90 sec |
| Task 8 | Describing an Unusual Situation | 30 sec | 60 sec |
Task 5 runs across three screens: the instructions, a 60 second screen where you choose your option and do not speak at all, then the preparation and recording windows for persuading. If you choose nothing, the computer chooses for you.
Someone you know has a decision to make and wants your recommendation. The longest recording window in the test opens the section, which is generous but unforgiving: 90 seconds is a long time to fill if your advice is one sentence and no reasons behind it.
Tell a story from your own life that fits the prompt. The trap is spending 40 seconds on background and never reaching what actually happened. A clear shape, what happened and why it mattered, beats a colourful opening every time.
Describe an illustrated scene to a listener who cannot see it. Because they cannot see it, where things are matters as much as what they are. Naming eight objects scores worse than placing four of them in relation to each other.
The same picture returns and you say what will most probably happen next. Everything must be answerable from the frozen frame, so look for the person caught mid-action who has not been answered yet. That is where your predictions live.
You choose between two options, then someone close to you counter-proposes a third one you never saw. You persuade them your choice is better by comparing it to theirs, not to the option you already rejected. That rejected option disappears and is never mentioned again.
An awkward situation with two people you could speak to, and you pick one. Text only, no pictures. This is the tone task: the same words that work on a friend will cost you marks with a supervisor, and tone is explicitly marked.
Take a position on an everyday issue and defend it for 90 seconds. There is no right side. What is marked is whether you commit to one, support it with reasons, and organise those reasons instead of circling the same point three times.
You describe something to a named person over the phone that has no obvious name, and you finish by asking them for something. The closing request is part of the task and it is the beat candidates most often forget. It is really a test of describing without the easy word.
A speaking test you cannot hear back is just a rehearsal. Every answer you record here is kept, written out, and rated against the four official criteria.
Your browser asks for the microphone once, then the test runs the real prep and recording clocks, task after task, exactly as the exam does.
Each recording is written out word for word, so you can read back what you actually said rather than what you remember saying. The two are rarely the same.
Content/Coherence, Vocabulary, Listenability and Task Fulfillment, the same four the exam uses, with expert feedback naming what cost you marks on each.
A level from CLB 1 to 12 across all 8 tasks, plus a replay so you can listen back to any task beside its transcript and its feedback.
Your CLB estimate is for practice. Only Paragon can issue an official CELPIP result.
What actually moves a speaking level, based on what the four criteria reward.
Thirty seconds of preparation feels generous until it is running. Practise with the timer on, from the very first test, so that the silence and the countdown are familiar rather than alarming. Speaking well when nothing is at stake proves very little.
Length is marked under Task Fulfillment. If you finish early, add a reason, an example or a consequence rather than stopping. Practise having one more thing to say, because the last 15 seconds of a task are where most candidates go quiet.
You cannot predict the prompts, so do not try. What transfers across all 8 tasks is the equipment: precise words instead of vague ones, clauses that join ideas, hedging when you are unsure, and a tone that fits who you are speaking to.
Listenability rewards control first. A simple sentence you finish cleanly beats a complex one that collapses halfway. Add longer structures only once the shorter ones are reliable, and self-correct briefly rather than restarting the whole thought.
Tone is scored. Your supervisor, your friend and your neighbour do not get the same sentences, and the higher bands are explicitly about adjusting to the listener. Before you speak, name the person in your head and let that set your register.
The most uncomfortable and most useful step. Read your transcript while the audio plays. The filler words, the abandoned sentences and the word you reached for three times are all obvious on the page and invisible in the moment.
Eight tasks, eight different jobs. Work through the one you freeze on, with sample answers for each.
Advise one person on one decision, in 90 seconds.
Practise this task → Task 2One story, told with enough detail to fill the time.
Practise this task → Task 3Describe a picture to someone who cannot see it.
Practise this task → Task 4Say what happens next in the same scene, and why.
Practise this task → Task 5Pick one of two options and sell it to a named person.
Practise this task → Task 6The hardest task: choose who to disappoint, then say it well.
Practise this task → Task 7Take a side on a statement and hold it for 90 seconds.
Practise this task → Task 8Describe an object with no obvious name, precisely enough to be understood.
Practise this task → SamplesSample answers across the 8 tasks, with what each one does right.
Practise this task →Your four section levels stay separate on your score report, so a strong speaking level cannot rescue a weak reading one. Every section is free and built for the 2026 format.
Problem solving, daily life, information, news, a discussion and viewpoints across all 6 parts, plus an unscored practice task.
Start listening →All four reading parts with instant CLB scoring and a rationale on every answer.
Start reading →Email and survey tasks in a timed editor, returned with expert feedback.
Start writing →Full-length CELPIP mock tests across all four skills in one place.
View mock tests →Start with Test 1 if the recording clock is new to you, or jump to Test 6 if you want the hardest set we have. All six are free, and every task comes back transcribed with expert feedback.