# Let an agent create, find and delete clones through the API or MCP

> Let an agent create, find and delete clones through the API or MCP: 0 companies fully do it, 11 with a limit, 111 carry the tag. Every listing in Voice cloning & custom voices with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/voice-cloning.agent-runs-clone-pipeline/
- Markdown: https://www.anchorterminal.com/niches/voice-cloning.agent-runs-clone-pipeline/index.md (~1,600 tokens)
- Slim: https://www.anchorterminal.com/niches/voice-cloning.agent-runs-clone-pipeline/index.min.md (~430 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/voice-cloning.agent-runs-clone-pipeline/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 | 111 | 8 | 4 |

## What counts as done

The clone tool takes a file path or upload from an MCP client and returns a voice id; own clones can be listed without paging the shared library; creation is idempotent or documented to check before retrying; status is pollable or sent to a webhook; deletion exists in the API; the cost of a clone is known before the call.

How agents ask for it: clone this voice with the ElevenLabs MCP; list my cloned voices; delete the clone we made last month; how much will this clone cost before I make it

## 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.)**: The MCP has clone_voice, update_voice and delete_voice, and a delete endpoint exists, but there is no idempotency key on clone and no pre-call cost estimate (acc-08, still open).
- **ElevenLabs Voice Cloning and Voice Design API (Eleven Labs Inc.)**: The official MCP saves output to files under ELEVENLABS_MCP_BASE_PATH and its voice_clone fix for corrupted uploads shipped (acc-05, closed July 2026); no idempotency key on voice create and no cost preview were found.
- **Fish Audio Voice Cloning API (Hanabi AI Inc.)**: The MCP gives get_upload_url for local files, create_voice_clone and search_voices with self=true for own voices, and TTS returns audio URLs; no delete-voice tool or idempotency on create was found.
- **Gradium Voice Cloning and Voice Design (Gradium SAS)**: List, update and delete voice endpoints exist and candidates can be deleted, but there is no MCP clone tool and no idempotency key on create; the free plan holds 5 custom voices.
- **Inworld Instant Voice Cloning (Theai, Inc. dba Inworld AI)**: List voices with filter owned="true" returns own voices, delete exists, voices export by copy token, and a full library fails with a named 400; no idempotency key on clone, and the response echoes up to 15 seconds of base64 audio.
- **MiniMax Voice Clone and Voice Design (Nanonoble Pte. Ltd.)**: The caller picks voice_id and a duplicate is refused, which makes a retry safe, and get and delete voice endpoints exist; acc-06 saw the MCP upload fail with a 502 (closed), and unused clones vanish after 7 days.
- **Resemble AI Voice Cloning API (Resemble AI, Inc.)**: Create, build and a callback_uri webhook let an agent run the job, but dataset_url must be a public URL rather than a file upload in that path, and no idempotency key or price before the call was found.
- **Smallest.ai Instant Voice Clone (Smallest, Inc.)**: List and delete clone endpoints exist and the clone limit fails with a named 400, but there is no idempotency key, and voice_clone_timeout says retry, which can create a duplicate unless clones are listed first.
- **Soniox Voice Cloning (Soniox Inc.)**: Voice names are unique per project, so a retried create with the same name is rejected rather than duplicated, status is pollable and delete exists; no MCP clone tool, and voices are capped per project and organisation.
- **Speechify API Voice Cloning (Speechify, Inc.)**: POST /v1/voices takes an Idempotency-Key so a lost response replays without spending the challenge, the voice is ready on return, clones can be listed, inspected and deleted, and paid plans have no per-clone fee. But the only MCP serves docs, so there is no clone tool for an MCP client.
- **Ultravox Voice Cloning (Theai, Inc. dba Inworld AI)**: One multipart call with a file path creates the voice and names must be unique, which stops duplicates; no MCP clone tool, and the one-clone limit means an agent cannot make a second without support.
- **VoiceStudio (VoiceStudio maintainers)**: clone_voice reads a path confined to OMNIVOICE_MCP_BASE_PATH, list_voices enumerates own profiles and cost is zero locally, but deleting saved content needs administrator access over HTTP, and no tool carries destructive annotations (listing notes).

- 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.
- not this job: Murf Voice Cloning API (Murf Inc.): A code search of the Python SDK finds no clone method, and the Murf MCP README covers text-to-speech voiceovers only, with no clone tool.

## What stops agents finishing

- no idempotency key, so a retried clone makes a duplicate (Cartesia, Inworld, Smallest)
- clone visibility defaults to unlisted on the MCP, usable by anyone with the id (Fish Audio)
- clone deleted if not used within 7 days (MiniMax)
- no SDK method for cloning, HTTP only (Murf)

## Around it

Before: file-storage.read-file. After: voice-cloning.cloned-voice-in-live-agent, text-to-speech.clone-voice-speak.

## Where this comes from

Weight 8 of 100 in Voice cloning & custom voices: 4 of 34 accounts: ElevenLabs MCP rejecting TTS-made audio as corrupted, MiniMax MCP clone failing with a 502 after account verification, a missing language field on Cartesia's create, and a request for pre-flight cost on Cartesia

As data: https://www.anchorterminal.com/api/v1/use-cases/voice-cloning.agent-runs-clone-pipeline.json, or the MCP tool get_use_case with id voice-cloning.agent-runs-clone-pipeline. 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
