AI voice catalog
Find your voice, then match it to the model and language your application needs.
Definition
The voice catalog is the library of voices available to your synthesis requests. It contains 900+ voices, with model and language compatibility recorded per voice. You select one by its voice ID in each request. For anything the catalog does not cover, a cloned voice slots in by ID the same way.
Breadth
A large catalog means you match voice to use case instead of forcing one voice everywhere. A warm narrator for an audiobook, a crisp voice for IVR, a distinct set for a game cast, selected from the same library. On Simba 3.5 every voice in the catalog speaks all 48 languages under one model id, and a voice keeps the accent of the recording it was made from. The older Simba 3.2 supports English and Simba 3.0 supports 6 languages across 7 locales.
Consistency
Reuse a voice ID to keep the selected speaker identity across chapters, episodes, and course modules. Keep your model and delivery settings consistent, and evaluate the output: synthesis is not a promise of byte-identical audio or zero variation.
Beyond the catalog
When you need a voice that is not in the library, clone one and reference its ID exactly like a catalog voice. See the voice cloning page and the model and language guide.
Frequently asked questions
How many voices are in the catalog?
How do I choose a voice?
Are voices consistent across requests?
Continue exploring speech.
Start building
Find your voice, then match it to the model and language your application needs.