TTS locale vs language field: getting an en-GB accent

Cartesia separates language (a base code like en) from locale (a regional code like en-GB) and ties accents to each voice. Sume's TTS request exposes a language field only, so it has no locale or accent field: pick a voice that already speaks the accent you want.

Cartesia facts are from its Multilingual voices guide; Sume facts are from the OpenAPI document behind the API reference. Both were read 2026-09-30.

What does Cartesia mean by locale?

Per the guide, most requests need only language or locale: the voice speaks that language in one of the accents it supports. locale needs sonic-3.6 or newer; on earlier models you use language with a base code like en. A request may set one or the other, never both.

Each voice reports an accents list, and each accent belongs to a locale. If you ask for a locale the voice has no accent for, the voice keeps its original accent. The guide's example is a voice whose only English accent is General American: a request for en-GB falls back to American English rather than failing.

What does Sume accept instead?

The schema describes language as the language the voice speaks the transcript in (BCP-47 / ISO-639, for example ko, ja, en), 2 to 16 characters long. Omitting it defaults to English at the provider. The request schema does not allow extra properties, so locale and accent are not part of the contract.

Can I send en-GB as the language?

The length limit allows it, since en-GB is five characters, and the description names BCP-47. The docs do not say what Sume does with a regional subtag, though, so do not assume it selects a British accent. Send en, listen, and if the accent is wrong, change the voice.

const response = await fetch("https://api.sume.com/v1/tts-1.0/generate", {
  method: "POST",
  headers: {
    Authorization: "Bearer " + process.env.SUME_API_KEY,
    "Content-Type": "application/json",
    "Idempotency-Key": "tts-en-001",
  },
  body: JSON.stringify({
    transcript: "Thanks for holding. Your account is up to date.",
    voice: { id: process.env.SUME_VOICE_ID },
    language: "en",
  }),
});
console.log(await response.json());

Can I still use sonic-3.6 on Sume?

Yes, as an explicit id. TTS 1.0 has no engine picker, but POST /v1/tts-router/generate requires a model, and its catalog enum includes sonic-3.6, sonic-3.5 and sonic-3. That choice sets the model, not a locale. The Sonic 3.6 languages post covers the language side in detail.