Questions we get about Portuguese transcription.
Short answers, most with a link to the page that goes further.
Which Portuguese we work in
Do you work with European or African Portuguese?
Our team is Brazilian, and Brazilian Portuguese is what we transcribe and translate day to day. Recordings from Portugal, Angola, Mozambique, or other Portuguese-speaking countries we take on by arrangement: send a short excerpt first and we’ll tell you whether our reviewers can do it well. If one study mixes speakers from Brazil and Portugal, mention it in the request, because the two spell some words differently.
Can you handle regional accents and slang?
Portuguese sounds very different in Recife, Porto Alegre, and the interior of São Paulo, and speech recognition tends to do worst with the accents it heard least in training. That’s why the draft goes to a native speaker, who corrects what the software got wrong. Slang stays as spoken in the Portuguese; in the English it gets a translator’s note when a plain translation would lose it. For heavily accented recordings, a pilot on a short excerpt shows what to expect.
How a transcript gets made
Is the transcript made by AI or by people?
Both, in that order. Whisper, run locally through whisper.cpp, makes the first draft, and pyannote separates the speakers, on our workstations in Brazil.1 A native speaker then goes through the draft and listens again wherever it reads wrong, correcting speaker labels, names, acronyms, and misheard words. The draft never reaches you. The sample transcript shows the kind of fixes a reviewer makes.
Should I choose verbatim or clean verbatim?
Verbatim keeps hesitations, repetitions, and false starts, as in “eu, eu passo na UBS.” Clean verbatim drops them and doesn’t correct how people speak. For thematic analysis, clean verbatim is usually enough; for discourse or conversation analysis, choose verbatim. You can switch between the two on the sample transcript and compare the same passage.
How are speakers labeled in a focus group?
The moderator is M and the respondents are R1, R2, and so on, with a timestamp at each change of speaker. If your screener or codebook uses other codes or first names, the labels follow it. When two respondents talk at once, each keeps a separate line with the same timestamp and a [crosstalk] mark. Speaker-separation software often merges those lines, and pulling them apart is part of the review. The focus group sample shows one at 00:41:16.
Which file formats do you deliver?
DOCX for reading and quoting, TXT for importing into NVivo, ATLAS.ti, or MAXQDA, and SRT caption files timed to the recording if you want them. Bilingual transcripts come as a two-column DOCX, Portuguese on the left and English on the right. Each delivery includes a short report listing the files processed, the conventions used, and who reviewed them.
Translation and translator’s notes
Can we get the translation side by side, or English only?
Either. The two-column version aligns each turn of the Portuguese with its English, sharing one timestamp and speaker label, so a Portuguese-speaking colleague can check the translation line by line. The English-only version keeps the same labels and timestamps. Spelling is American unless you ask for British. On the sample transcript you can switch between Portuguese, side by side, and English.
What is a translator’s note?
A short note beside the English when a word carries meaning the translation can’t hold on its own. In our sample, a community health worker says some families need her to estar em cima; the English reads “keep a close eye on,” and the note explains the colloquial sense. Notes cover language and context, health-system acronyms such as UBS included. They don’t interpret what a participant meant.
Can you translate our discussion guide or screener into Portuguese?
Yes. Discussion guides, screeners, and stimulus materials are translated from English into Brazilian Portuguese before fieldwork, by our in-house translators. For academic studies, the Portuguese interview guide is also what Brazil’s research ethics committees (the CEP) review. We translate consent forms and participant information sheets for the CEP as well. Send the documents with your fieldwork dates.
Data protection and ethics committees
Where does our audio go?
First to a private upload link set up for your project, then to our workstations in Brazil, where the first draft of the transcript is made.1 For review, translation, and delivery, we name the tools before the project starts, so your protocol can list them. The data handling page draws the whole path.
Will you sign our NDA and data processing agreement?
Yes, after reviewing them with you. An NDA is in place before we receive any file, and for projects under the GDPR, a data processing agreement in which we act as your processor. As a Brazilian company, Agência Rivera is subject to the LGPD, Brazil’s general data protection law. If your institution has its own templates, send them with the request; the quote form has a box to tick for each.
