Skip to main content
Autocalls assistants speak 100+ languages, on every plan, at the same per-minute rate as English. There is no charge for a language and nothing to request: pick it from the language dropdown, or, for a language you do not see there, tell the assistant which language to speak in the prompt (see below). Hear samples on the languages page.

​
Supported languages

Afrikaans, Albanian, Amharic, Arabic, Armenian, Assamese, Asturian, Azerbaijani, Bashkir, Basque, Belarusian, Bengali, Bosnian, Breton, Bulgarian, Burmese, Castilian, Catalan, Chinese (Mandarin), Croatian, Czech, Danish, Dutch, English, Estonian, Faroese, Filipino (Tagalog), Finnish, Flemish, French, Galician, Georgian, German, Greek, Gujarati, Haitian Creole, Hausa, Hawaiian, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Javanese, Kannada, Kazakh, Khmer, Korean, Lao, Latin, Latvian, Lingala, Lithuanian, Luxembourgish, Macedonian, Malagasy, Malay, Malayalam, Maltese, Maori, Marathi, Moldovan (Romanian), Mongolian, Nepali, Norwegian (Bokmål), Norwegian (Nynorsk), Occitan, Pashto, Persian, Polish, Portuguese, Punjabi, Romanian, Russian, Sanskrit, Serbian, Shona, Sindhi, Sinhala, Slovak, Slovenian, Somali, Spanish, Sundanese, Swahili, Swedish, Tajik, Tamil, Tatar, Telugu, Thai, Tibetan, Turkish, Turkmen, Ukrainian, Urdu, Uzbek, Valencian, Vietnamese, Welsh, Yiddish, Yoruba. English and Spanish come with several accents (US, UK, Australian, Latin American, Spain). Within a language you pick the accent through the voice, and you can tune speed and pitch. Which voices exist per provider (ElevenLabs, Cartesia, Inworld, Azure) is in Voice selection.

​
Setting the language

  1. Open the assistant’s Voice & speech settings.
  2. Pick the language in the Language dropdown. In Pipeline and Dualplex™ modes this selects the transcription and voice pairing for that language.
  3. Pick a voice for that language, run a test call, and read the transcript.

​
A language you do not see in the dropdown

The dropdown drives the transcription and voice pairing for Pipeline and Dualplex™ modes. Speech-to-speech mode does not need it, because the multimodal model listens and speaks natively in any language it is asked to use. So for a language that is not in the dropdown:
1

Set the assistant mode to Speech-to-speech

In the assistant’s settings, choose Speech-to-speech (see Assistant modes).
2

Leave the language on English

Speech-to-speech ignores the dropdown.
3

Tell the prompt which language to speak

Add a line such as “Always speak Nepali, regardless of the language the caller greets you in.”
Speech-to-speech uses a smaller voice set than Pipeline, and voice cloning runs on ElevenLabs and Cartesia, so a cloned voice is limited to the languages those providers support.

​
One assistant, several languages in the same call

Turn on Secondary languages in the assistant’s Voice & speech settings (see General settings). The assistant then detects the caller’s language and switches mid-call, for example a Spanish speaker calling a US clinic’s English receptionist. Two notes:
  • Filler audio turns off when secondary languages are on, because the fillers are recorded in one language.
  • Inworld voices do not support secondary languages; pick ElevenLabs or Cartesia for multilingual assistants.
If most of your callers speak one language, set that language as the primary one instead of relying on detection. A dedicated language per call transcribes more accurately than auto-detection, especially on short answers like names and postcodes.

​
Testing a language

Run a test call in the target language before going live and read the transcript. If names or numbers are mis-transcribed, slow the speech rate slightly or add spelling instructions to the prompt. Before you sign up, the free live demo call lets you hear the agent in English on your own phone.

​
Frequently asked questions

Yes. All three are in the language dropdown with built-in voices and transcription.
100+ languages, listed above, on every plan from Starter at $34 a month. Every language is billed at the same per-minute rate as English.
No. Every language is included, and any language you do not see in the dropdown works through Speech-to-speech mode with a line in the prompt.
Yes. The full list and Speech-to-speech mode are available in every white label platform.