Choose and manage voices

The Voices area has separate My voices and Built-in tabs. Your own voice cards show their cloning status and available samples; built-in voices can be browsed by characteristics and used without creating a personal clone.

Use a voice in a new project

1. From a ready voice card, use Create project to open a project with that voice preselected, or start from Projects and select the Voice field.

2. In the project editor, choose My voices or Built-in voices.

3. For a cloned voice with multiple samples, open the voice to choose a specific language. For a built-in voice, select the desired language row directly.

4. The selected sample sets the project's language. Confirm that your script is written in that language before generating.

Browse built-in voices

• Use the Built-in tab when you do not want to create or use a personal voice clone.

• Filter by gender, style, and language, then reset filters when you need a wider list.

• Use the play control when a preview is available. Preview a voice before selecting it, especially when choosing between similar styles.

Understand readiness and language samples

• Pending and cloning indicate that a personal voice is not ready to use yet.

• Ready indicates that its output samples are available. A multilingual voice can show progress as language samples become ready.

• Failed indicates that the clone did not complete. Review the original audio requirements and create a better source sample if necessary.

Maintain your voice library

• Use the voice actions menu to edit a display name, copy the voice ID, create a project, or delete the voice.

• Use names that identify the permitted speaker and intended use, such as “Product host – English”, rather than names that could be confused with an unapproved person.

• Deleting a voice is permanent. Check which projects rely on it before removing it; completed projects remain their own recorded results, but future selections will no longer include the deleted voice.

Language selection matters

A project's voice language and script should agree. Selecting a different language sample is the intended way to prepare multilingual output; changing only the script without choosing a matching sample can produce a result that is not what you expect. For any important publication, generate a short test project first.