FAQ

Frequently asked questions

Straight answers about the scoring, the course, the money and your data. If something is missing, open an issue in the Glosso repository.

What is Glosso, in one sentence?

Glosso is a free, open-source app that teaches pronunciation by listening to you speak and giving a verdict on every sound, not just on whether you picked the right words.

How does it actually measure my pronunciation?

A speech model trained on many accents turns your recording into the sequence of phonemes it heard. That sequence is aligned against the reference pronunciation of the exact sentence, sound by sound, and the differences are turned into a score and a list of named problems, for example "you said /t/ where the word needs /θ/".

Is my voice uploaded to a server?

By default, yes: on a normal install the recording is sent over HTTPS to our own server, scored in memory, and discarded as soon as the request finishes. It is never stored, sold or used to train models. If you prefer your audio never to leave the phone, you can download the optional on-device model (about 342 MB) from Settings; scoring then runs entirely on your device and also works offline. See the privacy policy.

How accurate is the score? Can it tell me how good my pronunciation is?

It gives you an honest estimate, not a verdict. The score measures how closely your sounds matched the reference for that sentence, and it is good at flagging real pronunciation problems. But it is an estimate, not a certificate, an exam result or a "sounds like a native" test. A top score is not automatic: even our own reference recordings average around 89 out of 100.

Is Glosso finished?

No, and it will keep growing. The pronunciation engine is the mature part. Lessons, languages and the iOS app are still being built, so expect some screens and numbers to change.

Is a Glosso score a CEFR certificate or an exam result?

No. The course structure is aligned with the CEFR framework, so levels run from A1 to B2 and each level is organised by what you should be able to do. A score inside the app is feedback for your learning, not an official assessment and not a qualification.

What happens if the app cannot hear me clearly?

It tells you so. If the recording has no usable speech, is too noisy or too quiet, Glosso says the attempt could not be evaluated instead of showing a low score. We never turn a listening problem into a 0/100.

Is Glosso really free?

Yes. Every lesson and every feature is free for everyone: no subscription, no trial that turns into a payment, no in-app purchases and no ads. There is no paid tier, and we have no plans to add one.

Then how is Glosso paid for?

By donations. Glosso is made by a small independent team, with no investors. Voluntary donations on Ko-fi pay for the servers that score your voice and serve the lessons, the audio for new lessons and languages, and the time to build what comes next. We do not sell or share your data, and we never will.

Does a donation unlock anything?

No. Everything is already free for everyone, whether you donate or not. A donation is a thank-you that keeps the project running, not a purchase.

Is Glosso open source?

Yes. The app, the phonetic engine and the server are free software under the GNU Affero General Public License (AGPL-3.0-or-later). The code is on GitHub: you can read how a score is computed, check what the app sends, build it yourself and change it. The Glosso name and logo are not covered by the licence, so a fork must use its own.

Can I run my own server?

Yes. The server is in the same repository, with a Docker setup for self-hosting. In the app, open Settings > Advanced > Server, enter your server's address and tap "Check and use". Your progress stays on the phone.

Is there a version without Google services?

Yes. The F-Droid version of the app contains no Google libraries, no Sign in with Google, and starts with usage statistics off. You can build it from the source today; see Get Glosso.

How can I help?

Use it and tell us what is wrong. Bug reports, fixes, translations and new languages are welcome in the repository; its CONTRIBUTING file explains how. Telling a friend helps too, and so does a donation if you can afford one.

Which languages and accents are available?

English (British and American), French, German and Russian. The engine is built so a new language is data plus a pack, not a rewrite, and contributions of new languages are welcome.

Does it work offline?

Lesson downloads let you read and listen without a connection. Scoring uses the cloud by default, so it needs a connection, unless you install the optional on-device model from Settings: then scoring also runs offline, without sending your voice anywhere.

Do I need to create an account?

No. There is no mandatory account and no sign-up wall. Progress is kept on your device. In the Google Play version, signing in with Google is optional and only lets you take your account to a new phone; we keep Google's account identifier, not your email or name. The app does not send an advertising ID or a hardware identifier; it uses a random installation id instead.

What devices are supported?

Android 8.0 or newer, with a microphone (built into every phone). An iOS version is in progress.

What data does the site and the app collect?

This website sets no cookies. It uses our self-hosted Umami analytics for aggregate page views and referrers, with no cookies, no IP address stored and no cross-site tracking. The app sends first-party usage statistics, including crash reports, to our own server, linked to a random installation id rather than to your name or email, and we do not store your IP alongside them. You can turn them off or delete them in Settings > Usage statistics. No advertising SDKs, no third-party trackers, no location, contacts or messages. Details in the privacy policy.

Is there a web app or a desktop version?

Not yet. Glosso is an Android app today, with iOS on the way. This site shows the real screens, but the full experience lives in the app.