Playto is a real-time game overlay translator and a language-learning tool at once. Local OCR reads any PC game's on-screen text and AI shows the meaning in place — dialogue, menus, tooltips, anywhere you point. Immersion learning, as you actually play. 23 languages · Offline · No subscription.

Sign in to add this item to your wishlist, follow it, or mark it as ignored

Early Access Software

Get involved with this software as it develops.

Note: This Early Access software is not complete and may or may not change further. If you are not excited to use this software in its current state, then you should wait to see if it progresses further in development. Learn more

What the developers have to say:

Why Early Access?

“Playto combines real-time game translation with language learning — a category that doesn't really exist yet. The translation engine works, but the learning system around it is still being built. Which games to prioritize, which languages to expand next, what study tools actually help retention — these decisions should be driven by real users playing real games. Early Access is the right place to build this together.”

Approximately how long will this software be in Early Access?

“Approximately 6–12 months. The timeline depends on how quickly we can ship the planned features (see below) and validate them with community feedback. We'll leave Early Access when the learning tools, language coverage, and community features reach a level where we're confident in the full experience.”

How is the full version planned to differ from the Early Access version?

“The full version will add significant features beyond the current Early Access build:

- **Expanded language pair support** — The EA version works best with English ↔ Japanese. Full version targets reliable quality across CJK languages (Chinese, Korean) and major European languages, with per-language tuning.
- **Community game glossaries** — A shared glossary system where players contribute and refine translation dictionaries for specific games. Each game has unique terminology that generic translation misses.
- **Learning analytics and progress tracking** — Vocabulary growth visualization, difficulty-adapted word suggestions, and data-driven study recommendations based on your play history.
- **Mobile companion app** — Review your saved vocabulary and practice quizzes on your phone, synced with your PC dictionary. Study on the go without your gaming PC.
- **Improved quiz and practice modes** — More exercise types beyond the current spaced repetition flashcards, including context-based recall and game-scene association.
- **Per-game translation profiles** — Optimized settings and prompts for popular games, developed with community feedback to handle game-specific fonts, UI layouts, and terminology.
- **Content engine** — A public database ranking games by language difficulty (CEFR level), helping learners pick games that match their skill level.

Some of these features depend directly on Early Access user feedback to shape their design.

What is the current state of the Early Access version?

“The translation foundation is functional: real-time OCR screen capture, cursor-point translation, text-to-speech, a personal dictionary, and basic spaced repetition quizzes. It works with any PC game. All translation processing runs locally on your machine using open-source AI models — no internet connection required during gameplay.

What's still early-stage: translation quality varies significantly by game (font style, art direction, and text density all affect accuracy). Language support beyond English ↔ Japanese is functional but not yet well-tuned. The learning tools (quizzes, progress tracking, vocabulary review) are basic — they work, but they don't yet adapt to your level or give you meaningful feedback on your progress. There are no community features yet (glossary sharing, game profiles). These are the areas where Early Access development is focused.”

Will this software be priced differently during and after Early Access?

“We plan to increase the price when Playto leaves Early Access, as the feature set will be substantially larger. Getting Playto during Early Access is the best deal.”

How are you planning on involving the Community in your development process?

“Community input directly shapes what we build next:

- **Game requests** — Tell us which games you want better support for. We prioritize per-game translation tuning based on what the community is actually playing.
- **Language pair priorities** — Which language combinations matter most to you? Community demand determines our expansion order.
- **Bug reports and accuracy feedback** — Reports on specific games help us improve the OCR and translation pipeline where it matters.
- **Glossary contributions** — We're building a community glossary system where players can share and refine game-specific terminology. Your corrections make Playto better for everyone playing the same game.
- **Feature feedback** — Steam discussions and our Discord server are where we hear what's working, what's not, and what's missing. Development priorities shift based on this feedback.”
Read more

Download Playto Lite: Game Screen Translator

Buy Playto

 
See all discussions

Report bugs and leave feedback for this software on the discussion boards

About This Software

Learn languages through the games you love. Playto reads the on-screen text of any PC game with local AI and shows the meaning right there as a translation overlay — JRPGs, visual novels, strategy games, anything with text on screen.


