Multilingual Voices

Every voice can speak every language the model supports. A single-accent voice sounds like a non-native speaker outside its own language; a multilingual voice has accents in several languages and sounds native in each, so one voice can serve customers in every language you support.

See which accents a voice supports

A voice sounds native only in the accents it supports. In any other language, it falls back to its original accent. Check which accents a voice supports in the dashboard or the API:

In the voice library, hover over the flag in a voice’s language column to list its supported accents, and click any accent to hear it.

In the API, call Get Voice and read the accents field:

// GET /voices/9626c31c-bec5-4cca-baa8-f8ba9e84c8bc
{
  "id": "9626c31c-bec5-4cca-baa8-f8ba9e84c8bc",
  "name": "Jacqueline",
  "accents": [
    { "accent": "general-american", "locale": "en-US", "is_native": true },
    { "accent": "standard-hindi", "locale": "hi-IN", "is_native": false }
    // ...and every other accent this voice supports
  ]
  // ...plus description, gender, created_at, and the rest of the voice
}

In this example, Jacqueline supports two accents, General American English and Hindi, and sounds equally native in both. Each accent belongs to a locale. is_native: true, like the flag shown in the voice library, marks the accent the voice was originally recorded in. It’s the accent the voice falls back to when you request a language outside its list.

Using a multilingual voice

Take Jacqueline from above and use that voice ID in a text-to-speech request. Most requests only need language or locale: the voice speaks that language in one of the accents it supports, and Cartesia normalizes the transcript for that language or locale so dates, numbers, and units are read correctly.

locale needs sonic-3.6 or newer. On earlier models, use language with a base code like en instead. A request may set one or the other, never both.

Speak a language natively

locale sets the language the voice speaks, how the transcript is read, and which of the voice’s supported accents it uses. In the example below, we set locale to en-US, and Jacqueline speaks American English natively, like a local.

{
  "model_id": "sonic-3.6",
  "voice": "9626c31c-bec5-4cca-baa8-f8ba9e84c8bc",
  "transcript": "Thanks for holding. I've pulled up your account, and the last three payments went through without any trouble. The one from Thursday is still pending, so it should clear by tomorrow morning.",
  "locale": "en-US"
}

Accent fallbacks

When you request a language this voice has no accent for, the voice falls back to its original accent: General American English in the case of Jacqueline. Jacqueline doesn’t support a Japanese accent, so listen to the audio and notice how she speaks Japanese with an American accent:

{
  "model_id": "sonic-3.6",
  "voice": "9626c31c-bec5-4cca-baa8-f8ba9e84c8bc",
  "transcript": "お待たせいたしました。お客様のアカウントを確認しましたところ、直近三件のお支払いはすべて問題なく完了しております。",
  "locale": "ja-JP"
}

The same fallback applies at the locale level. Jacqueline’s only English accent is General American (en-US), so a request for en-GB falls back to her American English rather than failing. The audio below is identical to the en-US request above:

{
  "model_id": "sonic-3.6",
  "voice": "9626c31c-bec5-4cca-baa8-f8ba9e84c8bc",
  "transcript": "Thanks for holding. I've pulled up your account, and the last three payments went through without any trouble. The one from Thursday is still pending, so it should clear by tomorrow morning.",
  "locale": "en-GB"
}

Add a deliberate accent

Setting accent only applies to a multilingual voice, and you only need it when you intentionally want the voice to sound accented, the way a non-native speaker would. Here the transcript is the same English as above, with accent set to standard-hindi. Listen to the audio and notice how Jacqueline speaks English with a Hindi accent:

{
  "model_id": "sonic-3.6",
  "voice": "9626c31c-bec5-4cca-baa8-f8ba9e84c8bc",
  "transcript": "Thanks for holding. I've pulled up your account, and the last three payments went through without any trouble. The one from Thursday is still pending, so it should clear by tomorrow morning.",
  "locale": "en-US",
  "accent": "standard-hindi"
}

sonic-3.6 makes voices sound more native when they speak another language, so a deliberate accent may sound lighter. For a deliberately non-native accent, use sonic-3.5.

accent must be one of the voice’s supported accents. Jacqueline has no japanese accent, so this request fails:

{
  "model_id": "sonic-3.6",
  "voice": "9626c31c-bec5-4cca-baa8-f8ba9e84c8bc",
  "transcript": "お待たせいたしました。お客様のアカウントを確認しましたところ、直近三件のお支払いはすべて問題なく完了しております。",
  "locale": "ja-JP",
  "accent": "japanese"
}

// HTTP 400
{
  "error_code": null,
  "message": "this voice does not support accent 'japanese'",
  "title": "Unsupported accent for voice",
  "request_id": "927f6ea3-22cf-426a-8767-446f41677e17"
}

Switching languages mid-conversation

Send one language per request. For example, send “For English, press one” and “हिंदी के लिए, दो दबाएँ” as two separate requests. Mixing languages inside a single generation works where it’s common, such as Hindi (Hinglish) and Tagalog (Taglish). Outside those cases the speech may sound accented.

