audiobooker 0.8.0a1

Audiobook scraper — search and stream from Librivox, LoyalBooks, and more

AudioBooker

Python library for searching and streaming free audiobooks from multiple sources. Parallel search across all sources, fuzzy matching, relevance scoring, and a unified AudioBook dataclass — one API regardless of where the book comes from.

Supported Sources

Source	Site	Catalogue	Native Search	Genres / Tags
Librivox	librivox.org	~18 000 books	title, author, narrator, tag (REST API)	30+
LoyalBooks	loyalbooks.com	~3 500 books	title, author (sitemap), tag (genre pages)	41
StephenKingAudioBooks	stephenkingaudiobooks.com	~113 books	full-text site search	—
GoldenAudioBooks	goldenaudiobook.co	~6 500 books	title, author, tag (linear scan)	—
AudioAnarchy	audioanarchy.org	~11 books	title, author, tag (linear scan)	Anarchy, Radio Drama
DarkerProjects	darkerprojects.com	~244 episodes	title, author, tag (linear scan)	Audio Drama
HPTalesAudioBooks	hpaudiotales.com	~20 books	title, author, tag (linear scan)	Harry Potter

YouTube sources (optional, requires pip install audiobooker[youtube]):

Source	Channel	Content	Tags
TheCybrarian	@TheCybrarian	Robert E. Howard fiction (Conan, Solomon Kane, Kull…)	Fantasy, Sword and Sorcery, Robert E. Howard
HorrorBabble	@HorrorBabble	Horror short fiction narrated by Ian Gordon	Horror, Lovecraft, Weird Fiction

Total indexed: ~28 000+ titles across 7 web sources + 2 YouTube channels.

LoyalBooks genres (41): Action and Adventure, Ancient Texts, Animals, Art Design and Architecture, Biography and Memoir, Children in Fiction, Children Non-fiction, Classics (Antiquity), Comedy and Humour, Drama, Early Modern, Fantasy, General Fiction, Historical Fiction, History, Horror and Supernatural Fiction, Humor, Instruction and How-To, Language, Literary Fiction, Love Romance and Marriage, Modern (19th C), Music and Theatre, Myths Legends and Fairy Tales, Nature and Wildlife, Non-fiction, Philosophy, Poetry, Politics and Economics, Psychology, Religion, Science, Science Fiction, Short Stories, Short Works, Spiritual and Inspirational, Sport and Recreation, Tragedy, Travel and Geography, War and Military, Westerns.

YouTube support

Install the optional YouTube extra:

pip install audiobooker[youtube]
# or
pip install tutubo

Use the pre-configured channel sources or define your own:

from audiobooker.scrappers.youtube import HorrorBabble, TheCybrarian, YoutubeChannelSource
from audiobooker.base import BookAuthor

# Pre-configured channels
for book in HorrorBabble().iterate_all():
    print(book.title, book.streams)  # streams = YouTube watch URLs

for book in TheCybrarian().search_by_title("Conan"):
    print(book.title, book.runtime)

# Custom channel
my_channel = YoutubeChannelSource(
    channel_url="https://www.youtube.com/@SomeChannel/videos",
    authors=[BookAuthor(last_name="Unknown")],
    tags=["Audiobook"],
    language="en",
    min_runtime=300,  # skip anything under 5 minutes
)
for book in my_channel.iterate_all():
    print(book.title)

# Custom playlist
from audiobooker.scrappers.youtube import YoutubePlaylistSource
playlist = YoutubePlaylistSource(
    playlist_url="https://www.youtube.com/playlist?list=PLxxxxxx",
    authors=[BookAuthor(last_name="Various")],
    tags=["Horror"],
)
for book in playlist.iterate_all():
    print(book.title, book.runtime)

When tutubo is installed, TheCybrarian and HorrorBabble are automatically included in ALL_SOURCES and participate in all unified search*() calls.

Install

pip install audiobooker

Unified search

Search all sources in parallel — results arrive sorted by relevance score.

from audiobooker import search, search_by_author, search_by_title, search_by_tag, search_by_narrator

# Search all sources, deduplicated, scored, timeout 30s
for book in search("Lovecraft", max_per_source=5, timeout=30):
    print(f"[{book.score:.2f}] [{book.source}] {book.title}")
    print(f"  author={book.authors}  streams={len(book.streams)}")

# Targeted searches
for book in search_by_author("Dickens", max_per_source=5):
    print(book.title)

for book in search_by_title("Sherlock Holmes", max_per_source=5):
    print(book.title, book.language)

for book in search_by_tag("horror", max_per_source=5):
    print(book.title)

for book in search_by_narrator("Frank Muller", max_per_source=5):
    print(book.title, book.narrator)

Search parameters

Parameter	Default	Description
sources	all 7	list of instantiated AudioBookSource objects to restrict search
max_per_source	10	max results collected per source before stopping that thread
timeout	30.0	seconds before slow sources are cancelled
deduplicate	True	skip books with identical title+author from a second source

Scoring

Each result carries a score field (0.0–1.0) computed by score_book(). Weights depend on the search method so cross-field contamination is avoided:

Method	Title	Author	Tag	Narrator
search_by_title	100%	—	—	—
search_by_author	—	100%	—	—
search_by_tag	—	—	100%	—
search_by_narrator	—	—	—	100%
search	55%	30%	10%	5%

Uses rapidfuzz WRatio — handles token reordering, typos, and partial matches. Title scoring adds a containment bonus when all query words appear verbatim in the title.

Results scoring below 0.45 are filtered out automatically.

