第1話

Chapter 1: OwlReader

“The Owl Who Listens”

A reading coach for early learners that hears every sound a child makes.

WAITLISTLanding page and waitlist live, families signing up
Visit owlreader.ca Play Treasure Talk

BRYAN

Inconsistent reading practice

The problem

We speak three languages at home. Who checks that the reading practice actually sticks?

A parent

Children in multilingual households get uneven reading practice. A parent can't always hear which sounds a child is missing, especially in a second language.

How it works

  1. Step 1: Listen

    The child reads aloud. Azure speech assessment scores the reading sound by sound (phoneme-level), in real time.

  2. Step 2: Coach

    Claude turns those results into short, personal coaching: what went well, and the one sound to try again.

  3. Step 3: Play

    An animated student UI, built from the API routes to the owl itself, so practice feels like a game.

Behind the ink

How it's built, and how the AI is kept honest.

One pipeline, two AI services

Speech assessment and LLM coaching joined end to end, shipped full stack by one developer.

Treasure Talk

A voice-controlled storybook game in English and French. Speech recognition runs in the browser first, with a server fallback. It makes no pronunciation scores it can't back up.

Accelerator

Completed Invest Ottawa's Venture Path Ignition program (Cohort 32) and pitched OwlReader at the Pitch Finale.

Built withNext.js · TypeScript · Azure Cognitive Services (Speech) · Anthropic Claude API