Ship Spelly in
your product.

The same engine that powers Spelly Learn, Spelly Practice and our white-label deployments. Pick how you want to use it.

Already have an account? Sign in to the operator console

your-product.com/widget
The Spelly widget playing a sentence inside a product: each word lights up as it is spoken and the mouth shows the sound being made.

How it connects

Inputs

  • Audio
  • Reference text
  • Language

Engine

Spelly model

Same model Spelly's own apps run on.

Outputs

  • Match score
  • Phoneme detail
  • Tips per sound

One request. One response. Same shape whether you are embedding the widget or calling the API directly.

One request. One response.

Multipart upload, X-API-Key auth, JSON back. Same shape whether you embed the widget, call the API directly, or run Spelly under your own domain.

Request
$ curl -X POST https://spelly.ch/api/assess \       -H "X-API-Key: $SPELLY_KEY" \       -F audio=@recording.webm \       -F reference_text="The quick brown fox." \       -F language="en-GB"
Example response200 OK
{  "accuracy": 94,  "completeness": 98,  "fluency": 87,  "clarity": 92,  "duration": 2.4,  "analyzed_text": [    { "word": "quick", "accuracy": 96, … }  ]}

Four embeds

curl
$ curl -X POST /api/assess
-H "X-API-Key: $KEY"
-F audio=@rec.webm
200 OK

For engineering teams

API

One request per recording. Per-phoneme feedback. The same engine that runs Spelly Learn, Spelly Practice and our white-label deployments.

your-domain.com

Practise this sentence

The quick brown fox jumps.

0:02 / 0:03

For product teams

Widget

Drop pronunciation practice into your page.

Under your brand
  • Your name and logo
  • Your colours
  • Your domain

For enterprises & resellers

Platform

White-label Spelly under your brand. Your domain, your data, your capacity. Already running for organisations outside Spelly.

Classes today
Year 7, Room B24
Year 8, Room A19
Year 9, Room C22

For schools & districts

Schools

Run Spelly across your school. Class rosters, teacher dashboards, homework. Every student gets a daily boost on top of their tier.

Why teams ship with us

Built at ZHAW

Research-backed speech AI

We started Spelly on our own while doing research at ZHAW in Switzerland, building on what we learned there and on our background as data scientists. Two co-founders, two academic advisors, one independent assessment engine.

23 stages · 500+ lessons per language

A curriculum with depth

Every lesson is carefully crafted for the path, so learners keep improving their pronunciation from first words to advanced, across twenty-three stages of everyday topics. Every sentence is scored sound by sound, and each day’s drill follows the sound the learner gets wrong most.

Speechprint per user

Personalisation that follows

Per learner, per language. Updates across every diagnostic and Learn session. The same Speechprint follows them wherever they practise on Spelly.

Pick a path. We will take it from there.

Tell us which surface fits what you are building. We will get back to you with the technical brief and a working demo.