Can you work with health information covered by HIPAA?
Only with specific approval. We don’t accept protected health information (PHI) under HIPAA unless the project has been approved for it in writing and a Business Associate Agreement (BAA) is signed before any file is sent. If your recordings may contain PHI, tell us when you ask for a quote.
Can EU research data be sent to Brazil?
Yes. Commission Implementing Decision (EU) 2026/179 finds that Brazil ensures an adequate level of protection for personal data sent from the EU, including health data, to organizations in Brazil that are subject to the LGPD, Brazil’s general data protection law, as we are. That means no standard contractual clauses for the transfer when the files go to storage in Brazil or the EEA (we name the upload service and its location in writing before you send anything), and the same applies from Norway, Iceland, and Liechtenstein under the EEA Agreement. The finding is about Brazil, not the countries beyond it: a further transfer to a service hosted outside Brazil and the EEA needs its own safeguard under Brazil’s transfer rules, and we name the tools for your project in writing before it starts. The GDPR still requires a data processing agreement under Article 28, which we put in place with you. The UK and Switzerland haven’t recognized Brazil as adequate, so for projects from either country, tell us which transfer mechanism your institution requires.
Can you de-identify transcripts?
Yes, on request. Names, places, and other identifiers are replaced with codes you define, such as [RESIDENT 1] or [STREET 1], and the codes are the same in the Portuguese and the English. We call this de-identification because a coded transcript can still count as personal data under the GDPR, and “anonymization” would promise more. Try the De-identified setting on the sample transcript.
Ethics committees and interview transcription for academic studies
When are files deleted?
When your protocol says. Let us know the retention period before the project starts; when it ends, we delete the recordings and transcripts and confirm the deletion by email. If your protocol doesn’t set a period, we agree one in writing at the start.
Files, schedules, and quotes
How long does it take?
Turnaround is agreed per project, because it depends on how many hours of audio there are, how many voices, how clear the recordings are, and whether you need English too. Send us your deadline, or the date of your debrief or analysis kickoff, and we reply with a quote and a timeline worked back from it. On a first project, a pilot on a short excerpt also shows how the work goes on your kind of audio.
How do you price a project?
Per project, once we know the recordings. The quote depends on total audio length, voices per recording, how clear the audio is, whether translation is needed, and the deadline. There’s no price list on this site because two 60-minute files can take very different amounts of work: a one-on-one interview in a quiet room is not a focus group of eight in a café. The quote form asks for exactly these details.
Can we test you first?
Yes. We can run a pilot on a short excerpt of one of your recordings, in the conventions you’d use for the study. Pick a passage that’s typical of your recordings, and include a hard one if you have it: people talking over each other, a noisy room, a strong accent. You see the transcript and translation before you commit the budget.
How do we send recordings?
Through a private upload link we set up for your project. If a field partner in Brazil holds the files, they can upload them directly. Please don’t send recordings by email or attach them to the contact form: the form doesn’t take audio, and an inbox isn’t where participant data should travel.
What time zone are you in?
Santa Fé do Sul is on UTC−3, Brasília time. That puts us one or two hours ahead of New York, depending on the time of year, and three or four hours behind London. The clock at the bottom of this page shows our local time.
Who is behind Rivera Brazil?
Rivera Brazil is the international service of Agência Rivera de Consultoria em Comunicação, Tradução, Edição e Publicidade Ltda., CNPJ 42.968.216/0001-95, in Santa Fé do Sul, São Paulo state. For five years the company has edited and translated for Brazilian scientific journals and their authors. It is a member of ABEC, the Brazilian Association of Scientific Editors, and holds an active editorial contract with Cadernos de Saúde Pública (Fiocruz). The work is done by in-house editors and translators, with postgraduate consultants in specific fields.
A question this page doesn’t answer? Write to hello@riverabrazil.com, or send us the details of your project.