The beginning of clear speech

Phoneme practice, one sound at a time.

A phoneme is the smallest unit of speech — the /m/ in “mother”, the ख in “खाना”. Phoneme practice in Sign & Talk gives you a place to see how a sound is made, record an attempt, and read back a score for that sound alone, with nothing else in the way.

Open the appThe Android app is not on Google Play yet. Use the web app in your browser today.

Why start with single sounds?

A word is several sounds running into each other. When a word comes out unclear, it is hard to tell which part went wrong — and harder still if you cannot hear the difference yourself. A single sound removes that guesswork: there are no neighbours pulling at it, so whatever the score reflects is the sound itself.

Every sound in the app comes with a mouth-position guide: where in the mouth it is formed, whether the air is stopped or squeezed through a narrow gap, and whether the vocal folds are switched on. Those three things separate most sound pairs that learners mix up. /s/ and /z/ are made in exactly the same place in the same way — the only difference is voicing, and you can feel that with two fingers on your throat without hearing a thing.

A recording is analysed for the sound you were asked to make, and comes back as a number from 0 to 100 with the acoustic detail behind it: voice onset time, pitch, aspiration, vowel shape. If the recording was too quiet, clipped or empty, the app says so and gives you the attempt back rather than scoring noise. A false start never costs you an allowance.

Most learners start with sounds they can check in a mirror — /m/, /b/, /p/, /f/, /v/ — because those are formed at the lips and teeth, where a visual cue is honest. Three at a time is the right number. Ten is a list, not a practice.

SEE

Understand the sound first

Visual mouth-position guides show the place of articulation, whether air is stopped or squeezed, and whether voicing is on. You learn what to do with your mouth before you are asked to do it.

TRY

Practise without pressure

Record when you are ready. No countdown. Quiet or invalid recordings are caught before analysis and returned for a free retry, so a cough or a bumped microphone costs you nothing.

SCORE

A number, not a verdict

Each attempt returns a 0–100 confidence for that sound with the measurements behind it. A 61 that used to be a 40 is progress you can point at; “incorrect” is not.

CHOOSE

Start with visible sounds

Lips and teeth first — /m/, /b/, /p/, /f/, /v/ — because you can verify them in a mirror without hearing them. Sounds made further back come once the visible ones are steady.

READY

Know when to move on

A sound is ready to leave behind when it is right on the first attempt across several separate days. Same-day repetition flatters you; the day-to-day record does not.

NEXT

Carry it into words

Once a sound is stable alone, word practice puts it back beside its neighbours, where coarticulation reshapes it. That is the real test, and it is the next stage.

Start practising today

Start with a sound you can see.

Open Sign & Talk in your browser. The Android app follows the same account, so nothing is lost when you switch.

Open the appThe Android app is not on Google Play yet. Use the web app in your browser today.

Questions people ask

The things worth knowing before you start.

What is a phoneme, exactly?

A phoneme is the smallest sound that changes the meaning of a word. Swap the first phoneme in “pat” and you get “bat” — different word, one sound apart. English uses about 44 of them; Hindi uses a different set. Practising them one at a time is how most speech work starts, because every word is built from them.

Can a deaf learner practise sounds they cannot hear?

Yes, and that is what the app is built for. Every cue is visual or physical: where the tongue and lips sit, whether air is stopped or flowing, whether the throat is vibrating. The feedback is a number and a picture, not a played-back recording you are asked to judge by ear.

How is the score calculated?

A phoneme model listens to the recording and returns a 0–100 confidence that the target sound was produced, along with acoustic measurements such as voice onset time, pitch and aspiration. It is an educational aid, not a medical assessment, and it does not diagnose anything.

How long should a session be?

Ten focused minutes on three sounds beats an hour on twenty. Practice that happens most days matters far more than practice that happens hard once a week, and nothing in the app penalises a missed day.

Does it work in Hindi?

Yes. The app works in Hindi and English, including sounds such as ख and घ that English does not have. You choose the language when you start and can change it later.

Where this fits

The rest of the path.

Word practice

Put the sound back into a whole word, where its neighbours start to reshape it.

Read more

How the analysis works

What the 0–100 score measures, and what it deliberately does not claim.

Read more

How Sign & Talk works

The whole path, from your first sound to sentences, in one page.

Read more