New Score Task models for PTE Describe-Image and Retell-Lecture — now with Score Explanation

A number between 0 and 5 tells a teacher a learner scored 2. It does not tell them that the learner named one country out of five, never compared any of them, and never drew a conclusion. That is the part a learner can act on, and until now you had to infer it yourself.

Today we are releasing new scoring models for describe-image and retell-lecture, and every scored response now comes back with a score_explanation: a short, plain-language account of what the response covered, what it missed, and which omission set the band.

What’s new

• New models for describe-image and retell-lecture (task_score.version 0.3). Both make fuller use of the 0–5 scale: the intermediate bands 1 and 3 are now used where the previous model tended to round to 0, 2 or 4, so partial responses are separated from empty ones and from strong ones.
• score_explanation on every response for these two task types. No new parameter, no extra call, no change to your request — if you are already calling Score Task, it is in the response.
• answer-question is unchanged and does not return score_explanation. Its score is binary (0 or 1), so there is little to explain.

Example 1 — Describe Image, a complete response

The task_context is the model description of the image: write it as the answer that would itself earn a 5. The model scores the candidate’s response against it.

Request

POST https://api.speechace.co/api/scoring/task/v9/json?key=YOUR_KEY&dialect=en-us&task_type=describe-image

task_context  = "The bar chart compares how five countries generated their electricity in 2020, split between renewable and fossil-fuel sources. The clearest feature is that Norway produced almost all of its electricity from renewables, far more than any other country. In contrast, Poland and India relied overwhelmingly on fossil fuels. Germany and the UK sat in the middle, each with a fairly even split. Overall, the chart shows a strong divide between a few renewable leaders and the fossil-fuel-dependent countries, suggesting that national energy policy, not just resources, drives the mix."
user_audio_text = "The bar chart compares electricity generation in five countries in 2020, renewable against fossil fuel. Norway is the clear outlier, producing almost all of its electricity from renewables, while Poland and India rely overwhelmingly on fossil fuels. Germany and the UK sit between them with a fairly even split. Overall the chart shows a sharp divide between a few renewable leaders and the fossil-dependent countries, which suggests policy rather than natural resources drives the mix."

Response

{
  "status": "success",
  "task_score": {
    "type": "describe-image",
    "version": "0.3",
    "score": 5,
    "score_explanation": "The response accurately describes all key features of the image, including the standout performance of Norway, the heavy reliance on fossil fuels in Poland and India, and the balanced approach in Germany and the UK. It also identifies the major trend of a divide between renewable and fossil-fuel-based electricity generation and offers a logical conclusion that policy, rather than natural resources, influences the energy mix. The response is concise but comprehensive, with no missing elements or inaccuracies.",
    "transcript": "The bar chart compares electricity generation in five countries in 2020, renewable against fossil fuel. Norway is the clear outlier, producing almost all of its electricity from renewables, while Poland and India rely overwhelmingly on fossil fuels. Germany and the UK sit between them with a fairly even split. Overall the chart shows a sharp divide between a few renewable leaders and the fossil-dependent countries, which suggests policy rather than natural resources drives the mix."
  },
  "quota_remaining": -1,
  "version": "9.17"
}

The explanation names the specific features the response covered, the trend it identified, and the conclusion it drew, which is why this one earns the top band rather than a 4.

Example 2 — Retell Lecture, a partial response

This is where the explanation earns its place. The same lecture, a much thinner retelling:

Request

POST https://api.speechace.co/api/scoring/task/v9/json?key=YOUR_KEY&dialect=en-us&task_type=retell-lecture

task_context  = "Today's lecture is about photosynthesis, the process plants use to make their own food. The speaker explained two main stages. First, in the light-dependent reactions, chloroplasts capture sunlight and split water, releasing oxygen. Second, in the Calvin cycle, the plant uses that captured energy to turn carbon dioxide into glucose. The lecturer stressed that the two stages are linked: the energy carriers produced in the first stage power the second. She concluded that photosynthesis is the foundation of most food chains because it converts sunlight into stored chemical energy."
user_audio_text = "The lecture was about photosynthesis. The speaker talked about how plants use sunlight and water, and said something about carbon dioxide and glucose. There are two stages in the process."

Response

{
  "status": "success",
  "task_score": {
    "type": "retell-lecture",
    "version": "0.3",
    "score": 2,
    "score_explanation": "The response mentions the general topic (photosynthesis) and the presence of two stages, but lacks specific details about the stages (e.g., light-dependent reactions and the Calvin cycle). It also fails to explain the relationships between the main ideas, such as how the energy from the first stage powers the second. Additionally, it omits the conclusion about photosynthesis being the foundation of food chains. The response lists isolated details without connecting them into a coherent summary of the lecture's argument.",
    "transcript": "The lecture was about photosynthesis. The speaker talked about how plants use sunlight and water, and said something about carbon dioxide and glucose. There are two stages in the process."
  },
  "quota_remaining": -1,
  "version": "9.17"
}

A bare "score": 2 gives a learner nowhere to go. The explanation turns it into three specific things to fix: name the two stages, explain how the first powers the second, and finish with the lecturer’s conclusion. That is feedback a teacher can hand over directly, or that your application can surface as next steps.

Writing a good task_context

Because the score is measured against your task_context, it is worth a few minutes of authoring. Two things matter most:

Write the model answer, structured in four parts. Name what the stimulus shows; list the key features with their values, most salient first; state at least one explicit relationship or trend; close with an overall takeaway. Those four parts are what the 0–5 bands reward. In particular, if your context never states a relationship or a conclusion, nothing in it rewards a candidate who provides one.

Keep it to what a candidate can actually say. In PTE Speaking the candidate has about 40 seconds, which is roughly 80–100 spoken words. Coverage is measured against your context as a whole, so every element you add that a candidate has no time to produce dilutes the share they can realistically reach. We aim for 80–110 words for describe-image (around 500–700 characters) and 90–140 words for retell-lecture (around 550–900 characters). The retell context is a transcript or close summary of a one-to-two minute talk, so it carries connective prose an image description does not. The 1024-character limit is a ceiling, not a target.

One thing worth knowing if you tune contexts: length on its own does not set the score. Rather, which elements the context contains does. Because coverage is proportional, trimming a context makes each remaining element count for more, so the same response can score lower against a shorter context if the trim removed elements it happened to cover. Edit for the right elements, not to a character count.

Full guidance for all three task types, including worked describe-image and retell-lecture contexts, the four-part structure, and what to leave out is in the Score Task API reference.

One note on using score_explanation

score_explanation is generated prose intended for people. Its wording varies between calls on the same input, so show it to teachers and learners, log it for review, use it to spot-check a score you disagree with — but do not parse it, key application logic off it, or diff it between runs. The numeric score remains the field to store and compare.

Availability

The new describe-image and retell-lecture models are live in all regions. Score Task is available to Speechace Premium subscriptions.

We would also like to hear which task types you want next. Send ideas to contact@speechace.com.

All the best, The Speechace Team