Per-source usage

All scrapers share the same interface via AudioBookSource.

from audiobooker.scrappers.librivox import Librivox
from audiobooker.scrappers.loyalbooks import LoyalBooks
from audiobooker.scrappers.goldenaudiobooks import GoldenAudioBooks
from audiobooker.scrappers.audioanarchy import AudioAnarchy
from audiobooker.scrappers.darkerprojects import DarkerProjects
from audiobooker.scrappers.hpaudiotales import HPTalesAudioBooks
from audiobooker.scrappers.stephenkingaudiobooks import StephenKingAudioBooks

# Common interface
source = Librivox()
source.search(query)               # title + author + tag
source.search_by_title(query)
source.search_by_author(query)
source.search_by_tag(query)
source.search_by_narrator(query)
source.iterate_all()               # every book in the catalogue
source.iterate_popular()           # front-page / curated selection
source.iterate_by_author(author)
source.iterate_by_tag(tag)

Librivox — REST API, fastest source

lv = Librivox()
for book in lv.search_by_author("Lovecraft", max_per_source=5):
    print(book.title, book.runtime, "s")
for book in lv.search_by_narrator("LibriVox"):
    print(book.title, book.narrator)

LoyalBooks — sitemap + genre pages

lb = LoyalBooks()
for book in lb.search_by_tag("Horror and Supernatural Fiction"):
    print(book.title)             # uses genre page, not linear scan
for book in lb.iterate_popular():
    print(book.title)             # front-page featured books

Linear-scan sources

GoldenAudioBooks, AudioAnarchy, DarkerProjects, HPTalesAudioBooks, and StephenKingAudioBooks all support iterate_all(). StephenKingAudioBooks also has a native site search for title/author queries.

for book in AudioAnarchy().iterate_all():
    print(book.title, book.tags)   # tags: ["Anarchy"] or ["Anarchy", "Radio Drama"]

for book in DarkerProjects().iterate_popular():
    print(book.title)              # front-page shows

AudioBook dataclass

@dataclass
class AudioBookChapter:
    title: str   = ""
    offset: float = 0.0   # seconds from start of book
    runtime: float = 0.0  # seconds
    stream: str  = ""     # per-chapter audio URL
    image: str   = ""

@dataclass
class AudioBook:
    title: str          = ""
    description: str    = ""
    image: str          = ""   # cover art URL
    language: str       = ""   # ISO 639-1 code (normalised from source)
    authors: List[BookAuthor]              = field(default_factory=list)
    tags: List[str]                        = field(default_factory=list)
    streams: List[str]                     = field(default_factory=list)  # direct audio URLs
    narrator: Optional[AudiobookNarrator]  = None  # primary reader
    narrators: List[AudiobookNarrator]     = field(default_factory=list)  # full reader cast
    chapters: List[AudioBookChapter]       = field(default_factory=list)
    genres: List[str]                      = field(default_factory=list)  # taxonomy genres
    year: int           = 0
    runtime: int        = 0    # seconds (where available)
    source: str         = ""   # e.g. "Librivox", "LoyalBooks"
    score: float        = 0.0  # relevance score from last search (0..1)
    codec: str          = ""   # e.g. "mp3"
    bitrate: str        = ""   # e.g. "128"
    external_ids: dict  = field(default_factory=dict)  # e.g. {"librivox_id": "47"}

    def has_live_streams(self) -> bool: ...  # HEAD-checks stream URLs

AudioBook supports == and hash() based on (title, sorted authors) — use a set to deduplicate across sources.

Utilities

from audiobooker import score_book, iter_sitemap_urls, check_url_availability, normalize_language

# Score a book against a query manually
score = score_book("Lovecraft", book, method="search_by_author")

# Walk any sitemap or sitemap index recursively
for url in iter_sitemap_urls("https://example.com/sitemap.xml"):
    print(url)

# Check if a stream URL is reachable
if check_url_availability("https://example.com/book.mp3"):
    print("live")

# Normalise language strings to ISO 639-1
normalize_language("English")   # → "en"
normalize_language("en-US")     # → "en"

mediavocab integration

mediavocab is a hard runtime dependency. Every AudioBook can be projected into the typed mediavocab.Release schema via audiobook_to_release():

from audiobooker import search, audiobook_to_release

# Search → typed mediavocab Release with parsed_license filtering
for book in search("Lovecraft", max_per_source=3):
    release = audiobook_to_release(book)
    if release.parsed_license and release.parsed_license.is_open():
        # public domain / CC-licensed: free to redistribute
        print(release.work.title, release.parsed_license.identifier)

The converter populates a wide swath of the Release / Work schema:

mediavocab field	Source data
Work.title, Work.year, Work.runtime, Work.language	direct
Work.content_genres	AudioBook.genres (e.g. LibriVox genres)
Work.credits	authors → RelationRole.CREATOR, every reader → RelationRole.PERFORMER
Work.external_ids	librivox_id and any other typed ID the source supplied
Release.chapters	AudioBook.chapters → Chapter(offset, end, title)
Release.codec, Release.bitrate	LibriVox publishes 128 kbps MP3 by policy
Release.audio_language	mirrors Work.language
Release.license	public_domain for LibriVox / LoyalBooks
Release.release_date	IsoDate-compatible YYYY from AudioBook.year

LibriVox emits one Release per book with full per-section chapters and a deduplicated reader cast. Other sources populate whatever subset their public data exposes — fields are only set when the source actually carries the data.

Error handling

Network failures and malformed pages are swallowed per-item — a bad page never aborts an iterate_all() run. If a source site is down or has restructured its HTML, that scraper silently yields nothing.

License

MIT