Forvo

Text-to-speech has gotten frighteningly good. Modern synthetic voices can read a paragraph with convincing rhythm and near-human intonation. So why does Forvo Pronunciation, an app built on the decidedly low-tech premise of just recording real people saying words, still matter?

Because a synthesizer can tell you what a word is supposed to sound like. Forvo tells you what it sounds like when a woman in Naples, a man in Buenos Aires, and a teenager in Osaka actually say it. That distinction is the entire pitch, and it’s a good one.

Developed by Forvo Media and released way back in June 2010, the app is the mobile front end for Forvo.com, which has quietly become one of the world’s largest crowdsourced pronunciation dictionaries. The scale does a lot of the talking here: a vast archive of recorded pronunciations spanning languages from the widely spoken to the obscure. This isn’t an algorithm approximating speech from phonetic rules. It’s a vast, human-contributed archive — the audio equivalent of Wikipedia, built one recording at a time by native speakers who wanted to help a stranger say Woerthersee or nauseous correctly.

Market positioning is refreshingly honest. Forvo doesn’t pretend to be a full language-learning suite. It won’t teach grammar, run anyone through spaced-repetition flashcards, or gamify streaks like Duolingo. It does exactly one thing — pronunciation — and focuses on it to the exclusion of everything else. That focus makes it a companion tool rather than a destination app, the utility to reach for mid-conversation, mid-translation, or mid-panic before having to read a foreign name aloud.

The audience is specific and loyal: language learners chasing an authentic accent, travelers who don’t want to butcher a place name, translators and teachers who need to be right, and the merely curious. For all of them, the value proposition is the same — reliable, real-world pronunciation from real human throats. The app is the messenger. The database is the miracle.

• Crowdsourced Native Audio: Every entry is a recording from a real native speaker — no synthetic voices. This is the whole game, and little else on mobile matches its authenticity.
• Broad Language Coverage: Coverage runs from Spanish and Mandarin down to languages most apps have never heard of. If a word exists and someone speaks it, there’s a decent chance Forvo has it.
• Accent & Voice Comparison: Hear the same word across regional accents within one language, and toggle between male and female voices. Genuinely useful for learners targeting a specific dialect.
• Request & Contribute: Missing a word? Request it and native speakers may record it. Speak a language yourself? Contribute your own pronunciations and follow contributors whose voices are trusted.
• Redesigned Interface with Dark Mode: A cleaner UI, a proper dark mode, and localization in English, Spanish, and French.

The Core Loop

Forvo’s utility loop is almost aggressively simple: search a word, get a list of recordings, tap to play, listen, repeat. That’s it. And for the common use case — “how do I say this?” — it’s exactly enough. Type “gnocchi,” tap the little speaker icon, and an Italian voice sets things straight before an embarrassing dinner moment. The design puts the payoff one tap away.

Where the app earns its keep is in the depth beneath that simple loop. Pull up a common word in a widely-spoken language and the result isn’t one recording — it’s a dozen, each tagged by speaker and often by region. This is the feature power users live for. Learning Portuguese but specifically the Brazilian variety, not European? Filter for it. Want to know whether a target word carries a hard or soft consonant in different parts of Spain? The recordings are right there, stacked, ready to be A/B tested against each other. No synthetic engine can replicate this, because the variation is the data.

Where It Gets Human

The contribution mechanics turn Forvo into something closer to a living community. The request system means the database isn’t static — gaps can be flagged, and the crowd fills them in. Following contributors whose voices are preferred adds a small but thoughtful layer of personalization; for anyone mimicking a specific accent, being able to return to a trusted voice matters more than it sounds.

But here’s where the skepticism kicks in. Crowdsourcing is a double-edged sword. The same open contribution model that gives Forvo its breadth also means quality is uneven. Popular words in major languages have many recordings and effective peer review. Obscure words in less-common languages might have one recording of questionable audio quality, or none at all. Forvo is honest about this in structure — the number of pronunciations is visible — but a newcomer expecting uniform polish will occasionally hit a wall.

The Friction

The app’s biggest weakness isn’t what it does — it’s how it feels doing it. The mobile experience comes across as clunky, with a persistent sense that the app is a wrapper around a website rather than a natively-conceived mobile experience. Navigation works, but it doesn’t delight. Some features and the API sit behind a paywall, which is fair — someone has to keep the servers running — but an ad layer in the free tier is built to interrupt what should be a frictionless mid-conversation lookup.

The redesign has helped. The interface is more intuitive than it was, and dark mode is a welcome, overdue addition. But “more intuitive than before” is a low bar when the underlying resource is this valuable. The database deserves a better vehicle than it currently has. The rough edges are tolerable because what’s underneath is irreplaceable — and that’s a slightly uncomfortable place for any product to live.

The redesigned UI is clean, functional, and unremarkable — which, for a utility, is mostly fine. Dark mode is well-implemented and a genuine relief for anyone reaching for pronunciation help late at night or in a dim café. Typography is legible, the search field is front-and-center, and playback controls are large enough to hit without precision.

The design prioritizes the thing that matters most — fast access to audio — and the layout reflects that. Where the cracks show is in the connective tissue: the interface is built for quick lookups rather than extended browsing, and the ad layer can feel intrusive on the free tier. This isn’t a beautiful app. It’s a serviceable one bolted onto an extraordinary database, and the aesthetic ambition tops out at “gets the job done.”

On both iOS and Android, the experience is functionally identical, which is a polite way of saying neither platform’s version feels lovingly tailored to it. The touch-first interaction — tap to search, tap to play — suits the format naturally; this was always a phone-shaped task, and the big playback targets work well with thumbs.

Both platforms inherit the same faintly web-wrapped feel. Neither version leans into platform-specific niceties — no standout widget story, no deep share-sheet magic. It works on each, but it doesn’t feel at home on either. For a tool often summoned one-handed in a hurry, it’s competent. It’s just not native-grade craftsmanship.

Forvo is a case study in the gap between a resource and a product. The resource — a crowdsourced archive of human pronunciations across an enormous range of languages — is one of the internet’s quiet treasures, the kind of thing that gets more valuable every year as more people contribute. Nothing synthetic comes close to the authenticity of hearing a word spoken the way people actually speak it.

The product is merely fine. The app is functional, occasionally awkward, sometimes ad-cluttered, and never quite as elegant as what it delivers. But — and this is the point — it earns its place anyway, because the alternative is guessing. When the question is how a word really sounds, Forvo is the answer, warts and all. Rough vehicle, priceless cargo — and a recommendation for anyone who asks that question often enough to forgive the clunkiness.