About Verbum
Last updated: 2026-08-26
Verbum reads the Bible and early Christian literature aloud
in the languages they were written in. The name comes from the Latin
Vulgate of John 1:1 — In principio erat Verbum.
Each chapter is a video: the original text on screen, read in a reconstructed
pronunciation, with subtitles you can switch between the original, Chinese, English,
or all three at once. Everything is aligned verse by verse — tap a verse
to jump to it, tap a word to see its gloss and morphology.
What's new
25 August 2026 — Word-level highlighting: each word turns yellow as it
is spoken
Chapter audio on the reading pages now marks your place
word by word, in all four languages — Hebrew, Greek,
Latin and Ge'ez. Until now it highlighted only the current verse, and a verse is often
twenty or thirty words; if you cannot yet recognise the script by ear, that still left
you guessing.
The timings are not estimated afterwards, nor obtained by forced
alignment: they come from the duration the speech model itself computes for every
phoneme as it synthesises, so they match the audio word for word.
Because this only surfaces data the model was already producing and discarding,
the voices themselves are unchanged — no weights were touched, and the
same text synthesises to a bit-for-bit identical waveform.
Four languages
| Language | Corpus |
|---|
| Biblical Hebrew | Hebrew Old Testament |
| Koine Greek | Greek New Testament, Septuagint, Greek Fathers |
| Latin | Vulgate, Latin Fathers |
| Ge'ez | Ethiopian Orthodox canon (including 1 Enoch and Jubilees) |
Titles name the corpus, not the language code: the Septuagint and the
Greek New Testament share the code grc, and Old Testament books exist in both
Hebrew and Septuagint Greek — a distinction that matters to readers.
Where the voices come from
All four voices are speech-synthesis models we trained ourselves —
no commercial text-to-speech product covers these languages. The training recordings come
from four readers, all licensed in writing:
Jesse Orloff (Greek); LibriVox "bedwere" (Latin, Public Domain Mark 1.0);
ንባብ ቤት / eathebook.org (Ge'ez); Rabbi Dan Be'eri (Hebrew).
Pronunciation follows scholarly reconstructions — Classical/early Koine for Greek,
Ecclesiastical for Latin, the Tiberian tradition for Hebrew, and Ethiopian Orthodox
chanting practice for Ge'ez. Specialists still differ on the details; our choices are
not the only defensible ones.
Where the text comes from
- Original languages: Public-domain texts. We cannot identify the specific critical edition behind each corpus — the texts come from public digital collections that do not record their upstream edition. Observable characteristics: the Latin matches the Clementine Vulgate (1 John 5:7 carries the Comma Johanneum; orthography ejus/majores); the Greek New Testament matches Westcott–Hort 1881 (John 1:25 reads Ἡλείας). For scholarly citation use a printed critical edition, not this site.
- Chinese: Chinese Union Version with New Punctuation (public domain)
- English: World English Bible (public domain)
- Imagery: public-domain artworks matched to each verse; where no suitable
painting exists, a calligraphy background we produce ourselves, styled after a manuscript page
More than video: a full toolkit for reading the originals
Video is produced one chapter at a time, but the text, the verse alignment and
the lexicon apply to the entire corpus — 623 books, 11,724 chapters, 748,966 verses. Three things are open to all of it, whether or not a video exists yet:
- Read the text — the original of any chapter with Chinese and
English alongside, and a gloss on every word: lemma, morphology, meaning.
Where a reading exists, the page links to it.
- Chapter audio (rolling out since August 2026; 5,621 chapters
so far) — hear the original read aloud on the reading page itself.
Click any verse to start from that verse; the verse being read is marked
and each word turns yellow as it is spoken, so you can follow along even
if you have never studied Hebrew or Greek. The speed is adjustable, and the whole chapter
can be downloaded as an mp3.
Same synthesised voices as the videos — but where a video takes a chapter at a time,
the reading-page audio can cover the whole corpus.
- Full-text search — in the original, in Chinese or in English.
Greek and Hebrew queries need no accents or vowel points, and any word can be
searched by lemma, listing every inflected occurrence (λόγῳ, λόγον …).
- Parallel columns — the same chapter in Hebrew, Septuagint Greek, the Vulgate and
Ge'ez side by side (e.g. Jonah 1). Where the versification differs,
cells are left blank rather than forced into line.
Lexicon coverage, measured in August 2026: Greek New Testament 99.7%,
Septuagint 99.9%, Latin Vulgate 98.6%, Hebrew Old Testament 97.9%,
Apocrypha 94.9%, Greek Fathers 92.1%, Latin Fathers 92.7%,
Ge'ez Bible 89.0%. Words that are not found say so; nothing is guessed from a
near match. Latin morphology comes from William Whitaker's WORDS (public domain);
for Greek, where this site’s own form tables have no entry, we fall back to
Perseus/CLTK morphological data (CC BY-SA 3.0). That layer is measurably less reliable
(about 93.5% in held-out testing, against 99.5% for the layers above it), so
the gloss panel says when an entry comes from it.
