What to expect from spoken Latin
The speech model has heard very little spoken Latin. Expect a rough draft with many errors. It works best on slow, clear, well-articulated reading by a single speaker. Fast conversational Latin, choirs and chant, or reading over background music will often produce text that needs rewriting.
If the recording is of a known text, such as a passage from Caesar, Cicero, Virgil or the Vulgate, the most practical approach is to use the draft for its timings and to correct the words against the printed text.
Upload the recording as it is: MP3, M4A, WAV, OGG, OPUS, FLAC, a voice message or a phone recording. You can also record from your microphone in the upload box. Nothing needs converting first.
Pronunciation systems and spelling
Latin is spoken with several pronunciations, and that changes what the model hears:
- Restored classical pronunciation (hard c and g, w-like v) is common in schools and universities.
- Ecclesiastical or Italianate pronunciation (soft c before e and i, as in Italian) is used in church Latin; the model may write some words with Italian-like spelling.
- National school pronunciations, such as traditional English or German ones, sound different again.
- Macrons marking long vowels are not written in the output, and u/v and i/j are not standardised.
Mixed Latin and modern languages
Many recordings mix Latin with a modern language: a lecture in English that quotes Latin, a Mass where readings and homily are in the local language, a class where the teacher explains in Spanish. Choose the language by what dominates. A lecture mostly in English with Latin quotations should be transcribed with English, and the quotations fixed by hand. A reading entirely in Latin should stay on Latin.
Auto-detect often mistakes Latin for Italian or Spanish, so set it yourself, as this page does.
The transcript opens next to a player that highlights the current line. Click a line to hear it again, correct the word in place, and your changes save automatically.
Who records Latin
Typical files are spoken-Latin courses and podcasts, recitations of poetry for teaching metre, recorded Latin Masses and readings, conference talks given in Latin, and students practising for oral exams or competitions. Teachers use the timed transcript to build listening exercises, and learners use the CSV export to line up text and time.
Export TXT, DOCX, PDF, CSV, SRT or VTT, with timestamps and speaker names switched on or off per export. Speaker labels are available on paid plans or with a Minute Pack.
For video files, see the Latin video to text page.