Supported languages
JustTranscribe detects the language of a recording by itself and transcribes 59 languages with word-level timestamps. There is nothing to choose before uploading: a Spanish voice note comes back in Spanish with its accent, an English lecture in English, a Brazilian podcast in Portuguese. For 16 languages the transcript also arrives with speaker labels at no extra cost.
The two lists below mirror what the pipeline does today. The first group runs on the speech engine that returns who spoke when; the second group is transcribed with the same timestamps but speaker labels are added on request. Any finished transcript can then be translated into 111 languages.
At a glance
- Languages transcribed
- 59, detected automatically
- Speaker labels included
- 16 languages (Spanish, English, Portuguese, French, Italian, Dutch, Russian, Turkish, Indonesian, Filipino, Hindi, Korean, Greek, Urdu, Bengali, Catalan)
- Spanish variants
- Rioplatense, Mexican, Colombian, Andean, Caribbean, Chilean, peninsular — same engine, no setting
- Portuguese variants
- Brazilian (all regions) and European
- Translation
- Any finished transcript into 111 languages
- Per file
- Up to 150 minutes and 500 MB; free account, no card
Transcribed with speaker labels included · 16
Timestamps per word, paragraphs, and a label for each voice ("Speaker 1", "Speaker 2" — rename them in one click). Files over two minutes are routed here once the language is detected; a short clip is sent here first and kept when the engine confirms the language.
- Bangla · bn
- Catalan · ca
- Dutch · nl
- English · en
- Filipino · tl
- French · fr
- Greek · el
- Hindi · hi
- Indonesian · id
- Italian · it
- Korean · ko
- Portuguese · pt
- Russian · ru
- Spanish · es
- Turkish · tr
- Urdu · ur
Transcribed (speaker labels on request) · 43
Same word-level timestamps and paragraphs. Speaker labels are one click away in the transcript view, computed on demand.
- Afrikaans · af
- Arabic · ar
- Armenian · hy
- Azerbaijani · az
- Basque · eu
- Belarusian · be
- Bosnian · bs
- Bulgarian · bg
- Chinese · zh
- Croatian · hr
- Czech · cs
- Danish · da
- Estonian · et
- Finnish · fi
- Galician · gl
- German · de
- Hebrew · he
- Hungarian · hu
- Icelandic · is
- Japanese · ja
- Kannada · kn
- Kazakh · kk
- Latvian · lv
- Lithuanian · lt
- Macedonian · mk
- Malay · ms
- Māori · mi
- Marathi · mr
- Nepali · ne
- Norwegian · no
- Persian · fa
- Polish · pl
- Romanian · ro
- Serbian · sr
- Slovak · sk
- Slovenian · sl
- Swahili · sw
- Swedish · sv
- Tamil · ta
- Thai · th
- Ukrainian · uk
- Vietnamese · vi
- Welsh · cy
Spanish and Portuguese from any country
- Spanish: the engine was measured on rioplatense, Mexican, Colombian, Peruvian, Chilean and peninsular recordings — real samples are published on the Spanish tool pages, errors included. No country setting exists because none is needed.
- Portuguese: Brazilian Portuguese is among the languages it transcribes best; European Portuguese is detected and transcribed the same way. The article "um" is kept (an earlier engine dropped it as a filler).
- Galician, Basque and Catalan: all three are detected. Galician and Catalan audio is checked against Spanish so a Spanish recording is never written as Galician; Basque is the weakest of the three and a page with real samples says exactly where it fails.
- Mixed recordings: the language is decided once per file from a sample in the middle, so a Spanish class with English quotes is transcribed as Spanish throughout. Recordings that switch language for long stretches are better uploaded in parts.
What people actually transcribe here
In the 60 days to 26 October 2026, people transcribed recordings in more than 60 languages on JustTranscribe. Spanish was 61% of completed transcripts, English 18%, Portuguese 14%; French, Hindi, Arabic, Filipino, Chinese and Russian followed.
Limits that apply to every language
- 150 minutes and 500 MB per file or link; longer recordings are uploaded in parts.
- Three new transcripts per hour per account, processed one at a time (the beta's abuse brake — write to us if you hit it doing real work).
- Audio: MP3, WAV, M4A, AAC, OGG/OPUS, FLAC. Video: MP4, MOV, WEBM, MKV. Public links from YouTube, TikTok, Instagram, Facebook, Pinterest and Google Drive.
- Automatic transcription: clear single-speaker audio comes back close to final; noisy rooms, overlapping voices and rare proper nouns need a read-through. Every page with a real sample names its errors.
Questions about languages
Do I have to select the language before uploading?
No. The language is detected from the recording itself and the transcript comes back in that language. You can translate it afterwards into any of the 111 target languages.
Which languages get speaker labels automatically?
Spanish, English, Portuguese, French, Italian, Dutch, Russian, Turkish, Indonesian, Filipino, Hindi, Korean, Greek, Urdu, Bengali and Catalan. For every other language the Speakers chip in the transcript view computes the labels on request.
Does it understand my country's Spanish?
Yes — rioplatense, Mexican, Colombian, Peruvian, Chilean, Caribbean and peninsular Spanish are transcribed by the same engine with no setting. Recording quality matters far more than accent: a voice close to the microphone comes back almost final.
Does it work with Brazilian Portuguese?
Yes, and well: Brazilian Portuguese is one of the languages it transcribes best. European Portuguese is detected and transcribed the same way.
What about a recording in two languages?
One language is chosen per file from a sample taken in the middle of the recording, so short quotes in another language are transcribed as heard in the main language. For long stretches in a second language, split the file.
Can I get the transcript in a different language than the audio?
Yes: transcribe first, then translate the finished transcript into any of 111 languages with one click. Both versions can be exported, including as subtitles.