Learn a language by reading what you love.
glosa is a free, open source reader. Bring your own books1 and articles, read them with every word marked by how well you know it, and keep what you learn: for free2, on your own computer3.
- ¹ EPUB, TXT, HTML, or paste any text
- ² no account, no subscription. AGPL
- ³ nothing leaves it unless you pick a cloud AI
A page, as glosa reads it
Every word carries its colour: new, being learned, or known. Click one and you get the meaning it has in that sentence. Detected expressions are marked too, so you know where they start and end.
In relating the circumstances which have led to1 my confinement2 within this refuge for the demented, I am aware that my present position will create a natural doubt of the authenticity of my narrative. It is an unfortunate fact that the bulk of3 humanity is too limited in its mental vision to weigh with patience and intelligence those isolated phenomena, seen and felt only by a psychologically sensitive few, which lie4 outside its common experience.
- new never looked up
- learning saved, coming back in review
- expression found for you
- plain: known
try it: click a word in the passage to change how well you know it
- ¹ led to → llevar a
- ² confinement → encierro
- ³ the bulk of → la mayor parte de
- ⁴ lie → estar situado. Not “mentir”: this sentence decides
Meanings come in your language; here, Spanish. Text: H. P. Lovecraft, “The Tomb” (1917).
Yours, all of it
Your own texts. Read the novel on your nightstand, the newspaper you follow, the blog you like. No catalogue of lessons chosen for you: import an EPUB, a text file, a web page, or paste anything.
Your own computer. glosa runs on your machine with Docker. Its dictionary, its translation model and its voices can all run locally, so your books and your vocabulary stay with you. A cloud AI is there if you want it, and you see what every request costs.
Free, for good. No account, no subscription, no premium tier. The code is open under the AGPL, and your vocabulary exports to a plain CSV whenever you like.
- the translation model runs on your CPU
- the graphics card is optional
Index
- dictionary, Wiktionary in two editionsmeanings ranked for the sentence they appear in
- expressions, about 13,000, detected as you readsee also phrasal verbs
- grammar, structures marked in each sentence and explained
- importing, EPUB, TXT, HTML or pasted text
- phrasal verbs, found even when inflected or split“looked it up” is one unit, not three words
- pronunciation, real recordings for words; local voices for phrases
- review, spaced; words come back when you are about to forget them
- translation, a local model on your CPU, LibreTranslate, or a cloud AIevery translation is kept; the same text is never translated twice
- vocabulary, four levels from new to learned; export to CSV
Appendix: installing glosa
All you need is Docker. Everything else runs in containers, and nothing is installed on your system.
Get the code and start it.
$ git clone \ https://github.com/CoKeFish/glosa $ cd glosa && docker compose up -d
Optional: for local translation, install Ollama and its translation model.
$ ollama pull translategemma:4bOpen localhost:5174, import a book, and start reading.