Latin has two main pronunciation systems: classical (restored) for classics courses, exams and verse, and ecclesiastical (Church) for liturgy, choirs and most medieval texts. Use this guide to choose the right one, hear the differences sound by sound, and check any Latin audio, from a text-to-speech app, a recording or a tutor, against the system you need.
Key Takeaways
- Classical and ecclesiastical Latin differ in about 8 sounds, most audibly C, G, V and AE. Neither is wrong; they serve different purposes.
- Most AI Latin voices you will find sound closer to ecclesiastical, because they are built on a Romance-language (often Italian-like) sound base.
- Use the 5 test lines below on any Latin audio source. You will hear which system it follows within seconds.
- For verse, use a text with macrons and write elisions into the input. No engine scans meter for you.
- Kveeky does not offer a Latin voice. It can voice the English explanations, vocabulary notes and instructions in your Latin lessons.
Which Latin pronunciation do you need?
Start from what the audio is for. The decision rarely needs more than one question.
- Classics class, exam, Virgil or CiceroClassical (restored)
- Mass, liturgy or choir rehearsalEcclesiastical (Church)
- Medieval or Renaissance textsEcclesiastical, usually
- Not sure what you needRun the 5 test lines below
Classical vs ecclesiastical: sound by sound
These are the differences you will actually hear. Use the example words as quick checks.
| Spelling | Classical (restored) | Ecclesiastical (Church) | Example word |
|---|---|---|---|
| C before e, i, ae, oe | Always hard k | Soft ch as in “church” | Cicero: KIK-er-oh vs CHEE-cheh-roh |
| G before e, i | Always hard g | Soft j as in “gem” | gens: GENS vs JENS |
| V | w | v | veni: WAY-nee vs VEH-nee |
| AE | eye | eh | caelum: KYE-lum vs CHEH-lum |
| OE | oy | eh | poena: POY-na vs PEH-na |
| SC before e, i | sk | sh | descendit: des-KEN-dit vs deh-SHEN-dit |
| TI before a vowel | tee | tsee | natio: NAH-tee-oh vs NAH-tsee-oh |
| GN | g + n | ny as in “canyon” | agnus: AG-nus vs AH-nyus |
| Vowel length | Long and short vowels differ, and it matters for meter | Length is mostly ignored | mālum (apple) vs malum (evil) |
Where each is used. Classical pronunciation is the standard in most university classics departments and English-speaking school Latin. Ecclesiastical pronunciation is used in Catholic liturgy and in most Latin choral music, from plainchant to Renaissance polyphony.
Test which pronunciation any Latin audio uses
Play or generate these lines with the Latin audio source you plan to use: a text-to-speech app, a recording or a tutor. Listen for the sound in each note. If two or more come out “ecclesiastical”, that source follows the Church system.
Test 1 · Hard C or CH? Listen to the first sound of each name
Test 2 · W or V? Classical sounds like "way-nee, wee-dee, wee-kee"
Test 3 · GN and liturgical text Ecclesiastical says "AH-nyus"
Test 4 · Hard G Classical keeps G hard in "Gallia"; listen to "gens" too
Test 5 · Does it read macrons? Long vowels should sound held longer, and nothing should be skipped
Teaching Latin? Kveeky can voice the English explanations, instructions and vocabulary notes in your lessons. Free, with no credit card.
Voice your lesson notes freeMacrons, vowel length and meter
Written Latin usually leaves out macrons, the bars that mark long vowels. For prose this rarely matters. For verse it changes everything, because classical meter is built on long and short syllables, not on stress.
| Your text | What to do | Why |
|---|---|---|
| Prose (Caesar, Cicero’s letters, medieval Latin) | Generate as written | Vowel length barely affects understanding |
| Verse you want to sound metrical (Virgil, Ovid, Horace) | Use a macronised edition | Gives the voice a chance to lengthen the right vowels |
| Voice ignores or skips macron letters | Respell long vowels, for example canoo for canō, then listen again | Forces audible length when macrons are not read |
| Teaching vowel length | Generate pairs such as mālum / malum in one file | Students hear the contrast side by side |
Elision: the fix no engine does for you
In verse, a word ending in a vowel (or vowel plus m) before a word starting with a vowel or h is elided: the first vowel effectively disappears. Text-to-speech does not scan meter, so it reads every vowel.
As written The voice reads every vowel
Edited for elision Paste this version to hear it scan
How to check Latin audio before you use it
- Pick the system using the decision guide above, and note it at the top of your text.
- Run the 5 test lines on the audio source so you know which system it follows.
- If a word sounds wrong in a text-to-speech app, change your text and generate again. You can't edit the sound itself, so fix the spelling instead: for example, write Kikero instead of Cicero if you want a hard K. For poems, type the long-vowel marks (ā, ē, ī, ō, ū) and remove vowels that are skipped when the line is read aloud (see "Elision" above).
- Keep files short. One sentence or one line of verse per file is easier to replay, check and correct.
- Check with a trained reader before students copy the audio for exams or recitation.
How Kveeky helps Latin teachers and students
For pause and emphasis control in lesson audio, see the pause, emphasis and pacing guide. For course narration, see e-learning voiceover.
Accuracy: what AI Latin audio is good for
| Use | Is AI Latin audio enough? |
|---|---|
| Vocabulary drills and prose reading | Yes, once you have confirmed the pronunciation system |
| Choir rehearsal tracks (ecclesiastical) | Yes as a practice aid; check against your director’s preferences |
| Metrical verse for exams | Only with macrons and edited elisions, then checked by a teacher |
| Public recitation or recordings students will imitate | Use it as a draft, and check it with a trained reader first |
| Regional medieval traditions (German, French, English) | No engine we know of reproduces these |
Frequently asked questions
Does Kveeky have a Latin voice?
No. Kveeky does not offer a Latin voice. You can use it to voice the English parts of your Latin lessons, and use this guide to check Latin audio from any other source.
Which Latin pronunciation should I use?
Use the one your context expects. Classical suits classics courses, exams and classical verse; ecclesiastical suits liturgy and choral music. If your syllabus or choir specifies one, follow it.
What is the difference between classical and medieval Latin pronunciation?
Medieval Latin is usually read with ecclesiastical, Italian-style sounds: soft C and G, V as in “very”, and AE as “eh”. Classical pronunciation keeps C and G hard, says V as W, and treats vowel length as meaningful.
Why does my Latin text to speech sound Italian?
Most Latin voices are built on a Romance-language sound base, which produces ecclesiastical-style pronunciation. Run the test lines above to confirm which system you are hearing.
Can text to speech read Latin verse with correct meter?
Not on its own. Meter depends on vowel length and elision, which engines do not scan. Use a macronised text and edit elisions into the input to get much closer.
Is AI Latin audio accurate enough for teaching?
For vocabulary, prose reading and general familiarity, generally yes, once you have checked the pronunciation system. For verse and recitation that students will copy, check it against a trained reader first.
Sources and how we checked
- Classical sounds follow W. Sidney Allen, Vox Latina: A Guide to the Pronunciation of Classical Latin (Cambridge University Press, 2nd edition, 1978).
- Ecclesiastical sounds follow the Italianate Church pronunciation used in Catholic liturgy and choral practice.
- No audio source was ranked or endorsed in this guide; the test lines work with any of them.