Voxshelf

Every classic, read aloud. Voxshelf is a free, open source Android app for listening to the 20,000+ public domain audiobooks recorded by LibriVox volunteers. No ads, no tracking, no account.

Website: voxshelf.app

What it does

• Browse and search the whole LibriVox catalog, offline: rows by subject, Popular, New, Surprise me, and filters for length, reader type and era.
• Pick the best recording. Many books have several readings; Voxshelf ranks them by how much they're listened to and shows listener ratings and Open Library star ratings.
• Listen anywhere: background playback with lock screen and Bluetooth controls, downloads for offline listening (with a Wi-Fi-only option), playback speed, a sleep timer, and smart rewind after a long pause.
• Your shelf: Listening, To listen, Finished, Downloaded and your own lists, with progress and time left for each book.

Voxshelf is an independent app and isn't made or endorsed by LibriVox. If you enjoy the books, consider volunteering with LibriVox, or donating to the Internet Archive, which hosts the recordings (LibriVox itself no longer takes donations).

How it's built

Folder	What's in it
app/	The Android app: Kotlin, Jetpack Compose and Media3 (ExoPlayer).
catalog/	Python scripts that build the catalog the app ships with: LibriVox catalog, Internet Archive download counts, Open Library ratings and small cover images.
server/	A small Cloudflare Worker with a D1 database: recording ratings, problem reports, crash reports and cover images.
site/	The voxshelf.app website: plain HTML and CSS.
.github/workflows/catalog.yml	The nightly job that rebuilds the catalog and publishes it to the catalog release.

Building the app

You need JDK 21 and the Android SDK (compile SDK 37). Then:

./gradlew assembleDebug

The app expects a catalog at app/src/main/assets/catalog.db. Download the latest one from the catalog release (catalog.db.gz, then unzip it), or build it yourself with the scripts in catalog/ (see the nightly workflow for the exact steps).

Debug builds talk to the live ratings server at api.voxshelf.app. To run your own, see server/: npx wrangler dev runs it locally.

Building the catalog

Python 3.12, plus Pillow for covers. Roughly:

cd catalog
python fetch_catalog.py raw                 # LibriVox catalog (slow and gentle on purpose)
python fetch_popularity.py popularity.json  # Internet Archive download counts
python build_catalog.py raw catalog.db 20261023 popularity.json openlibrary.json

Privacy

Voxshelf collects almost nothing: anonymous recording ratings, optional problem reports, and crash reports without personal details. Problem and crash reports go to a separate private repo. The full policy is in PRIVACY.md and at voxshelf.app/privacy.

Built with AI

Voxshelf is developed with a lot of help from AI. Much of the code was written by an AI coding assistant (Anthropic's Claude), with a person directing the work, deciding on the design, and reviewing and testing the changes. If you spot something that looks wrong, please open an issue.

Contributing

Bug reports and ideas are welcome as issues. For bigger changes, please open an issue first so we can talk it through. Questions: [email protected].

Credits

• Recordings by LibriVox volunteers, in the public domain in the USA.
• Audio and cover art hosted by the Internet Archive.
• Book ratings from Open Library.
• Font: Nunito, SIL Open Font License (see licenses/Nunito-OFL.txt).

License

GPL-3.0. See LICENSE.