Choose the voice of each language

Every content language of your space is read by its own AI voice. Muvo uses native voices, so the accent comes with the voice: a Mexican voice for Mexican Spanish, a British one for British English. You only choose once per language, and every artwork in that language uses it.

Only administrators can change voices. Editors generate audio with whatever voice is chosen.

How voices are organised

Go to Settings → Voice and music. The Narration voice card splits your active languages into two tabs:

Each tab lists which of your languages belong to it. Only the languages active in your space appear; to add more, see Add or remove languages. A few languages have no native voices yet: they show a notice and are read with a multilingual voice instead.

Pick a voice

1. 1

   Go to Settings → Voice and music.

2. 2

   In Narration voice, open the tab of the language you want to change.

3. 3

   Under Voice for your language, press the play button on a voice card to hear its sample. Each card shows whether it is a female or male voice and, when known, its age.

4. 4

   Select the voice you prefer.

5. 5

   Repeat for any other language and press Save.

Choosing a voice costs nothing. Credits are only spent when audio is generated with it.

Hear how it sounds

The first card on the screen, Hear how it sounds, plays your choices together the way visitors will hear them. Press Listen with voice: your background music starts alone, then the sample of each language's voice plays on top, one after another with the main language first, and the music fades out at the end. While it plays, it shows which language and voice is speaking. Press Stop to end it.

It follows what you have selected below, even before you save, so you can try several voices and compare. It needs a background music track to play; see Add background music.

What happens to existing audio

Every AI audio remembers the voice it was made with. When you change the voice of a language and save, the paragraphs of that language voiced with the previous voice become Out of date. Other languages, recordings and uploaded audio are not affected.

After saving, Muvo tells you how many paragraphs have audio with the previous voice and what regenerating them costs. You can:

• Press Regenerate · N credits to see the confirmation and generate them all again with the new voice.
• Press Not now and regenerate later, from each artwork with Generate pending and outdated (N) in the arrow next to Generate what is pending (N).

Low-resource languages

In artificial intelligence, «low-resource» means there is less text and audio available to train voices in that language. It says nothing about how important the language is. These voices are somewhat more limited: they read each paragraph on its own, do not support intonations and may make more pronunciation mistakes. Listen to the result before publishing, or record your own voice for the paragraphs that matter most.