There's no fiddly setup that gets between you and the game. Pick how it reads to fit how you play: keep a fixed region in view, follow your mouse, read the whole screen, or show a translation only when you ask. Save the words and sentences you care about to your Word Book, and the ones you register quickly become study material — flashcards and more — that deepen your grasp of the game. This is immersion learning: play with Playto, and the words you run into again and again start to make sense without stopping to look them up.

Translate any spot you point at — dialogue or UI


Cursor-follow mode reads the text wherever you point your mouse — dialogue, menus, item descriptions. Fixed-region mode watches a set area such as subtitles or a status bar. Window mode (beta) translates the whole screen in the background and shows the overlay only when you want it. Whatever the game, you read any spot you like — not just subtitles. Captured text appears as an overlay with the original and its meaning, without pausing the game.

Fits your screen — and your taste

Choose where the overlay lives: on the original text, lined up in a bottom bar, or anywhere you place it. Tune background opacity, font size and difficult-word highlights until it blends into the game. Watch several spots at once with multiple capture regions — dialogue, name plates and quest logs each get their own translation.


Local AI, arranged your way: it works out of the box with recommended defaults, and you can go further — pick from bundled models such as Gemma and Qwen, keep a lightweight model on translation so your game never slows down, and let a larger model handle word and sentence explanations. In the translation log, words carry difficulty badges (JLPT N5–N1 for Japanese, CEFR A2–C2 for other languages) so you can tell at a glance which ones deserve your attention.

After playing, grow your Word Book


When you stop playing, a session summary lets you look back over the text you met today. Nothing is saved automatically — only the words and sentences you choose to keep enter your Word Book. So your Word Book becomes a learning record you selected yourself.

Keep the scenes you played, in an album


The screen as you saw it, the original text, and its meaning. Look back after you play, and keep the scenes in your album.

Make it stick with review and quizzes


Saved items go into SRS (spaced repetition) timed to the forgetting curve, and 8 quiz modes turn them into lasting memory. The more you play, the more your own dictionary grows — as wide as the games you've played.

From reading the screen to logging every line

Playto's work goes well beyond the translation itself. Playto reads the screen three ways, and you pick which one fits the game. Text runs OCR alone — the fastest and lightest. Text+ runs OCR with AI vision behind it, and applies your glossary and context. Image hands the whole screen to a vision model that reads the picture itself: stylized fonts, brush calligraphy, vertical Japanese text, and lettering painted into a texture are read this way — and it picks up who is speaking and adjusts the tone to the scene. Playto then sorts what it read into dialogue, menus, and item descriptions and breaks it into units a language model can translate cleanly. The meanings are produced by a local LLM running on your own PC — you choose which model to run, and you can change it whenever you like. And every translation that appears on screen is kept in a log you can read back after you play. A per-game glossary pins character names and terms the way you want them.

Game Audio (Beta)

Playto can now listen to the game itself. Voice lines are transcribed while you play, translated, and collected in their own log — tap any word for a dictionary lookup, or save a line to your Sentences with its original audio clip attached.


Works per-game: only the game's own audio is captured, never your mic, music apps, or voice chat.


Currently verified with Japanese and English game audio. Other languages may work — let us know how it goes. Available in the paid edition on Windows 10 (2004) or later.

Playto Cloud (Beta)

Prefer not to wait for the translation model to download — or running one isn't an option on your PC? Playto Cloud translates in the cloud instead. Opt in and start translating right away, while the local model keeps downloading in the background. Free trial credits are currently included.


Cloud translation is optional and off by default. Playto stays local-first — the cloud receives only what's needed to translate; see our privacy policy for details.

Key features

  • Real-time translation overlay — Reads on-screen text and shows the original together with its meaning, without pausing the game and without Alt-Tab.

  • Capture modes — Cursor-follow works like a mouse tooltip translator, reading wherever you point — dialogue, item tooltips, menus, UI labels. Fixed-region watches a set area; window mode (beta) translates the whole screen in the background. Read any spot you like.

  • Automatic log — Everything you encounter is recorded. Look back anytime during play to check meanings or pronunciation.

  • Word Book + SRS + 8 quiz modes — Save the words and sentences you want to remember, and make them stick with spaced repetition and quizzes (Flashcard, Multiple Choice, Fill Blank, Dictation, and more).

