MyLanguageVoiceTutor Try it free Features Contact

Features

Version 2.0.3 · free · no account · runs in your browser

MyLanguageVoiceTutor is a speaking-practice partner. You talk (or type), an AI tutor answers out loud in the language you're learning, and a set of practice tools helps you get from "I can't say anything" to holding a real conversation.

Three ways to practice

💬 Free chat

An open conversation with a friendly tutor who replies in the language you're practicing, keeps sentences at your level (Beginner, Basics or Intermediate), gently corrects mistakes and always ends with a question to keep you talking.

🎭 Role-play

Eight real-life scenes — café, asking for directions, hotel check-in, market, restaurant, train ticket, doctor, meeting someone new. The tutor plays the other person and stays in character; you get a short list of goals that tick off automatically as you complete them, with a 💡 example for each goal.

🔁 Repeat after me

Listen to a phrase (normal speed or 🐢 slowly), say it, and see word by word what was heard clearly, nearly, or missed — with a score and your best result saved. Practice 36 everyday phrases, the 20 most common verbs, your own vocabulary, or fresh sentences from the AI.

Help while you learn

  • Translate any tutor message into your native language with one tap, with a ↻ retry if it doesn't look right.
  • 💡 Suggested replies if you go quiet for a moment — a sentence you could say, spoken aloud, with its meaning in your language. You can switch this off.
  • Speak or type in your native language when you're stuck — the tutor still answers in the language you're practicing.
  • 🔊 Replay any message, hint or word out loud, at the speed you choose (0.5× to 1.2×), with a choice of voices.
  • Right-to-left support for Arabic, both as the language you practice and as your native language.

Vocabulary builder

  • Every word the tutor uses is collected automatically into your list, with how often you've seen it.
  • Tap Explain for a short meaning in your own language, 🔊 to hear it, ✕ to remove it.
  • Add the 20 most common verbs of any language with one tap.
  • Practice your own words in Repeat after me.

12 languages

English, Turkish, Spanish, Russian, Korean, Arabic, French, German, Italian, Japanese, Portuguese and Chinese (Mandarin) — each available both as the language you practice and as the language explanations are given in.

Private by design

  • The AI model runs on your own device's graphics card (WebGPU). Nothing you say or type is sent to a server.
  • No account, no sign-up, no ads, no tracking.
  • Your vocabulary, scores and settings are saved only in your browser. See the Privacy Policy.

Smart setup

  • Checks how much graphics memory your device has before downloading, picks the best model that fits, and warns you about models that would download and then fail.
  • Remembers the model that worked for you, so next time it loads in seconds from your browser's cache.
  • Keeps your screen awake during the download, and explains clearly what to do if your device runs out of graphics memory.

A note on accuracy — model size matters

The tutor is only as good as the AI model your device can run. Bigger models write more natural sentences, follow instructions better and translate more reliably. Smaller models are faster and fit on more devices, but they make noticeably more mistakes — odd word choices, grammar errors, wrong translations or suggestions that don't quite fit.

A more powerful device means a better tutor. Which models you can run depends on your device's graphics card (GPU) and how much graphics memory it has: a laptop or desktop with a stronger GPU can run the larger, more accurate models, while phones and older or basic computers are limited to the smaller ones. The app checks your device and picks the best model it can handle.

ModelGraphics memoryWhat to expect
Qwen2.5 7Babout 5–6 GBBest Most natural and accurate
Qwen2.5 3Babout 2.5–3 GBGood Occasional mistakes
Qwen2.5 1.5Babout 1.5–2 GBBasic Frequent mistakes — fine for simple practice
Qwen2.5 0.5Bunder 1 GBRough Many mistakes — a quick try only

Whichever model you use, treat corrections, translations and explanations as helpful guidance, not as an authority — double-check anything important. The hand-written parts of the app (the everyday phrases, opening greetings and role-play examples) don't depend on the model.

What you need

  • A recent Chrome or Edge (desktop works best) with WebGPU support.
  • A graphics card with enough free memory for the model you choose — see the table above.
  • A microphone for speaking practice (typing works too).
  • A one-time model download of about 1–5 GB, cached afterwards.

Try ver 2 now — it's free

Home Try it free Features Privacy Policy Terms of Service Contact

© 2026 MyLanguageVoiceTutor. Runs in your browser — nothing you say is sent to a server.

Version 2.0.3