For speech pathologists

Run it in your session. Send it home for the week.

A profile per child on your caseload, queues that mix sounds and positions, a drill mode when you want trials rather than turns, and a session summary you can paste straight into your notes.

Every lever, on one screen

Sounds, multi-select

All 23 sounds and 24 blends, ordered developmentally in the picker, blends grouped L / R / S. Select several and the queue interleaves them.

Positions, multi-select

Beginning, middle and end, offered only where the sound actually occurs — /ng/ has no initial set; /h/, /w/ and /y/ have no final set.

Three levels

Word, phrase and sentence. No reading-heavy fourth tier, and the jump between them is deliberately small.

Loaded or carrier phrases

At phrase level, switch between silly sound-loaded phrases and plain carriers — “my sail”, “a big soup” — where only the target word carries the sound.

Opponent, named

Games with a second player run against an automatic Buddy or against a real person you name, so the prompt reads “Pass the dice to Miss Kate!”

Listening, not just speaking

Switch the activity to Listening for minimal-pair discrimination: 14 contrast sets, 94 pairs, indexed by phonological process. You say the word, the child picks between two pictures — word level only, and still no audio from the app.

Drill mode

Rapid Drill strips the game layer out entirely — fast cards, highest productions per minute, for when you want trials rather than turns.

The content checks you would otherwise do yourself

Position purity, phonetic not orthographic. No word in a phrase or sentence carries the target sound in a position other than the one being practised. Silent letters, soft c, x as /ks/, ph as /f/, ng versus g and silent b are all handled.

Nothing is ever spoken by the app. No text-to-speech, no recorded model, no microphone. The adult in the room is the model, which is the only model worth copying.

Every target word is underlined. At phrase and sentence level the child sees exactly which words carry the sound, rather than inferring it.

Impure items are excluded, not fudged. Words carrying the target in two positions at once are dropped from both sets rather than assigned to one — “cake” never appears as a /k/ item.

A worked example

Practising /s/ in the initial position, at sentence level, the app can offer:

The sun sat on a sock.

Three underlined targets, all initial. No six (that x is /ks/), no seals (final /s/), no nice (soft c). The checks are phonetic, not spelling-based, and run over all 992 items.

What you get back at the end

Accuracy, defined

Correct divided by correct plus retries. Skips are excluded deliberately, so an abandoned item does not read as an error.

Needs-practice list

Any word retried twice or more, or skipped, is named in the summary — the shortlist for the next session.

History per profile

The last fifty sessions per child: date, target sounds, positions, level, mode, totals and accuracy. Clearable per profile.

Plain-text export

One tap copies the whole summary as text — target, counts, accuracy, needs-practice and the carryover idea — to paste into your notes.

Copied to your clipboard
Say & Play session — 12/08/2026 Target: /s/ (initial) — word level Productions: 28 | Correct: 22 | Retries: 5 | Skips: 1 Accuracy: 81% Needs practice: sofa, seven   Carryover ideas: - Go on a snake sound safari — find 5 things with /s/.

What it does not do

Worth knowing before you put it in front of a family, so nothing surprises you later.

It does not sync. Everything is stored on the device it was used on. A profile built on your clinic iPad does not appear on the family’s phone, and there is no login to restore it from.

It does not assess. Scoring is adult judgement, not an objective measure. Treat the accuracy figure as a practice log, not a probe result.

It does not listen. No speech recognition, by design. If you need instrumental feedback, this is not that tool.

History is capped. The last fifty sessions per profile are kept; older entries roll off. Export anything you need to keep.

Export is clipboard text. There is no PDF report or caseload dashboard — the summary copies as plain text for you to paste wherever your notes live.

Handing it to a family for the week

1

Set it in the room

Pick the sound, position and level with the family watching. The app remembers it, so they never have to make that choice themselves.

2

Run one session together

Two minutes of you pressing the buttons shows a parent what counts as Correct far better than any instruction sheet.

3

Give them a number

Thirty says a day is the built-in goal. Tell them if you want something different — they can stop whenever you say.

4

Ask for the summary

They can copy the end-of-session text and send it to you, or just bring the needs-practice words to the next appointment.

One subscription. Everything in it.

Every sound, every blend, all twelve ways to play, unlimited child profiles. No coin packs, no locked levels, no ads.

$20 a month, or $200 a year · cancel any time

Coming soon!

Subscriptions are not open yet. Gabby is still getting the tills working — email hello@sayandplay.app and we will tell you the day it opens.