PTE speaking · pronunciation + oral fluency
Free PTE pronunciation checker: an AI speaking analyser
Record any PTE speaking task, get an instant AI read on your pronunciation and oral fluency traits, and see the exact words that lost you marks. The pronunciation checker mirrors the two AI-only speaking traits Pearson scores on every spoken item, so what it flags is what a real test marker would flag.
Last reviewed: 17 July 2026. Facts verified against the official PTE Academic Test Taker Score Guide (July 2025) and pearsonpte.com pages.
Why this matters
Pronunciation is where AI-scored speaking is won or lost
PTE Academic is one of the few English tests where pronunciation is scored purely by machine. There is no human on the other end filtering out prejudice about your accent or giving you the benefit of the doubt on a mumbled ending. Pearson's Score Guide (July 2025) is explicit: pronunciation and oral fluency are AI-scored on every speaking task, always, and even the seven task types that get human content review never have a human touch those two traits.
For most retakers who reach a 65 plateau or a 72 plateau on speaking, the issue is not vocabulary or ideas. It is the two AI-scored traits: hesitations, restarts, missed word stress, and swallowed final consonants. A checker that isolates those signals gives you a tighter feedback loop than a human tutor listening in real time and being polite.
The catch: you cannot fix pronunciation by hearing yourself once. You need repeated recordings, a machine read on each one, and a targeted drill on the specific sound or the specific pause that keeps failing. That is the loop this page runs.
The three speaking traits
What the AI actually listens for
Every PTE speaking answer is scored on up to three traits. Two of them are AI-only, and those two are where a checker adds the most value.
| Trait | Scoring | What it measures | Why it matters |
|---|---|---|---|
| Content | AI, then human review on open tasks | Whether what you said matches what was asked or shown | Zero content collapses the whole item, no matter how fluent you sound |
| Oral fluency | AI only, always | Even rhythm, natural phrasing, minimal hesitation, no restarts | The single biggest cap between 65 and 79. Choppy delivery holds you at fluency 3 |
| Pronunciation | AI only, always | Clarity of individual sounds, word stress, and intelligibility to an international listener | Any accent is accepted. What loses marks is mumbled endings, missed stress, and swallowed vowels |
Source: PTE Academic Test Taker Score Guide (July 2025). If Content scores zero on an open task (a memorised template, wrong topic, off-prompt), Pearson stops scoring the item and the whole answer scores zero, no matter how clean your pronunciation is.
The numbers
Speaking sub-scores you need for each band
Speaking is measured per skill, so your final speaking score has to clear the band cut-off on its own, no averaging with the other three skills.
| Goal | Speaking cut | Fluency trait | Pronunciation trait | Who this fits |
|---|---|---|---|---|
| Competent (0 PR points) | 54 | 3 | 3 | Baseline for most work visas and undergraduate admission |
| Proficient (10 PR points) | 76 | 4 | 4 | Comfortable middle target for skilled migration and most master's programmes |
| Superior (20 PR points) | 88 | 5 | 5 | Competitive PR pool, medical registration bridging, top scholarships |
The trait bands run from 0 to 5. A pronunciation trait of 3 is the common ceiling for retakers stuck around 65 speaking, and it is almost always the two AI-scored traits (fluency and pronunciation) doing the capping, not content. Superior speaking (88) needs both traits at 5, which means at most one small stumble across the whole Read Aloud set and clean final consonants throughout.
For the full per-skill cut scores across all four skills (listening, reading, writing, speaking), see the PTE score chart or, if you are on the PR pathway, the PTE for Australia PR guide.
What to record
Speaking tasks the checker practises
Every task below feeds both the pronunciation and oral fluency traits. The counts are the Score Guide ranges for a real test.
| Task | Items per test | Audio stimulus | Timing |
|---|---|---|---|
| Read Aloud | 6 to 7 | None (you read a text) | 30 to 40s prep, ~40s to speak |
| Repeat Sentence | 10 to 12 | 3 to 9s of speech, no tone | ~15s to answer |
| Describe Image | 5 to 6 | None | 25s prep, 40s to speak |
| Retell Lecture | 2 to 3 | Lecture up to 90s | 10s prep, 40s to speak |
| Summarize Group Discussion | 2 to 3 | Discussion up to 3 min | 10s prep, 2 min to speak |
| Respond to a Situation | 2 to 3 | Situation cue up to 60 words | 10s prep, 40s to speak |
Answer Short Question is technically a speaking task but since August 2025 it is scored under listening only, so the pronunciation trait does not apply. Drill it separately on the Answer Short Question practice hub.
