Voice catalogs

Both catalog routes require voices:read.

Preset voices

curl --fail-with-body \
  'https://app.verbatik.com/api/v1/voices/catalog?model=verbatik%2Fspeech-lite&language=en&gender=female' \
  -H "Authorization: Bearer $VERBATIK_API_KEY"

Optional query parameters are model, language, gender, and search. The default model is Speech Lite. A language filter such as en matches English locales; an exact locale such as en-US narrows the result.

The response contains object: "list", model, and data. Voice entries include id and label, with additional metadata such as language, gender, styles, and preview information where the catalog provides it.

Use an ID from the catalog for the same model as your generation. Voice labels are display names, not substitutes for IDs.

Custom voices

curl --fail-with-body https://app.verbatik.com/api/v1/voices \
  -H "Authorization: Bearer $VERBATIK_API_KEY"

This lists workspace custom voices, not preset catalog voices. Retrieve an individual voice with GET /voices/{id}. Its resource ID can be used with the Speech Cloning model.

Create voices using cloning or voice design. Deleting a custom voice requires voices:write; update dependent integrations before deleting a voice they use.