- Free allowance
- Free public-domain audiobooks, read by volunteers, with a JSON/XML catalogue API.
- Verdict
- ★ qualifiesusable as stated; the read path was probed
- Cost
- free tier
- Identity
- anonymousno card, no phone
- Binding limit
- per-ip-shapingaudio is fetched from archive.org, which shapes the caller (measured 1.99-3.17 MiB/s on this machine)
- Evidence
- documented4 sources; a first-party page was read and the sentence quoted
- Freshness
- stale on 2026-11-23checked 2026-10-23 (live); stale on 2026-11-23
| Free tier | Free public-domain audiobooks, read by volunteers, with a JSON/XML catalogue API. The API caps a single response at 500 records ('Can no longer be set above 500', added 2026-09-26) and pages with offset. unit unlimited; metered on none; The catalogue size is not published; the highest project id returned by sort_order=desc was 23017, which bounds it near 23,000 projects. |
|---|
| Price | USD 0 |
|---|
| Limits | - request-rate 500 records maximum per request; 'If you intend to access a large number of records this way, please be considerate, and add a delay between your requests.' source
- per-ip-shaping (binding) audio is fetched from archive.org, which shapes the caller (measured 1.99-3.17 MiB/s on this machine) source
|
|---|
| Binding | per-ip-shaping The API is metadata and would take minutes to walk at 500 records a request; the bytes come from archive.org, so the Internet Archive's per-IP ceiling is what decides a TiB-scale pull. |
|---|
| Identity | anonymous No key and no account: the API answered an anonymous GET with 200 and 3,084 bytes for limit=2. |
|---|
| Egress | Metadata is free and unmetered from the API. The audio is not served by librivox.org: each record's url_zip_file points at archive.org/compress/<item>/..., so audio bytes are Internet Archive egress and inherit its per-IP shaping. free allowance: no byte allowance published |
|---|
| Per-object cap | not applicable: one MP3 per section, one ZIP per book. The ZIP is served by archive.org, so the object size is the Internet Archive item's size. |
|---|
| Retention | permanent: public-domain recordings 'donated to the public domain', with the audio stored as Internet Archive items |
|---|
| Interface | rest, cli GET https://librivox.org/api/feed/audiobooks with id, since, author, title, genre, fields, extended, coverart, limit, offset, format, sort_order. Parameters may be path segments instead of query strings. |
|---|
| Tooling | curl/python-requests; any HTTP client. No SDK and no rclone backend is needed because the metadata is a plain JSON API and the audio is a plain archive.org URL. |
|---|
| Composes with | Harvest the catalogue through the API (a few hundred requests), then pull the MP3s from archive.org - the same egress path and the same shaping as any other Internet Archive item. |
|---|
| Reliability | The API changed in 2026-09-26 to cap limit at 500; older clients that asked for more are silently clamped rather than rejected. |
|---|
| Notes | The interesting part is where the bytes are not: librivox.org serves a small JSON catalogue and archive.org serves the audio. A plan that expects a fast audiobook mirror from this API alone will move metadata and no media. |
|---|
| Evidence note | documented The API contract is the vendor's own page and was exercised here for two records; the audio path is read from the record's url_zip_file field, not downloaded. Downloading one ZIP would raise it. |
|---|
| Parsed numbers | - has-free-tier 1
- per-ip-shaping 3.17 MiB
- price USD 0
|
|---|