On the day
Five pronunciation and fluency rules for the real test
- Do the microphone check properly. Speak at the same volume, distance and pace you plan to use in the test. If the pre-test mic check shows your voice weakly, the scoring engine hears you weakly for the whole 2 hours.
- Start speaking within 3 seconds. If you stay silent for more than about 3 seconds, the microphone auto-stops and anything after is lost. Even a filler phrase like 'the image shows' is better than dead air, especially on Describe Image.
- Finish every sentence you start. A stumble that ends the sentence early hits both fluency and content. A stumble followed by a completed sentence only nicks fluency slightly. Push through.
- Land the endings. The single most common pronunciation loss is dropped final consonants ('developmen' for 'development'). Consciously voice the final t, d, s, and th sounds on every word for a week and it becomes habit.
- Ignore the beep, or the lack of one. Read Aloud, Describe Image, Retell Lecture, Summarize Group Discussion and Respond to a Situation play a short tone before recording starts. Repeat Sentence and Answer Short Question do not. Do not wait for a tone that is not coming, and do not startle at the tone that is.
For the wider on-the-day playbook, see our PTE exam day tips. For the underlying speaking strategy across all seven tasks, the speaking practice hub walks through each task in order.
FAQ
PTE pronunciation checker, answered
A pronunciation checker is a tool that records your spoken answer, runs speech-to-text plus acoustic analysis, and returns a per-word confidence score. Ours mirrors the two AI-only speaking traits Pearson uses: oral fluency and pronunciation. You get a score band, the exact words that sounded unclear, and the pause markers where your rhythm broke.
Every spoken answer in PTE Academic is scored on up to three traits: content, oral fluency and pronunciation. Pronunciation and oral fluency are AI-scored on every speaking task, always, with no human review. Content is AI-scored and then reviewed by a human examiner on the more open tasks. That is why matching Pearson's automated marker matters more than sounding good to a human coach.
No. Pearson explicitly accepts any regional accent that is intelligible to an international English listener. What lowers the pronunciation trait is missed word stress, dropped endings ('understandin' instead of 'understanding'), swallowed vowels, and sounds that could plausibly be mistaken for another word. An Indian, Nigerian, Filipino, or Australian accent scores the same as a British one when the sounds land clearly.
Pronunciation is about the sounds themselves: whether an 'e' sounds like an 'e', whether the stress lands on the right syllable, whether the ending consonant is audible. Oral fluency is about the timing between the sounds: whether your pace is even, whether you pause naturally or awkwardly, whether you restart. You can have great pronunciation and poor fluency (a slow, choppy reader) or great fluency and poor pronunciation (a smooth speaker who mumbles endings). Both traits are scored on every speaking task.
The scoring engine used by Pearson is a large speech model trained on thousands of test-taker recordings, and independent Pearson research shows very high agreement with expert human raters on both fluency and pronunciation. Our checker uses a comparable automatic speech recognition model, which means the words it fails to recognise are almost always the words a real PTE marker would flag as unclear. It is a signal, not the exact final score, but it is a useful signal.
For Superior English the whole speaking skill must reach 88, which in practice means both oral fluency and pronunciation need to be at trait band 5. A trait band of 3 caps your overall speaking around 65, and a trait band of 4 sits in the Proficient range. There is no way to hit Superior speaking with a pronunciation trait of 3, even if your content is perfect.
Three drills work faster than anything else: shadow one minute of clear English audio every day (match the speaker's stress and rhythm word for word), record every Read Aloud attempt and listen back for dropped endings, and practise Repeat Sentence with the tone off so you build the reflex to speak immediately after the audio stops. Two focused weeks lifts most candidates one trait band.
Every speaking task feeds the pronunciation trait: Read Aloud, Repeat Sentence, Describe Image, Retell Lecture, Summarize Group Discussion and Respond to a Situation. Answer Short Question also involves speaking but since August 2025 it is scored under listening, not speaking. If you want the fastest pronunciation practice loop, Read Aloud and Repeat Sentence give the most reps for the least prep time.
Recommended next step
Score all four skills in a real mock, not just pronunciation
The pronunciation checker isolates one trait. A full scored mock puts every speaking task next to your listening, reading and writing scores, so you can see whether pronunciation is really the ceiling or whether another skill is dragging your overall down first. Free for a limited time.
Drill the tasks that feed pronunciation and fluency
Read Aloud practice
Highest-reps loop for pronunciation and rhythm
Repeat Sentence
Fluency under memory pressure, no tone
Describe Image
40 seconds of continuous speech, structured
Retell Lecture
Content + fluency after a 90-second listen
Speaking hub
All seven speaking tasks in one place
PTE 79 strategy
The full push to Proficient and Superior