Making your voice multilingual

You can give a voice native accents beyond the one it was recorded in. Both ways make the voice sound native in the new accent. Which one you use depends on the voice:

Method	Applies to	Result
Add Voice Accents	Private Instant Voice Clones you own	Same voice ID, now speaking more accents natively
Localize Voice	Any voice, including public Cartesia voices and Pro Voice Clones	A new voice ID that speaks only the new accent

An added accent makes the voice sound native when it speaks that language. How closely it matches the original speaker varies by voice and accent, so preview the accent before you use it in production.

Add accents to your Instant Voice Clone

Cartesia dashboard: find the voice in the voice library, open the More actions menu next to the voice, then select Manage voice > Supported languages & accents > Add.
API: call Add Voice Accents with the accent ids.

A voice can support up to 10 accents. To free a slot, remove one with Delete Voice Accent.

Adding accents requires the voice’s original accent on record. If Get Voice returns no accents field, set it first with Update Voice. The dashboard asks you for it automatically. To skip this step, set accent when you clone a voice.

Localize any other voice

Cartesia dashboard: use Localize.
API: call Localize Voice with the voice ID and an accent id.

Use the new voice ID in your requests.

Localizing a Pro Voice Clone creates an Instant Voice Clone, not a new PVC. It still sounds like the speaker, but it won’t match the PVC’s tone, timbre, and pacing as closely. To keep that fidelity in another language, train a separate PVC for each language.

Common questions

What do a voice's supported languages and accents mean?

A voice can speak every language the model supports. Its supported accents are the ones it speaks natively, like a local. In any other language, it speaks with its original accent.

What is a voice's original accent?

The accent the voice was recorded in. It’s the accent the voice falls back to in any language it has no supported accent for. Two ways to tell which one it is:

Voice library: the flag shown in the voice’s language column.
API: the item with is_native: true in the accents array of the Get Voice response.

Can a voice speak a language outside its supported list?

Yes. Set locale to the language you want on the voice you already use, and listen. The voice keeps its identity and speaks with its original accent.If the accent isn’t native enough, add the accent with Add Voice Accents, or Localize Voice for voices you don’t own, so it sounds like a local.

Will a multilingual voice sound like the same person in every language?

Mostly. How closely it matches the original speaker varies by language and by voice. Listen to every language you plan to use before production. If an added accent makes the voice stop sounding like the original speaker, remove it with Delete Voice Accent. The voice then speaks that language in its original accent.

When should I set locale vs. accent in a text-to-speech request?

Most requests only need locale. It sets the language, how the transcript is read, and the voice’s matching accent. Set accent only when you want a multilingual voice to sound deliberately accented.

How do I generate deliberately accented speech?

Voice with one accent: set locale to the transcript’s language. The voice’s accent carries through.
Multilingual voice: set locale to the transcript’s language and accent to one of the voice’s supported accents. See Add a deliberate accent.

For the strongest accent, use sonic-3.5. Sonic 3.6 makes voices sound more native in other languages.

Why does my voice sound less accented on Sonic 3.6?

Sonic 3.6 makes voices sound more native when they speak another language. Two cases where you’ll hear it:

No supported accent for the language: the voice falls back to its original accent, which carries over more lightly than on Sonic 3.5.
A deliberate accent: the accent you set comes through subtler.

If keeping a deliberate accent matters for your use case, use sonic-3.5.

Sonic

Choose a voice

Control the voice

Streaming

Audio output

Older models

See which accents a voice supports

Using a multilingual voice

Speak a language natively

Accent fallbacks

Add a deliberate accent

Switching languages mid-conversation

Making your voice multilingual

Add accents to your Instant Voice Clone

Localize any other voice

Common questions

​See which accents a voice supports

​Using a multilingual voice

​Speak a language natively

​Accent fallbacks

​Add a deliberate accent

​Switching languages mid-conversation

​Making your voice multilingual

​Add accents to your Instant Voice Clone

​Localize any other voice

​Common questions

See which accents a voice supports

Using a multilingual voice

Speak a language natively

Accent fallbacks

Add a deliberate accent

Switching languages mid-conversation

Making your voice multilingual

Add accents to your Instant Voice Clone

Localize any other voice

Common questions