Voices
Find voices by model, provider, language, or descriptive search.
Use the public voices API to find a voice and confirm which models and providers offer it. No API key is required.
Search voices
/v1/voicesNativeSearch in plain language:
curl "https://api.allmodels.io/v1/voices?q=british+woman"Or list voices for a model:
curl "https://api.allmodels.io/v1/voices?model=elevenlabs%2Feleven-multilingual-v2&page_size=20"Each result contains:
model: the AllModels model ID and name.voice: the voice ID, name, description, preview, languages, and labels.providers: the providers that can serve that voice with the model.
Use total_count for the number of matches. When has_more is true, pass next_cursor unchanged in the next request:
curl "https://api.allmodels.io/v1/voices?model=elevenlabs%2Feleven-multilingual-v2&cursor=NEXT_CURSOR"Filter and sort
| Parameter | Use |
|---|---|
q | Search names and descriptive metadata. |
model | Return voices available for a model ID. |
provider | Return voices available from a provider. |
language | Filter by language ID, such as en. |
category | Filter by category, such as official or community. |
label | Filter by facet:value, such as gender:female. |
first_seen_after | Find voices added after an RFC 3339 timestamp. |
sort | Use featured, relevance, newest, popular, or name. |
page_size | Return 1–100 results; the default is 20. |
cursor | Continue from a previous response's next_cursor. |
include_facets | Set to true to include available filters and counts. |
Repeat model, provider, language, or category to match any supplied value. For labels, repeated values in the same facet match either value; different facets must all match.
Without a search query, results use featured sorting, which places official voices ahead of community voices. With q, results default to relevance.
If pagination returns 409 catalogue_changed, restart without a cursor to use the latest catalogue.
Use a voice
Pass voices[].voice.id as the voice and voices[].model.id as the model in your text-to-speech request. If the result lists multiple providers, you can let AllModels choose or use provider selection.
See Text to speech for complete request examples, or the voices API reference for the full response schema.
Using the ElevenLabs SDK? See ElevenLabs voices.