Community library

The library is new. The first books are still being produced, so the shelves may be empty or nearly empty when you visit. That is expected for now. This page describes how the library works, not what is on its shelves.

What it is

Firebrat Community is a small website with one job: help people find and share audiobooks that anyone may freely copy and listen to. It has three parts that fit together.

The audio is hosted on the Internet Archive. A book is one item there: a zip (or a folder) in the ordinary Firebrat book format. Downloads come straight from archive.org.
The listings are kept in Firestore. Title, author, language, category, licence, the chapter list, ratings, comments and reports live in the project’s Firebase database. No audio is stored there.
It is a static website with sign-in through Firebase Auth (Google or email). The same site is built into the web app as the Discover tab and into the Android app as the Community screen.

Because every book uses the standard package format, a book from the library opens in Firebrat exactly like one you converted yourself: sentence-by-sentence highlighting, tap-to-seek, and offline playback once downloaded. See DATA_SCHEMA.md for the format.

Listen and download

You can browse the library without an account. Signing in is needed to rate, comment, report, submit or save favourites.

On the web

Open the live library directly, or use the Discover tab inside the Firebrat web app.
Search by title or tag, and filter by category, language and licence. Sort by newest, top rated or longest.
A book page shows the cover, rating, the “How this audio was made” card, and a chapter list. Chapters play right in the page, and a mini-player stays at the bottom while you browse.
Download gives you the zip from the Internet Archive. Import it in the Firebrat app (the import button in your library) and it works fully offline.

On Android

Open Community from the library screen, pick a book and choose Add to my library. It downloads from archive.org hosts only, and respects the Download over setting (Wi-Fi only, Wi-Fi and cellular, or ask on cellular).
A download that is interrupted restarts from the beginning.
Not yet tried on a device The Community screen is covered by unit tests and static analysis, but has not yet been exercised on a phone.

Submit a book

You need a Firebrat account, a finished audiobook package, and the right to share it (see what counts as copyright-free). The Submit page is a four-step wizard: Source, Package check, Details and provenance, then Rights attestation and review.

The package

A zip (or an unzipped folder) containing manifest.json (schema version 1.0, book id, title, author, narrator voice, sections), sections/<id>/audio.m4a with its segments.json, and an assets/ folder. This is the same output the Firebrat converter makes, so any book you converted yourself already has the right shape. The wizard validates the manifest before you go on.

Two routes

Route A: I already have it on the Internet Archive

Paste the archive.org URL (a /details/ or /download/ link). The site checks live, through the Archive’s metadata API, that the item really contains your package.

Route B: upload it for me

The site uploads your zip to the Internet Archive from your browser, using your own Internet Archive S3 keys from archive.org/account/s3.php. The keys are kept in memory only: never stored, never logged, never sent to Firebrat’s database. If the upload fails, the page falls back to manual upload instructions and then to route A.

Why you choose your local file. archive.org downloads send no CORS headers, so a web page cannot read manifest.json from the Archive after the fact. Instead the wizard reads the manifest from your local copy of the zip or file, in your browser only, and stores the chapter list with the listing. Nothing from your machine is uploaded in route A. Audio then streams from the Archive with a plain audio element, which needs no CORS.

Status. Uploading from the browser (route B) has not yet been run against the live Internet Archive; only the preflight request was checked.

The “How this audio was made” section

Every submission must say how its audio was produced. It is shown to listeners on the book page, so nobody has to guess whether a voice is human or synthetic. You provide:

Field	What to enter
Generation method	Firebrat TTS, human-narrated, mixed, or other.
Narration engine and models	Which voice engine and which models (for example, the language model that cast the characters) were used.
Pipeline notes	Anything a listener or reviewer should know about how the book was made.
Source text	The title and a link to the text the audio was made from.
Licence	Public domain, CC0, CC BY or CC BY-SA, with a link to the licence or the evidence for it.
Attestation	A required tick box: you confirm you have the right to share the work under that licence.

What counts as copyright-free

Only submit works you are confident anyone may copy and share. In practice that means:

Public domain in the United States. Works first published in the US before 1931, and older works generally.
CC0 (public-domain dedication) and CC BY or CC BY-SA works, where you give the credit those licences require.

Read these cautions before you submit.

Public domain is not the same everywhere. A work that is free in the United States may still be protected elsewhere, for example in countries where copyright lasts for 70 years after the author’s death. If the author died less than 70 years ago, think carefully, because the audiobook may not be freely shareable worldwide.

Translations and editions can be copyrighted even when the original is not. A modern translation, an annotated edition, an introduction or an illustrator’s work can each carry its own copyright. Use a source whose own licence you can point to.

Project Gutenberg. Most Gutenberg texts of public-domain works are free to use, but the “Project Gutenberg” name is a trademark and its boilerplate and licence text come with conditions. Do not use the name to suggest endorsement, and follow the licence shown with the text you used. A stripped, clean text is still the same work; check the individual book’s licence page.

This page is general guidance, not legal advice. If you are unsure, do not submit.

Ratings, comments, reports and bugs

Ratings. Signed-in listeners give a book 1 to 5 stars. You can change your rating later.
Comments. Signed-in listeners can comment on a book and delete their own comments.
Report an offensive comment. Use the report link on a comment and choose offensive, spam, harassment, misleading or other. Admins review reports and can hide a comment.
Report a book the same way if its description or audio is misleading.
Bug reports and feedback. The Report a bug page sends a note to the admins. You can also open an issue on GitHub.

Copyright takedown notices

If you believe a book in the library infringes your copyright, you can send a notice. You do not need an account.

Fill in the takedown form: the book, who you are, the work you own, the links you believe infringe, a good-faith statement, an accuracy statement and your signature.
Admins review every notice. While it is open, an admin may disable the book by marking it disputed, so it stops being offered while the question is resolved.
If the book’s submitter disagrees, a counter-notice can be sent through the GitHub issue link given on the legal page. No personal email address is published.

Please be accurate. A knowingly false claim can have legal consequences for the person who files it.

The copyright, terms and privacy pages in the library carry the full wording.

What is not finished. The database security rules for the library are written but have not yet been run in the Firestore emulator, and there is no rate limiting because the site has no server-side functions. Both are on the to-do list before the library is widely promoted.