Inside Ekavo

What Speech match measures (and what it doesn't)

October 17, 2026 · scoring

Every scored take in Ekavo comes back with a number called Speech match. It is an automated estimate of how that recording aligned with Ekavo’s American English practice target. Numbers about your voice deserve a clear explanation, including where they can be useful and where they can be wrong.

What goes into the score

Speech match can use these components:

Sound accuracy: how closely the sounds in your take aligned with Ekavo’s American English practice target.
Flow: how smoothly the take moves through the text, including pauses between words.
Coverage: how much of the text on screen you read. Skipped words lower this component.
Melody: stress, rhythm, intonation, and speaking rate in sentence and passage takes. Word takes do not include this component.

The components work together, so a high result in one area does not erase a lower result in another. A take can receive a high sound score and still show that flow or coverage needs another look.

What it does not measure

Speech match describes one scripted reading. It does not directly test whether another person understood you. It does not assess your vocabulary, grammar, ideas, or ability to speak without a script. It is not a proficiency level, a teacher’s judgment, or a test of whether you sound like a native speaker.

Ekavo currently teaches American English as a practice target. That is one useful target, not the correct way for everyone to speak. Ekavo’s goal is greater clarity and control, not erasing where you are from.

The bands in the app (Developing, Solid, and Refined) make the numbers easier to scan. They are not mastery certificates.

Why one take is never the whole story

Speech match is produced by an automated speech system. Microphone distance, background noise, recording quality, speaking volume, and the system’s interpretation can all affect a result. If a score surprises you, make another recording in a quiet place before using it to choose what to practice.

For each tracked sound that has appeared in your recordings, Ekavo can show a starting baseline, recent performance, recent variation, and a personal best. Sounds that have not appeared yet remain New. Once there is enough evidence, the app can also show a steadiness label: Steady, Settling, or Varies. That label describes consistency, not level.

Before Ekavo has enough personal evidence, your first-language estimate may guide the starting order. As your recordings build a clearer picture, recent performance takes priority. A recommendation is not based only on a personal best or one unusually low take.

How to use Speech match

One score tells you how the current take aligned with the target. It cannot tell you whether you improved until you compare it with another take or view the trend. Look at the component that limited the score, then check the particular word or sound you want to change.

If Explain this score is available, tap it for a short explanation and practical next steps.