Part 1 is two strangers, one helping the other out of a jam. Play two short clips with no timer, decide your answer, then reveal the full transcript and the reasoning behind every option, right and wrong. This is the learning half of listening, not the timed exam.
Problem Solving is the first thing you hear in the whole exam, and it is built to be the friendliest. Two people who have never met fall into conversation because one of them is stuck: a bike that has died on a trail, a bus map that will not give up its secret. The other person offers to help, and the clip is the back and forth of that help being worked out. Because they are strangers, they spell things out for each other, and every explanation they give each other is an explanation given to you.
On the real 2026 test the conversation arrives in three short sections of roughly a minute each, and after every section two or three questions are read aloud, eight in all. You hear each question once, read its four options, and choose inside about thirty seconds before the next one appears. There is no going back up the list. That structure rewards a particular habit: listen to a whole section as a story first, and only then answer, rather than hunting for one fact while the speaker moves on.
The trap in Part 1 is that the person with the problem rarely names it cleanly. They describe a symptom, they worry aloud, they mention three things at once. It is the helper who tidies the problem into a sentence, usually by repeating it back, and that tidy sentence is what a general-meaning question is testing. If you answer from the first flustered thing the stuck person says, you answer the wrong question.
The two clips below are shorter than a full Part 1 and each is a single helper exchange rather than a three-section conversation, which suits a course better than an exam: one problem, cleanly turned over, with every answer explained. Learn the shape here, then take the full eight-question version under the clock when you are ready.
How the Problem Solving part is built on the timed 2026 test, so you know what these clips are preparing you for.
The single most useful move in Part 1 is to wait for the helper. The person in trouble opens the scene, but the helper is the one who states the problem in plain terms before offering a fix, and that restatement is where the general-meaning answers live.
Two of those deserve a second look. The condition rule matters because Part 1 is where relief makes you careless: the moment the bike restarts or the route becomes clear, your attention drops, and the test knows it, so the very next detail is often the one it asks about. Stay switched on through the happy ending.
The tone rule matters because Problem Solving is warm by design. These are polite strangers being kind to each other, and the feelings run under the surface. A candidate who only files away facts will nail the specific questions and miss the inference ones, which is exactly the split that separates a good listening band from a middling one.
Each clip opens in the player. Listen once or as many times as you like, answer the questions shown as text, then reveal the transcript and a reason for all four options on each question. Nothing here is timed and nothing is scored.
A woman's e-bike has quit partway along the Galloping Goose Regional Trail near Victoria, and a passing cyclist stops to help. The fix turns out to be almost embarrassingly simple, which is the whole point of the last question: how she feels once the bike wakes up.
A visitor is puzzling over the bus map at a downtown Victoria stop, and a local waiting for her own bus helps him work out how to reach Butchart Gardens. The help narrows from a route to a specific ticket, and the reason for that ticket is the detail most people miss.
On the real 2026 test the Problem Solving audio plays exactly once, starts by itself, cannot be paused, and the eight questions are spoken to you one at a time with about thirty seconds each. That is a memory task under pressure, and it is the right thing to rehearse close to test day.
This course does the opposite on purpose. You may replay each clip while you are still learning what to listen for, the questions are shown as text so you can reread them rather than holding them in your head, and instead of a score you get the full transcript and a written reason for every option once you reveal. Only the clip is audio here; on the real test the Part 1 questions are spoken, which is worth remembering as you practise.
One honest note on the voices. The accents you hear are North American, because there is no Canadian text-to-speech voice on any provider we can reach. The scripts are Canadian through and through, which is why the trail is the Galloping Goose and the ticket is a transit DayPass. When you want the real conditions, the timed listening tests put all six parts back under the clock.
The specific ways confident listeners still drop marks on the Problem Solving part.
The stuck person opens with worry, not a diagnosis. Grab that and you answer a general question with a symptom. Wait for the helper to name the actual problem, which for the e-bike is simply that it lost power.
Once the solution lands, attention drops, and the condition attached to it slips past. The DayPass is suggested for a reason, and that reason, not the ticket itself, is the question.
A correct answer paraphrases; it does not quote. If one option uses the speaker's exact words, be suspicious of it rather than comforted, because the test writer usually built it as a decoy.
Part 1 hides one or two inference questions about mood inside a factual conversation. If you only collect what was said and not how it was said, the surprise-and-relief question has nothing to bite on.
A dead e-bike sounds like a flat battery, so a candidate answers battery. The clip says otherwise. Answer from this conversation, not from what a bike problem is normally about.
On the real test the questions come after the audio, not during it. Splitting your attention to hunt for an answer mid-section is how you miss the line that actually settles it. Listen first, answer second.
In the first few seconds decide who has the problem and who is solving it. Everything the helper says is a candidate answer; the person in trouble mostly supplies context. Knowing which voice to trust for the fix saves you on every specific question.
You get paper on test day. Write the solution in two words and the condition beside it: bus, needs DayPass. That pairing is where the test loves to plant a question, and a scribbled note beats a fading memory.
The final line of a section often carries the mood: a laugh, a thank you, a sheepish admission. Do not mentally check out at the solution, because the inference question is usually built on that last beat.
This part mimics an overheard exchange between two people who have just met. A podcast interview or a help desk call trains the same ear: someone explains a problem, someone else reframes and solves it. Listen for the reframe.
When you reveal a clip, the value is not the letter of the answer, it is the reason each wrong option was written to catch a specific mishearing. Read all four notes. That is where you learn what the Problem Solving distractors are made of, so you stop falling for them.
Part 1 opens the listening section with a conversation between two people who do not know each other, one woman and one man, where one of them helps the other with a problem. On the real 2026 test the clip is split into three short sections and you answer eight questions, two or three after each section. It is the first and gentlest of the six listening parts.
Eight, and they are spread across three sections of audio, roughly three, three and two. Each is a multiple-choice question with four options and exactly one correct answer. In this practice course we show you two shorter worked clips instead, each with its own questions and full explanations.
On the real test the question is spoken. You hear it once, read the four options on screen, and have about thirty seconds to answer before the next question appears, in a fixed order. In this course the questions are shown as text so you can reread them, because there is no clock and the aim is to understand the answer, not to race it.
The two speakers in Part 1 are strangers, so they explain themselves as they go and you can lean on that. In Part 2 they already know each other and skip the introductions. The Part 1 helper usually restates the other person's problem in their own words before fixing it, which is exactly the sentence a general question is built on.
No. Nothing on this page is timed and nothing is scored, ever. You play a clip, choose an answer in your head, then reveal the transcript and the reasoning for all four options, including why each wrong one is the trap it is. Marking, and the CLB band that comes with it, belong to the timed listening tests instead.
The accents are North American, because no Canadian text-to-speech voice exists on any provider we can reach. The scripts, though, are written in Canadian English with Canadian settings, so the e-bike stalls on the Galloping Goose trail near Victoria and the visitor is trying to reach Butchart Gardens by transit pass. Everything here follows the 2026 CELPIP format.
Part 2 keeps two speakers but makes them old friends, which changes everything you listen for. Then step up to the parts with a single voice, and take the whole section under the real clock.
Two people who know each other, and the feelings sit under the arrangements.
Learn Part 2 → 3One speaker holds the prices and dates the other needs, and so do the distractors.
Learn Part 3 → 4A single news reader, facts in order, and the questions switch to sentence completion.
Learn Part 4 →All 38 questions across the six parts, played once under the real conditions.
Start a listening test →Play one Problem Solving clip, decide your answers with no clock running, then reveal the transcript and the reason behind every option. Both clips are free.
