Which languages does Muvo cover?

«Languages» means three different things in Muvo, and each has its own list. Your team may manage the space in French, a visitor may see the player buttons in Japanese, and the commentary they hear may be in Basque. Knowing which is which avoids most surprises.

Three kinds of language

The language you work in

The dashboard, the public website and Muvo's emails are available in Spanish, English, French, Italian and Portuguese (Brazil). Each person on your team chooses their own with the language selector in the user menu, and it does not affect visitors at all. The Help Center and the legal texts are in English only.

The player interface

The player is what visitors open when they scan a QR code. Its interface, meaning the play button, the speed, the tour stops, the language notice and similar texts, is translated into 12 languages: Spanish, Catalan, Basque, Galician, English, French, German, Italian, Portuguese, Chinese (simplified), Japanese and Dutch. If the phone is set to any other language, the interface appears in English.

The interface is not your commentary. You can publish an audio guide in a language the player interface does not have: a visitor with a Korean phone sees the buttons in English and hears your Korean commentary. See Which language visitors hear.

Languages for your audio guides

Your audio guides can be translated and voiced in more than 100 languages and regional variants. Most major languages come with their accents as separate choices, such as Spanish from Spain, Mexico or Argentina, British, American or Australian English, Portuguese from Portugal or Brazil, or Arabic from Saudi Arabia, Egypt or Morocco. Catalan, Valencian, Basque and Galician are there too.

Almost every variant has its own native AI voices, so a Mexican Spanish text is read with a Mexican accent. Valencian comes in two variants, Valencian (Valencian pronunciation) and Valencian (Catalan pronunciation): both are written in Valencian and visitors see «Valencià» either way.

Your space chooses which ones to use. Administrators add and activate languages in Settings → Languages, and your plan sets how many can be active at once.

High-resource and low-resource languages

In artificial intelligence, languages are often described as high-resource or low-resource. It is a technical term about training data, not about how important a language is or how many people speak it.

A speech model learns to talk by listening to a huge amount of recorded speech together with its text. From those hours it learns how each word is pronounced, where the stress falls, how a sentence rises and falls, and the rhythm of the language. Languages with fewer hours of public recordings and text give the model less to learn from, so its pronunciation, prosody and rhythm are less polished. This says nothing about the language itself, and quality improves as models are trained with more data.

Today, Basque and Valencian (Valencian pronunciation) are low-resource languages in Muvo. They are marked Few resources in Settings → Languages, and when you add one, the panel warns you that its AI voice is somewhat more limited, reads each paragraph on its own without carrying the intonation over from the previous one, and may make more pronunciation mistakes. In Settings → Voice and music they have their own tab, Low-resource languages, with their own voice.

The kind of mistakes to listen for

Much of the recorded speech that AI voices learn Basque and Valencian from comes from radio and television where presenters and guests switch between those languages and Spanish. The voice picks up that mix, so its usual slip is to sound Spanish: some words, sounds or stresses come out as they would be said in Spanish rather than in Basque or Valencian.

A personal note. Muvo is a Spanish company born in Alicante, in the Valencian Community, so Valencian is our language too. We work every day so that languages like Valencian and Basque sound as good as any other, because promoting culture and the languages that carry it is in our blood.

Low-resource languages are fully supported for translation, publishing and the player. To get the best result from their AI voice:

• Review the text carefully. Short, clear sentences are easier to read aloud well.
• Listen to every paragraph before you mark the language Ready to publish, and generate again any paragraph that sounds wrong.
• Record your own voice for the paragraphs, or the whole text, where the AI voice does not convince you. Recording is free: see Record your own voice.