Translation Aggregator sends one line of text to many translators at the same time and shows every answer in one window. Each translator gets its own pane. It’s a cross-platform rebuild, in Python, of the classic Windows Translation Aggregator. It does everything the original did except text hooking.

Why Compare Translators?

Machine translation of Japanese gets things wrong in different ways. One engine drops the subject, another gets the politeness level wrong, and a dictionary pane shows you what the words actually mean. With several answers side by side you can see where they agree and where they don’t, and work out which reading fits.

Panes fill in as each engine answers, so a slow engine never holds up the others. Most web engines answer in about a second.

            [ Japanese line (typed, pasted or copied) ]
                              │
                              ▼
                   [ Translation Aggregator ]
                              │
      ┌─────────┬─────────┬───┴─────┬──────────┬──────────┐
      ▼         ▼         ▼         ▼          ▼          ▼
   [Google]  [DeepL]  [Papago]  [ATLAS]   [JParser]  [OpenAI-
                                 (local)  (dictionary) compatible]
      └─────────┴─────────┴─────────┴──────────┴──────────┘
                              │
                              ▼
                 [ One window, one pane each ]

Who Is This For?

• Built for: Anyone reading Japanese, whether it’s game text from a text hooker, web pages or anything else you can copy, who wants several machine translations and a dictionary breakdown of the same line all at once, on Windows or Linux.
• Not for: Anyone who needs text hooking. AGTH-style process hooking is out of scope. Use a separate hooker such as Textractor and turn on Auto Clipboard mode.

Key Features

1. 15 Engines in Parallel

• Web translators: Google (switches to a fallback when rate-limited), Bing, DeepL (free web mode or your own API key), Yandex, Papago, Caiyun 彩云小译, Systran, Babelfish, and Baidu.
• Local engines (Windows): Fujitsu ATLAS V14 (Japanese⇄English) and LEC Nova from Power Translator 15 (Japanese→English).
• Dictionaries: WWWJDIC, Jisho, Babylon, a JParser pane (readings above every word, hover for the meaning, handles conjugations), and a MeCab pane with furigana.
• LLMs: Any OpenAI-compatible chat endpoint, such as OpenAI, OpenRouter or LM Studio.

Baidu is slow, about 10–20+ seconds per line. It drives a real browser through Playwright because its API is locked behind an obfuscated per-request token. You can untick it if you don’t need it.

2. Layout You Can Rearrange

Choose 1–3 columns, drag panes anywhere, or tear a pane off into its own floating window and drag it back later. The layout is saved between runs. You can also set always-on-top, window opacity and the pane font, and lock the layout.

3. Auto Clipboard and History

Tick Auto Clipboard and anything you copy is translated straight away. If lines arrive quickly they are queued, not dropped. Alt+Left and Alt+Right step back and forward through lines you’ve already translated.

4. Command Line

The same engines are available without the GUI:

transagg "日本語の文章です" -s ja -d en --translators google,deepl,papago

The default set is google,bing,deepl,yandex,wwwjdic.

Install

• Prebuilt: The releases page has a Windows .exe and a Linux AppImage, with the JParser dictionaries included. You don’t need Python. The AppImage runs the command-line tool if you pass --cli. MeCab isn’t bundled because its dictionary is large; install it from Tools → Install MeCab (about 50 MB, downloaded once).
• From source: Python 3.9+ (3.11+ recommended):

git clone https://github.com/PatchScratch/Translation-Aggregator
cd Translation-Aggregator
python -m venv .venv && source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -e ".[full]"
transagg-gui

• Baidu engine: Also run playwright install chromium (about 150 MB).
• ATLAS / LEC: Install the Windows software to its default location, and install a 32-bit Python (py --list should show a -32 entry, or set TA_ATLAS_32_PYTHON). Both engines are 32-bit DLLs, so the app talks to them through a small bridge process.

Settings are saved in ~/.config/TranslationAggregator/config.json.

Project Info

• Repository: github.com/PatchScratch/Translation-Aggregator
• Latest release: v1.1.1 (7 Oct 2026). Main has one later commit (7 Oct 2026) that adds the in-app MeCab installer.
• License: GPL-2.0-or-later

Translation Aggregator builds on the original Translation Aggregator project, EDRDG dictionary data, jisho.org, MeCab and PyQt6.