Generate the AI voice of your texts
Turn your paragraphs into natural speech with the AI voice, see the cost before confirming and regenerate only the paragraphs that changed.
- For:
- Admins, Editors
- Updated
- Updated October 20, 2026
- 3 min read
- 3 min read
Muvo reads your text aloud with a natural AI voice, one paragraph at a time. Each paragraph has its own audio file, so the voice sounds as good at the end of a long text as at the start, and editing one paragraph only regenerates that paragraph.
Generate one paragraph
Open the artwork from Artworks and choose the language in the language row.
Open the paragraph menu with its … button, or right-click the paragraph.
Choose Generate audio.
Check the Cost and the Balance afterwards, and press Confirm · 1 credit.
The audio is generated in the background. You can keep writing while each paragraph shows its progress, and then press its play button to listen.
Generate a whole text
To voice every paragraph that still needs audio, press Generate what is pending in the editor. The number in brackets is how many paragraphs will be generated. The confirmation also tells you how many are reused for free and how many recorded paragraphs are left as they are.
To do it for several artworks at once, select them in Artworks and choose Generate pending audio. The total cost is shown before you confirm.
One paragraph costs 1 credit, whether it is the first time or a regeneration. If your space already has an audio with exactly the same text, language and voice, it is reused at no cost. See How credits work.
Audio status of a paragraph
Each paragraph shows the state of its audio, and the bar at the bottom of the editor counts how many paragraphs have audio, for example 8 of 10 paragraphs with audio.
| Status | What it means | What to do |
|---|---|---|
| No audio | The paragraph has never had audio. | Generate it or record it. |
| Queued | The request is waiting its turn. | Nothing: it starts on its own. |
| Generating | The voice is being created. | Keep working; it usually takes seconds. |
| Ready | The audio matches the current text. | Listen to check it. |
| Out of date | The text, the language or the voice changed after the audio was made. | Generate it again. |
| Error | The audio could not be generated. Nothing was charged. | Press Try again. |
When you save, only the paragraphs whose text actually changed become Out of date. Moving, deleting or reordering paragraphs costs nothing. Merging two paragraphs leaves the result without audio, so generate it again.
A recorded paragraph is never replaced by the AI voice on its own. Generate what is pending leaves recordings alone; if a recorded paragraph's text changes, record it again. To swap a recording for the AI voice on purpose, use Replace with the AI voice in its menu.
Choose the voice
The voice is a setting of the whole space, so every artwork sounds consistent. Only administrators can change it.
Go to Settings → Voice and music.
Under Narration voice, listen to the samples and pick the Space voice.
If you want a different voice for one language, set it under Voice per language.
Save. If some paragraphs already have audio with the previous voice, Muvo tells you how many and offers Regenerate with its cost. Choose Not now to keep them as they are.
Each language in Muvo comes with pronunciation and accent guidance for the voice, such as Valencian with a Valencia accent or Spanish from Spain. Abbreviations, numbers and acronyms are read naturally.
Publish to update what visitors hear
New audio does not reach visitors until you publish. When you press Publish or Publish changes, Muvo joins the audio of all the paragraphs into one continuous track, with even volume and natural pauses. While it is being prepared you see Publishing… and visitors keep hearing the previous version.
If the track cannot be prepared, the editor shows The audio track could not be prepared. Press Try again: it costs no credits.
To avoid misuse, a paragraph can be generated up to 20 times a day. If you reach the limit, you can generate it again tomorrow or record it yourself.