Textbooks: learning to read the originals from scratch
Reading aloud and glossing words help someone who can already work through the
original. Starting from nothing calls for a course, so this site also writes
textbooks. There are two. A Latin course on the Vulgate, in an English
and a Chinese edition of 40 chapters each, with a lecture video per chapter (80 in all),
written notes, and a quiz (710 questions across the two editions); and a
Gəʿəz course on the Bible, 45 chapters in each language (90 videos, 804 questions
across the two editions), running from the shapes and syllables of the Fidäl script through
to syntax and sustained reading.
It is not a course in Classical Latin. It is written to get you reading
the Vulgate, and its choices differ accordingly:
- Ecclesiastical pronunciation, matching the Latin readings on this site
- Every example sentence is taken from the Vulgate itself (478 per edition), each
traceable back to chapter and verse
- Vocabulary chosen by frequency in the Vulgate (1,188 entries per edition, none
repeated across chapters) rather than by frequency in Classical prose
- The syntax biblical Latin actually uses: indirect statement with
quia / quoniam plus the indicative rather than the Classical
accusative-and-infinitive; parataxis with et in place of deeply nested
subordination; and the Hebraisms that come with it, such as et factum est
and in aeternum
The course is written by Peng-Chan Huang. The explanations, exercises and
vocabulary are original throughout — no text or figure is taken from any
current textbook, and the examples come from the public-domain Vulgate. So the course is
free to share and to show publicly, on the same terms as the readings, and
its videos are being published to this site's YouTube channel.
Courses in Hebrew and Greek are planned on the same pattern.
Neither has been started, and there is no timetable — this is stated to
show the direction, not to promise a date.
Frequently asked questions
- What is Verbum?
- A free, ad-free site with no accounts, where the Bible and early Christian literature
are read aloud in the languages they were written in — and where the full text of
those originals can be read, searched and parsed word by word. Four languages: Biblical
Hebrew, Koine Greek, Latin and Ge'ez; 623 books, 11,724 chapters, 748,966 verses.
- Are the voices human or synthetic?
- Synthetic. All four are speech models we trained ourselves — no commercial
text-to-speech product covers these languages. The training recordings come from four
readers, all licensed in writing. Synthetic speech can misplace stress or mispronounce
rare words.
- Which pronunciation is used?
- Scholarly reconstructions: Classical / early Koine for Greek, Ecclesiastical for
Latin, the Tiberian tradition for Hebrew, and Ethiopian Orthodox practice for Ge'ez.
Specialists differ on the details; our choices are not the only defensible ones.
- Which translations are the subtitles?
- Chinese: Chinese Union Version with New Punctuation. English:
World English Bible. Both are public domain. For books outside the 66-book canon
(such as 1 Enoch) no public-domain Chinese translation exists, so the Chinese is machine
translation by this site, not reviewed by a professional translator; the English is
R. H. Charles (1917).
- How much of the text does the lexicon actually cover?
- Measured in August 2026: Greek New Testament 99.7%, Septuagint 99.9%,
Latin Vulgate 98.6%, Hebrew Old Testament 97.9%, Apocrypha 94.9%, Greek Fathers 92.1%,
Latin Fathers 92.7%, Ge'ez Bible 89.0%. Words that are not found say so —
nothing is guessed from a near match. For Greek, where our own tables have no entry we
fall back to Perseus/CLTK morphological data; that layer is less reliable and the gloss
panel says so.
- I have never studied these languages. Where do I start?
- With a textbook. There are two: a Latin course on the Vulgate, 40 chapters
in English and 40 in Chinese, and a Gəʿəz course on the Bible, 45 chapters in each language.
Both run from the script and pronunciation through to reading continuous passages; each chapter
has a lecture video, written notes and a quiz. Every example sentence comes from the biblical
text in that language, so everything you learn is a sentence you will actually meet.
Courses in Hebrew and Greek are planned on the same pattern, but neither
has been started and there is no timetable.
- May I use this in a church, a classroom, or my own work?
- Yes. Free to watch, share and play publicly. Credit Verbum if you redistribute the
video files; ask first about commercial resale. See the Terms.
- Do you track me?
- No. No tracking cookies, no third-party advertising, and
no raw IP addresses are stored — only a two-letter country code and anonymous
view counts. Site traffic is measured with Cloudflare Web Analytics, which uses
no cookies and no cross-site identifier and cannot identify individual visitors.
See the Privacy Policy.
Sources and acknowledgements
This site is built on these open resources:
How you may use it
Verbum is free, ad-free and requires no account. It sets no tracking
cookies and stores no raw IP addresses. You are welcome to use it for personal study,
teaching, sermon preparation and research, and to play it publicly in churches, classrooms
and study groups. See the Terms and Privacy Policy.
Where to find it
- This site — video, verse navigation and click-to-parse lexicon
- Read the text and search — all 623 books, not just
those with video
- YouTube channel —
with trilingual subtitles burned in
- Podcast / RSS: feed.xml (audio only)
Contact
This is a personal project. Corrections, requests for particular books, and licensing
enquiries are all welcome at
u97001062@gmail.com.