Accent Test — Explore Your Speech Patterns Online

What makes an accent?

Vowel quality and consonant articulation are central to pronunciation. Different speakers may shape the same written word with different vowel positions, consonant releases, or sound reductions. A fundamental-frequency tracker does not identify those details; pitch is not a substitute for a phonetic transcription.

Intonation, pace, stress, and rhythm also affect how speech sounds. The contour here provides a limited picture of intonation, while the fixed passage permits a timing estimate. Vocabulary and grammar contribute to broader language varieties, but this tool neither transcribes nor analyzes your word choices.

What this tool actually measures

Speaking rate is estimated from the number of words in the displayed passage divided by the elapsed recording time in minutes. Confirm that you read the entire passage once before using that estimate. It cannot count spoken words, detect omissions, or separate your starting delay from reading time.

Pitch range shows the minimum and maximum accepted fundamental-frequency estimates; median pitch summarizes their centre. Pitch movement uses the central P10–P90 span expressed in semitones, which reduces the influence of isolated extremes. These are descriptive summaries, not accent benchmarks.

The rhythm panel summarizes the durations of runs with detected voicing and gaps without it. Unvoiced consonants, breaths, silence, and tracking failures can all create gaps. This is a voicing-pattern proxy, not syllable timing, stress analysis, or a reliable count of linguistic pauses.

Why accent detection is hard for a browser tool

A useful classifier would need appropriate speech data, a trained model, clearly defined categories, and evaluation across microphones and speakers. The local YIN algorithm instead estimates repeating periods in audio. It has no knowledge of words, phonemes, or the language variety a speaker intended.

Accents also vary continuously rather than falling into neat boxes. A person may change their speech with context, audience, language, or code-switching. A short scripted reading samples only one situation and cannot establish how someone speaks in every setting.

Microphones, room reflections, background noise, and automatic device processing alter the signal. A low pitch estimate or a missing segment may reflect recording conditions rather than pronunciation. Keep the setup consistent when comparing readings.

How to use these measurements

Compare your own readings of the same text at a relaxed pace and a deliberately slower pace. Then compare a formal delivery with a conversational one. Inspect which parts of the pitch curve move, how long the recording lasts, and whether gaps appear at similar places.

If you want to follow changes over time, keep a private written record of the displayed metrics and the recording setup. The page does not store a history. For listening feedback, use a separate local recorder and consider what you hear alongside the measurements.

Use the observations to explore your own delivery, not to rank other people or decide which accent is better. For pronunciation goals, a qualified speech or language coach can provide sound-specific feedback that this tool cannot supply.