# Keep a cloned voice when it speaks another language

> Keep a cloned voice when it speaks another language: 0 companies fully do it, 11 with a limit, 55 carry the tag. Every listing in Voice cloning & custom voices with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/voice-cloning.clone-speaks-other-language/
- Markdown: https://www.anchorterminal.com/niches/voice-cloning.clone-speaks-other-language/index.md (~1,550 tokens)
- Slim: https://www.anchorterminal.com/niches/voice-cloning.clone-speaks-other-language/index.min.md (~430 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/voice-cloning.clone-speaks-other-language/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 0 | 11 | 55 | 9 | 4 |

## What counts as done

Docs list which languages a clone can speak and whether the source language must be declared, the clone keeps the speaker's identity in the target language without importing a foreign accent, words shared between scripts are read in the target language, and a language or accent parameter exists where the model needs one.

How agents ask for it: make my voice speak Spanish; dub this video in the speaker's own voice; clone from Japanese, speak Chinese; keep my accent out of the French version

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **Cartesia Voice Cloning API + MCP (Cartesia AI, Inc.)**: Each clone is built from one declared language; other languages need the separate Add Voice Accents call, and a PVC needs a dataset per language.
- **ElevenLabs Voice Cloning and Voice Design API (Eleven Labs Inc.)**: A clone can speak any supported language; on Eleven v4 same-language speech keeps the accent and other languages are fluent in the target language, so the speaker's accent does not carry over. Samples in unsupported languages are rejected.
- **Fish Audio Voice Cloning API (Hanabi AI Inc.)**: The model carries a languages field, but acc-14 found characters shared between Chinese and Japanese read in the wrong language with no language parameter on TTS (closed as a feature request).
- **Gradium Voice Cloning and Voice Design (Gradium SAS)**: The source language must be declared, and Voice Localization makes new candidates in another language or accent from an existing voice; only five languages are supported.
- **Inworld Instant Voice Cloning (Theai, Inc. dba Inworld AI)**: The clone declares its source language (Other for Tier 2 languages); which languages a clone can speak is on a separate multilingual page that was not read.
- **MiniMax Voice Clone and Voice Design (Nanonoble Pte. Ltd.)**: language_boost (or auto) helps recognise the sample's language and dialect; nothing states which languages a clone keeps its identity in or how accent carries.
- **Resemble AI Voice Cloning API (Resemble AI, Inc.)**: A voice takes a default language code at creation; which other languages it can speak was not documented in the pages read.
- **Smallest.ai Instant Voice Clone (Smallest, Inc.)**: The clone declares a language checked against the model pool (400 Invalid language); Lightning v3.1 Pro covers English, Spanish and Hindi in the deprecation table, and cross-language identity is not described.
- **Soniox Voice Cloning (Soniox Inc.)**: Docs say a clone speaks the 60+ supported languages and keeps a consistent identity when switching, with language set per request; accent carry-over and shared-script handling are not described, and no account tests it.
- **Speechify API Voice Cloning (Speechify, Inc.)**: On simba-3.5 a clone speaks all 48 languages under one model id, but the docs say it speaks them with the accent of its recording, and on simba-3.2 a clone is English only.
- **VoiceStudio (VoiceStudio maintainers)**: generate_speech takes a language per call (Auto overrides the profile language) and the README claims 646 languages across engines; whether a clone keeps the speaker's identity across languages depends on the engine and is not documented.

## Couldn't check

The docs were closed to us, or silent. These carry no mark either way.

- **Murf Voice Cloning API (Murf Inc.)**: Murf's terms bar automated access to murf.ai, so its API docs were not read this run, as in the 5 November check; no first-hand account names Murf.
- **Ultravox Voice Cloning (Theai, Inc. dba Inworld AI)**: The cloning page says included voices cover all supported languages but says nothing about which languages a clone speaks.

- not this job: Hume Octave Voice Design and Cloning + MCP (Hume AI, Inc.): Hume's docs index now lists only Audio Decisions and Expression Measurement, and the voice overview and changelog redirect there; the TTS and EVI APIs, cloning and design included, are no longer documented after the 13 November 2026 end of access.

## What stops agents finishing

- a clone is built from one language and other languages need a separate accents step (Cartesia)
- only five languages supported (Gradium)
- clone languages tied to a pricier model (Speechify Simba 3.5 at twice the rate from 21 November 2026)

## Around it

Before: translation.translate-text, speech-to-text.subtitles-timestamps. After: video-generation.speech-and-sound, text-to-speech.timed-to-video.

## Where this comes from

Weight 9 of 100 in Voice cloning & custom voices: 4 of 34 accounts: an Indian accent on non-native clones, a Spanish reference whose accent was lost, Japanese-cloned voices reading Chinese characters in Japanese, and a dubbing app built on consistent translated voices

As data: https://www.anchorterminal.com/api/v1/use-cases/voice-cloning.clone-speaks-other-language.json, or the MCP tool get_use_case with id voice-cloning.clone-speaks-other-language. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
