Describe Image
- 25 seconds to prepare, then 40 seconds to speak.
- Machine-scored
- Original practice material
A bar chart compares total annual rainfall in four cities. City A has about 600 millimetres, City B about 1,800, City C about 450, and City D about 2,400. Each city's bar is split into a wet-season portion and a dry-season portion, and in City B and City D the wet season accounts for roughly four fifths of the total.
Press record, answer out loud, then press again.
One free take on this device · no account · no card
Phrases worth using: the clearest feature; more than five times; the second wettest; concentrated into a few months.
sayit has no image to show you, so the chart is written out above — the same information you would read off the screen. The analysis covers your delivery across the 40 seconds; whether you picked the right features is the content mark, and that is not something we can check.
What the machine listens for
- Content: whether you named the key features the image actually shows, rather than describing it in general terms.
- Oral fluency: 40 seconds of continuous speech with no long silences — this is where most candidates lose the marks.
- Pronunciation: sound-level accuracy while reading numbers and labels out loud, which is harder than it sounds.
Described in our own words from the published format. Pearson's scoring guides are copyrighted and are not reproduced here.
- Preparation
- 25 seconds
- Recording
- About 40 seconds
What this page practises
sayit has no image for you here, so each prompt below is a written description of a chart — the same information you would read off the screen. Speak from it for 40 seconds and the analysis reports pace, pauses, fillers and every sound. What it cannot tell you is whether you picked the right features, which is the content mark. An estimate from fluency and pronunciation only — not an official band, and not a mark for vocabulary or grammar.How to do Describe Image well
Open with what it is, in one sentence
'This bar chart compares rainfall in four cities across one year.' That sentence buys you five seconds and tells the scorer you understood the image.
Give the extremes, then the trend
Highest, lowest, and the direction of change. Three specific facts fill the middle of the recording and are what the content mark is looking for.
Keep talking to the end
Fluency is scored across the whole 40 seconds. If you run out of features, describe what you would expect next or restate the overall pattern.
Practise saying numbers
Percentages, years and large figures are where fluent candidates suddenly stumble. Say them out loud before test day.
Common Describe Image mistakes
Describing the image instead of reading it
'There are some colourful bars' scores nothing. Name the values.
Finishing at twenty seconds
Half the recording empty means half the fluency evidence missing.
Long silence at the start
Have a first sentence ready before the microphone opens; the template is the same every time.
Reading every data point
Selecting matters. Two or three significant facts beat eleven rushed ones.
Straight answers
Do I need to interpret the image or just describe it?
Describe it accurately and state the obvious pattern. A long interpretation is not required and takes time from the features the content mark is counting.
What if I do not understand the chart?
Say what you can see — the title, the axes, the largest value — and keep speaking. A partly correct description with good fluency scores considerably better than an accurate one delivered in fragments.
Is a memorised template allowed?
An opening sentence structure is fine and sensible. A whole memorised answer is not, because the content mark depends on the specific image, and generic sentences earn nothing.
Hear the sounds an automated scorer hears
Record one item and read the per-sound breakdown — the endings and clusters a recogniser marks as a mismatch. Free, nothing blurred.