Perslis.AAC
OPEN SOURCE • ON-DEVICE • BUILT FOR REAL PEOPLE OFFLINE • PRIVATE • FOR EVERYONE

The conversation
engine your AAC app
has been missing.
The conversation
engine your AAC
app has been missing.

Turn any input into natural, thoughtful, context-aware AAC conversations. Offline-first, private, and modular — for families, therapists, schools, and developers.

Turn any device into a smarter, more thoughtful AAC experience — offline-first, private, and built for real life. For families, therapists, schools, and caregivers.

Works offlinePrivacy firstBring your own model
HomeI want
Type a message or tap a tile
Simple.
Powerful.
Human.⤶
Android iOS Web Clinical & Emergency
MORE THAN A BOARDBUILT FOR REAL-WORLD AAC

Everything you need to build thoughtful, natural AAC conversations.Everything you need
to build thoughtful,
natural AAC
conversations.

From core communication to real-world independence — Perslis.AAC gives you the tools, privacy, and flexibility to communicate anywhere.

From single-device apps to full clinical deployments — Perslis.AAC gives you the building blocks to create flexible, intelligent, and private communication tools.

Multi-Modal Input

Text, speech (Vosk), images (OCR), or touch.

Smart Responses

Context-aware, natural, child-friendly output.

Fully Offline

Works without internet. Your data stays yours.

Multi-Language

English, Spanish, Hindi, Chinese, Russian + more.

Easy to Integrate

Use our API or drop-in components.

Open & Extendable

Open source. Bring your own model.

Natural Language

Context-aware conversations with offline AI.

Visual-First

Icons, photos, OCR, and custom boards.

Private & Offline

Your data stays on your device.

Built for Everyone

Children, adults, stroke, aphasia, and more.

Explore All Features →
SEE IT IN ACTION

A conversation
in seconds.

Tap to communicate, or let voice and images help. TinkySpeak adapts to you — at home, at school, in therapy, or on the go.

LIVE DEMO

See it in action.

Try the conversation engine right here. Type a message, pick a few options, and see how Perslis.AAC builds a response.

Try an example:
❝

“This gives my son a voice. It’s simple, fast, and it works anywhere.”

— Parent

That sounds great! 🌳
Would you like to go now or later?

Live model · online demo
DEVELOPER FRIENDLY

Integrate in minutes.

Use our API or SDK with any app. Bring your own model, run it locally, and keep full control.

Use our APIs or SDKs to bring AAC intelligence into your app, device, or workflow — model agnostic.

import json, urllib.request

url = "http://localhost:7437/v1/sessions"
def post(url, data):
    req = urllib.request.Request(url,
        json.dumps(data).encode(),
        {"Content-Type": "application/json"})
    return json.load(urllib.request.urlopen(req))

session = post(url, {"language": "en"})
reply = post(url + "/" + session["id"] + "/hear",
    {"text": "Did you like the game?"})
print(reply["choices"])
View Full DocsView Developer Docs →
TinkySpeak tablet with six communication tiles, an I want apple speech bubble, and on-device, offline and multilingual badges.
REAL IMPACT

Built for people. Trusted by builders.

10K+Conversation boards
supported

30+Downloads/month
(Hugging Face)

5+Languages today
(+ more coming)

100%Open source free tier
(BYO model key)

A child using an AAC board
FOR ALL AGES

Communication
without limits.

From first words to complex conversations, TinkySpeak grows with you.

Watch the Story ▶
A child using a colorful AAC communication board on a tablet.
FOR FAMILIES, THERAPISTS, AND SCHOOLS

Communication without limits.

Give every learner a voice. Designed for neurodivergent children, stroke survivors, and anyone who needs a more natural way to communicate.

Families

Everyday use at home.

Therapists

Custom boards & goals.

Schools

Classroom ready.

Clinics

Private & secure.

See Use Cases →
REAL EXAMPLE

19 routes. One session. A complete conversation.

From single prompts to multi-turn dialogue, the engine never fetches or draws artwork — it builds meaningful responses from your input and board context.

View Full Example →
1

User Input

“I want to go
to the park”

2

Intent Detection

Identifies
“go to park”

3

Context Expansion

Time, location,
companions

4

Response Options

Now / Later /
With mom / With dad

5

Natural Reply

“That sounds great!
Would you like to go...”

JOIN THE MOVEMENT

Ready to build a more inclusive tomorrow?

Open source. Privacy first. Built for real people.Join families, therapists, and developers using Perslis.AAC to make communication more human.

More voices.
A brighter world.⤴