Portuguese audio to text
Drop a Portuguese recording in and get a transcript in Portuguese, produced in your browser — nothing uploaded, nothing metered.
Both Brazilian and European Portuguese are handled, with Brazilian noticeably stronger because of how the training data is distributed.
Language and translation need the Multilingual model.
Drop an audio or video file
or , paste with Ctrl + V, or
- MP3
- MP4
- WAV
- M4A
- MOV
- WEBM
- OGG
- FLAC
- AAC
The file is read by this page and never uploaded — no account, no size limit but your own RAM.
Preset: multilingual model, language set to Portuguese.
Download
Click a line to jump the audio there. Click the text to fix a word — your edits go into every download.
How it works
- Open this page — the tool is already set up for “Portuguese to text”.
- Drop the file in, browse for it, or paste it with Ctrl+V.
- The model runs on your own device; the transcript appears as it goes.
- Correct anything misheard, then download text, subtitles or notes.
Brazilian and European Portuguese
The two varieties differ enough in rhythm and vowel reduction to matter for recognition. Brazilian Portuguese, which dominates the available training audio, transcribes noticeably better. European Portuguese — with its heavier vowel reduction and faster connected speech — is workable but produces more errors, particularly on unstressed syllables.
The model writes what it hears rather than converting between varieties, so Brazilian spelling conventions appear for Brazilian speech and European ones for European speech. It does not enforce either orthographic standard.
Nasal vowels and running speech
Nasal vowels and the diacritics that mark them are produced correctly in ordinary cases. Where accuracy suffers is fast connected speech with heavy elision, which is common in informal recordings and is the same weakness every language exhibits under the same conditions.
Recording distance matters more than any of this. A close microphone in a quiet room produces a strong transcript in either variety; a phone across a room produces a weak one.
English output
Setting the output to translate produces English text directly from Portuguese speech in one pass. As with every language here, translation is only into English.
It is a working gloss — enough to understand a recording and find the parts that matter, not a publishable translation.
Things that save a re-run
- European Portuguese is harder than Brazilian; expect more corrections on unstressed syllables.
- Leave detection on unless the clip is short or switches language part-way.
- Proper nouns repeat — fix once, then search for the rest.
Read the transcript against the audio before you rely on it. Every speech model — this one and the paid cloud ones — mishears names, numbers and crosstalk, and it does so confidently. The player and editor here exist so that check takes minutes rather than an afternoon.
Frequently asked questions
Does it handle both Brazilian and European Portuguese?
Yes, though not equally. Brazilian Portuguese transcribes better because it dominates the training data. European Portuguese works but produces more errors on fast, vowel-reduced speech.
Which spelling convention does it use?
It writes what it hears rather than converting between varieties, so the output follows the speech. It does not enforce a single orthographic standard.
Can I get English out instead?
Yes — set the output to translate. The model produces English text directly from Portuguese speech; it does not translate in the other direction.
Is the audio sent anywhere?
No. The model runs inside your browser and the recording never leaves your device.