How Accent Detection Works

When you speak into My Accént, the system identifies your specific English accent — American, British, Australian, Irish, Scottish, Indian, South African, or Nigerian. But how?

The Basics of Accent Identification

Every English accent has a distinctive "fingerprint" — a set of vowel positions, consonant productions, and rhythmic patterns that distinguish it from other accents. Accent detection analyses your speech to match these patterns.

What the System Listens For

Rhoticity

The most immediate accent marker: do you pronounce R after vowels?

• Rhotic accents (American, Irish, Scottish, Indian): "car" has an audible R
• Non-rhotic accents (British RP, Australian, South African): "car" → "cah"

This single feature divides the English-speaking world in half.

Vowel Positions

Each accent has characteristic vowel sounds:

• The "bath" vowel: broad "ah" (British RP, Australian) vs flat "æ" (American)
• The "lot" vowel: rounded (British) vs unrounded (American)
• The "goat" vowel: diphthong starting position varies dramatically by accent

Rhythm and Timing

Stress-timed vs syllable-timed tendencies vary by accent:

• American and British English are strongly stress-timed
• Nigerian and Indian English tend toward syllable timing

Consonant Features

• T-flapping (American)
• T-glottalisation (British urban)
• TH-stopping (Irish, some others)
• Retroflex consonants (Indian)

From Detection to Personalisation

Once your accent is identified, the system loads your accent's phoneme inventory — the complete set of sounds you produce. This inventory is then compared to the target language's phoneme inventory:

1. Matching sounds → marked as transfer (you already make them)
2. Similar sounds → marked as adjust (small modification needed)
3. Missing sounds → marked as new (requires new muscle memory)

This comparison is what makes learning personalised. An American speaker learning French has a different transfer/adjust/new ratio than a British speaker learning French, because their starting sound inventories are different.

Privacy and Processing

Your speech is processed to extract acoustic features — formant frequencies, voice onset times, rhythm metrics. The raw audio is not stored after analysis. The system needs your accent category and phoneme inventory, not recordings of your voice.

Accuracy

No accent detection system is perfect. Accents exist on a continuum, and many people have mixed or modified accents. The system identifies your closest match from its accent categories and uses that as your starting profile, which you can adjust if needed.

---

Explore more:

• French pronunciation guide
• Spanish pronunciation guide
• Take the free accent quiz
• French pronunciation for your accent