Why We Built AccenTuner: The Coach We Wished Existed

The origin of AccenTuner is simple, even if the problem it addresses is not. We kept meeting people — in Canada, in professional contexts, in conversations about language and opportunity — who were held back by something they knew how to name but did not know how to fix.

Their English was strong. Their vocabulary was expansive. Their grammar was precise. But the way they sounded in professional settings created friction that their knowledge and capability could not overcome. And when they looked for help, what they found was either inaccessible or insufficient.

What Was Available

Human accent coaches are excellent. The good ones listen carefully, identify patterns quickly, and create a learning environment that is encouraging without being dishonest. They are also expensive — typically $80 to $150 per hour — and available only at scheduled times that rarely align with the moments when practice is most useful.

Pronunciation apps existed. Most of them worked by recording a sentence, running it through an analysis algorithm, and returning a score. The feedback was better than nothing. The learning loop was too slow. By the time you saw your score, the production was gone. The correction was abstract. The improvement was slow.

Neither solution matched the actual mechanism of phonetic change: immediate correction, immediate retry, in the context of real-world speaking situations.

What Changed

In late 2024, voice AI reached a threshold. Real-time voice interaction — sub-200ms latency, natural turn-taking, context-aware responses — became practically achievable with a new generation of real-time AI infrastructure. The thing that had been technically possible in theory became actually deployable.

We built a prototype. A real-time AI accent coach that listened during a simulated job interview, identified the exact phrase where clarity broke down, explained the issue specifically, modelled the target sound, and asked for a retry before moving on. The feedback loop that human coaches produce — but at a fraction of the cost and available any time.

The results were faster than we expected. Not because the AI was doing something magical, but because the mechanism was right: corrected repetition, immediately, in a real-speaking context.

Who We Built It For

AccenTuner is built for people who are already strong in English but want to close the gap between how well they speak and how clearly they are heard. Specifically:

• New Canadians navigating the job market or regulated profession accreditation
• Professionals preparing for interviews, client presentations, or leadership roles where presence matters
• Anyone who has been told their English is good but is asked to repeat themselves more than they should be

We operate from Canada because Canada is where the need is particularly concentrated and well-documented. More than 20% of Canadians speak a language other than English or French at home. Immigration is not an edge case here — it is a central part of the national story. Accessible accent coaching is not a luxury in this context. It is an equity issue.

What We Are Not Trying to Do

We are not trying to erase accents or make everyone sound the same. Accent is part of identity. The goal is not homogeneity — it is clarity. Specifically, the kind of clarity that removes the friction between what someone means and what they are heard as saying. That is a different goal from "sound like a native speaker", and it is a more achievable and more useful one.

We are also not trying to replace human connection in language learning. The relationship between a good teacher and a motivated student is irreplaceable for motivation, nuance, and the social dimensions of language. What we are trying to replace is the inaccessibility of that relationship — the cost barrier, the scheduling constraint, the lag between need and access.

Where We Are Going

AccenTuner is early. The core coaching loop works well. The scenario library is growing. The pronunciation assessment layer adds a measurement dimension that most coaching tools lack. And the feedback we get from people who have used it for three or four weeks — who notice the change in how they feel opening their mouth in a meeting — is the kind of signal that tells us the mechanism is right.

We are building toward a tool that is genuinely useful across the full arc of accent development: from the first session through to the point where speaking clearly in your target accent is simply how you speak. That is the coach we wished existed. That is what we are building.