Real Speech Analysis by HackerNoon
Read aloud for 30 seconds. See where your vowels come from.
🔒 Your voice stays yours
- · Your recording is analyzed once, then deleted the moment we’re done.
- · We never save your audio, what you said, or your results.
- · No account, no sign-in. Nothing is tied to you.
What We Analyze
Each word in the passage was picked to reveal one part of your accent.
Regional Origins
Your measured vowels are compared with the dialect regions mapped in the Atlas of North American English.
Vowel Mergers
Do “cot” and “caught” sound the same? “Pin” and “pen”? These mergers split the country into clear regions.
Vowel Shifts
Raised short-a (“cat”), a fronted “oo” (“food”), a smoothed “I” (“five”) and Canadian raising (“about”).
The R Sound
Whether you pronounce the r in “car” and “farm”, measured from the dip in your third formant.
Accent Family
An AI model trained on thousands of speakers recognizes 16 English accents, from American to Indian to Scottish.
Language Background
A second model hears the influence of a first language: Romance, Germanic, Slavic, Arabic or Hebrew, East, Southeast or South Asian.
How It Works
Read Aloud
Read a short, silly passage at your normal pace. It takes about 30 seconds.
Measure
We find each target vowel and measure its formants, the resonances that make an “ah” different from an “aw”.
Compare
Your vowels are matched against dialect regions, and two AI models place your accent on a world map.
The AI Models
Two open models from Hugging Face listen alongside our vowel measurements.
CommonAccent
🤗 Jzuluaga/accent-id-commonaccent_ecapa
Recognizes 16 English accents, from American and Canadian to Indian, Scottish and Filipino.
SpeechBrain ECAPA-TDNN trained on Mozilla Common Voice English.
View on Hugging Face ↗
Voxlect English Dialect
🤗 tiantiaf/voxlect-english-dialect-whisper-small
Hears your English variety or the influence of a first language, across 16 groups.
Built on OpenAI’s Whisper-small speech encoder. We run a compressed (8-bit) copy.
View on Hugging Face ↗
Also powered by open-source speech tools: sherpa-onnx (speech recognition and speaker grouping), Silero VAD (finding speech), Praat (vowel measurement).