Most people can't hear their own accent, because to them it just sounds like talking. This accent test listens to a recording of your speech and names the accent it most likely belongs to, whether that's a national variety like Australian English, a regional one like the US South, or a second-language accent shaped by your first language. It doesn't stop at a label. The report walks through the vowel shifts, consonant habits, rhythm and intonation that led to its guess, rates how strong the accent is compared with a standard reference pronunciation, and offers a geographic estimate with a probability spread when more than one origin fits. Use it to satisfy curiosity, check progress in accent training, or understand why people keep asking where you're from.
Example result
Input
A 35-second voice memo of a woman describing her morning commute in English.
Output
Accent identification: North American English, most consistent with the Upper Midwest (Minnesota or Wisconsin), with a lower-probability match to Canadian Prairie speech. Key markers: the vowel in "bag" is raised toward "beg", and the long o in "go" and "road" is monophthongal, sounding close to a pure vowel rather than a glide. "Cot" and "caught" are merged. Rhoticity is full, and medial t is flapped in "better". Rhythm is stress-timed with a mild sing-song lift on sentence-final words. Accent strength: moderate. A listener from the coasts would likely notice it, while someone from the region would not. Comparison to General American: the main differences are the raised short a and the steadier o vowel; consonants are nearly standard.
Scores
How it works
How Accent Analyzer works
- 01
Vowels first
Vowels carry most regional identity. The model listens for things like whether "cot" and "caught" merge, how you say the vowel in "bath" or "goat", and whether you drop or keep the r after vowels.
- 02
Consonant habits
It notes t-flapping ("water" as "wadder"), glottal stops, dental th sounds, clear versus dark l, and any substitutions typical of a particular first language.
- 03
Rhythm and melody
Stress timing, syllable length and the rise or fall at the end of statements separate accents that share similar sounds, such as Irish and Scottish English or Brazilian and European Portuguese speakers of English.
- 04
Origin estimate and strength
The features are combined into a best-guess region with alternates, a 0 to 100 score for how marked the accent is, and notes on what sets it apart from a standard reference accent.
Use cases
Who it is for
Satisfying the "where are you from?" question
You moved a few times growing up and nobody can place you. The report tells you which region your speech leans toward and which features are mixed in from elsewhere.
Tracking accent reduction or accent training
Record the same passage every few weeks. A falling accent-strength score and fewer flagged features show whether practice on specific sounds is carrying over into natural speech.
Actors building a dialect
Check whether your stage Texan or your RP reads as the target, or whether your home accent is leaking through on particular vowels.
Language learners curious about their L1 influence
See which habits from your first language shape your English, such as syllable-timed rhythm or v and w swaps, so you know which ones matter most for being understood.
Tips
Get better results
- Talk naturally about something familiar for 30 to 60 seconds. Read-aloud text makes most people shift toward a more neutral accent.
- Include words that tend to reveal accents: "water", "car", "bath", "about", "thought" and "right".
- Record in a quiet room with the phone about a hand's width from your mouth so vowel quality comes through clearly.
- If you speak more than one language, test each in a separate recording; mixing them confuses the origin estimate.
- Treat the region as a starting point. Look at the listed features to see why the model chose it.
Limitations
What it can't tell you
An accent guess is a probability, not a fact about where you were born or raised. People who have moved, grown up bilingual or trained their speech often produce blended features that fit more than one place. Short clips, background noise and reading from a script all reduce accuracy. The analysis comes from an AI model interpreting audio, not from phonetic lab measurement, and it should never be used to make judgements about someone's nationality, immigration status or ability.
Frequently Asked Questions
Accent Analyzer questions
What accent do I have?
Record 30 seconds or more of relaxed speech and the tool names the most likely accent, lists close alternatives, and explains the specific vowels, consonants and intonation behind the guess. If your speech mixes regions, the probability spread shows that instead of forcing one answer.
Can an accent detector guess where I'm from?
It can estimate where your accent is typical, which often matches where you grew up but not always. Moving, a parent's accent, media exposure and deliberate practice all shift pronunciation, so treat the location as where your speech sounds at home rather than a statement about your history.
Does it work for non-native English speakers?
Yes. For second-language speakers it points to the likely first-language influence, such as Spanish, Hindi, Mandarin or German patterns, and notes the sounds that carry it. It also analyzes speech in other languages, though regional detail is richest for widely spoken varieties.
What does the accent strength score mean?
It's a 0 to 100 rating of how far your pronunciation sits from a standard reference for that language, where low means close to the reference and high means clearly marked. A high number isn't a problem. It only tells you how noticeable your accent is to outsiders.
Is this the same as an accent guesser game?
The idea is similar, but instead of a single guess you get the linguistic reasoning: which features were heard, how they compare to a standard accent, and how confident the model is. That makes it useful for training, not just a one-off guess.
How long should my recording be?
Thirty seconds to two minutes of continuous speech works well. Very short clips give the model too few vowels to compare. One voice, one language, natural pace.
Related tools
You might also need
Language Detector
Detect spoken languages and provide language identification with confidence scores
English Pronunciation Checker
Assess and improve English pronunciation accuracy
Speech Clarity Assessor
Evaluate articulation, enunciation, and intelligibility
Celebrity Voice Matcher
Compare voice to celebrities and identify which celebrity you sound most like
Voice Age Estimator
Estimate age from voice characteristics including pitch, resonance, and vocal maturity
Accent Analyzer uses credits from a paid LindaleAI plan. See pricing