Portuguese is spoken by more than 250 million people across Brazil, Portugal, and several other countries, yet it is missing from some of the biggest built-in text-to-speech language lists. ReadLoud reads Portuguese through more than one engine. Here is how to use it, what to expect honestly, and what a real listen costs.
How to listen to Portuguese text in ReadLoud
Getting Portuguese text into ReadLoud works the same as any other language: paste it directly, paste a link, or upload a file. PDF, EPUB, DOCX, HTML, Markdown, CSV, and plain text are all supported, up to 40 MB per file and 900,000 characters per document, enough for a single article or a full book.
ReadLoud detects the document's language automatically and offers voices in the matching language, so a Portuguese document is not read by a voice guessing its way through English pronunciation rules. A link to a Brazilian or Portuguese news site is extracted like a reader view, dropping navigation, ads, and comment sections, leaving the article itself. A PDF, common for Portuguese-language academic papers and reports, loses headers, footers, page numbers, and citation brackets in extraction, and anything skipped is shown dimmed in the text rather than disappearing without a trace.
What to expect from the voices
Portuguese, like Spanish, is covered by more than one engine on ReadLoud, and it is worth knowing what each is actually good at. The Fast, Studio, and Gemini 3.1 engines, built on Gemini models, are multilingual and produce solid Portuguese output, since the underlying models were trained broadly. That said, they are strongest in English overall, and the difference shows up in small ways at the edges, an occasional pacing or intonation choice that reads slightly less naturally than the same engine's English output.
The Narrator engine, built on Google's Chirp 3 HD voices, offers dedicated Brazilian and European Portuguese locales specifically, and tends to sound closer to a native speaker for sustained narration, since it was built with that regional variation in mind. The honest recommendation is to try Narrator first if natural-sounding Portuguese matters most to you, especially for something long like a book or a lecture transcript. If you want a Gemini engine's delivery styles, narrator, newsreader, calm, teacher, podcast host, storyteller, briefing, bright, or quiet, those apply regardless of language and are worth trying on Portuguese text too, just with the expectation that they are a notch behind their own English output rather than behind Narrator specifically.
Brazilian or European Portuguese
Brazilian and European Portuguese differ enough in accent, some vocabulary, and pronunciation patterns that a mismatched voice is noticeable, similar to the difference between American and British English. Narrator's separate Brazilian and European locales let you pick the one that matches your document or your own ear. A Brazilian news article or novel will generally sound more natural in the Brazilian locale, and a European Portuguese document, or one written in the Portuguese used across parts of Africa that follows the European standard, tends to fit better with the European locale. If you are unsure which suits you, preview both, since this is genuinely a matter of familiarity and taste.
Where the built-in options fall short
This is a clear and somewhat surprising gap. Kindle's text-to-speech covers only five languages: English, Spanish, French, Italian, and German. Portuguese, despite having more native speakers worldwide than either Italian or German, is not on that list, so a Portuguese ebook on a Kindle device with a screen has no built-in voice option, and e-ink Kindles have no speaker regardless.
ReadLoud's scope here is specific: paste or upload Portuguese text, choose between Brazilian and European Narrator locales or a Gemini engine, and preview before committing to something long.
Fixing a name or term that keeps coming out wrong
Portuguese documents often carry proper nouns, anglicisms, and technical terms, an author's name, a brand, a scientific term borrowed from English and used unchanged in a Portuguese paper, that a general-purpose voice can mispronounce the first time it meets them. Once it gets one wrong, it tends to repeat that same mistake every time the word shows up again.
ReadLoud's pronunciation dictionary fixes this once: correct the term, and the fix applies going forward across every document you listen to afterward, not just the one open when you noticed. For an academic paper series or a book with recurring character or place names, fixing a handful of terms early saves hearing the same mispronunciation over and over later.
Cost example
Credits on ReadLoud are minutes, not a flat plan, and the conversion holds regardless of language: as a rough guide, 1,500 words of text is about 10 minutes of listening at 1x speed, in any voice.
A Portuguese news article or blog post of a few hundred words runs one to two minutes. A longer piece, a book chapter, a research paper, closer to 1,500 words, lands around 10 minutes. On the Taste pack ($9 for 60 minutes, $0.15 a minute), that is roughly $1.50. On the Regular pack ($36 for 300 minutes, $0.12 a minute), closer to $1.20. A longer academic paper or several book chapters might run 20 to 40 minutes, a few dollars on the Regular pack. Anything already generated is cached and free to replay, so coming back to the same chapter later costs nothing the second time.
Try it before deciding
The best way to know which locale and engine actually suit your ear is to try them on your own Portuguese text. Try one minute free with no account and compare a Brazilian voice, a European voice, and a Gemini engine on the same document before deciding. A free account adds 20 minutes total, enough to get a real feel for all three.