  • Album — Save captured scenes as full images with their translation, to look back on memorable moments anytime.

  • Anki export — Take your study out of Playto: export your Word Book as a TSV file, or package Album scenes with their images as an Anki deck (.apkg) for spaced-repetition review anywhere. A natural fit for sentence mining.

  • Three reading modes — Text for OCR-only speed, Text+ for everyday play (glossary and context aware), and Image, where a local vision model reads the screen itself: stylized fonts, brush and fantasy lettering, vertical Japanese writing, and text painted into textures, plus speaker and scene awareness. Switch any time.

  • Per-game glossary — Register character names and custom terms, and they're reflected in that game's translations.

  • Multiple capture regions — Watch dialogue, name plates and quest logs at the same time, each with its own overlay.

  • Overlay customization — Position (on-text / bottom bar / custom), background opacity, font size, difficult-word highlights.

  • Difficulty badges — Words in the log carry JLPT N5–N1 (Japanese) or CEFR A2–C2 (other languages) marks.

  • Flexible local models — Pick from bundled models such as Gemma and Qwen; keep a light model on translation and a larger one on explanations. Recommended defaults out of the box.

  • Guides and support built in — In-app guides for each use case, and support contact right from the app.

  • Fully local — Translation runs on your own PC and works offline while you play — nothing on your screen leaves your machine unless you opt in to Playto Cloud. Supported languages: Chinese (Simplified), Chinese (Traditional), Czech, Danish, Dutch, English, Finnish, French, German, Greek, Hungarian, Italian, Japanese, Korean, Norwegian Bokmål, Polish, Portuguese, Romanian, Russian, Spanish, Swedish, Turkish, Vietnamese.

  • Game Audio (Beta) — voice lines transcribed, translated, and saved with their original audio (paid edition).

  • Playto Cloud (Beta) — optional cloud translation with free trial credits; start before the model finishes downloading.

Before you start

  • Setup is guided — A setup wizard walks you from install to your first translation, model downloads included.

  • Try Playto Lite first — Local AI needs GPU VRAM in addition to your game (typically 2–4 GB extra). Playto Lite is the free version — please confirm compatibility before purchasing.

  • Text-to-speech — Hear captured sentences read aloud. Available for these languages: English, Japanese, Spanish, French, Italian and Portuguese.

  • How translation works — Word meanings come from a dictionary. Sentences are translated by the local LLM you choose, shown alongside the original so you always have the real text in front of you.

AI Generated Content Disclosure

The developers describe how their game uses AI Generated Content like this:

Playto uses AI models (LLMs) to generate real-time translations of on-screen game text. All AI processing runs entirely on your machine using local models — no data is sent to external servers. The AI is also used for OCR (optical character recognition) to read text from game screens. No pre-generated AI content is included; all translations are generated live during use.

System Requirements

    Minimum:
    • Requires a 64-bit processor and operating system
    • OS: Windows 10 64-bit
    • Processor: Intel Core i5 / AMD Ryzen 5 (with AVX2 support)
    • Memory: 8 GB RAM
    • Graphics: DirectX 11 compatible GPU. Local AI runs on the CPU by default; an NVIDIA (CUDA) or Vulkan-capable GPU enables faster acceleration
    • DirectX: Version 11
    • Network: Broadband Internet connection
    • Storage: 5 GB available space
    • Additional Notes: AI models (about 2–6 GB) download on first run. For GPU acceleration, keep your GPU driver up to date — 4 GB+ VRAM recommended. No dedicated GPU? CPU mode works (slower)
    Recommended:
    • Requires a 64-bit processor and operating system
    • OS: Windows 11 64-bit
    • Processor: Intel Core i7 / AMD Ryzen 7
    • Memory: 16 GB RAM
    • Graphics: NVIDIA GPU (CUDA) or Vulkan-capable GPU, 6 GB+ VRAM, with an up-to-date driver
    • DirectX: Version 11
    • Network: Broadband Internet connection
    • Storage: 10 GB available space
    • Additional Notes: Keep your GPU driver current (especially on the latest-generation GPUs). If CUDA isn't supported, Vulkan and CPU modes are available.
Review Filters