# Clone a voice on your own hardware, with no upload

> Clone a voice on your own hardware, with no upload: 0 companies fully do it, 1 with a limit, 105 carry the tag. Every listing in Voice cloning & custom voices with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/voice-cloning.local-private-clone/
- Markdown: https://www.anchorterminal.com/niches/voice-cloning.local-private-clone/index.md (~1,250 tokens)
- Slim: https://www.anchorterminal.com/niches/voice-cloning.local-private-clone/index.min.md (~280 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/voice-cloning.local-private-clone/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 | 1 | 105 | 16 | 8 |

## What counts as done

The app or model installs from documented steps, exposes the clone through an HTTP or MCP interface an agent can call, writes audio to files rather than into context, states the licence of each cloning model and whether commercial use is allowed, and keeps the same speaker across a long text.

Measured, for our own tests: a 200-word passage in the cloned voice renders on the stated hardware in under twice real time with the speaker unchanged from start to end.

How agents ask for it: clone my voice locally; voice cloning without sending audio to a cloud; run voice cloning on my GPU; offline voice clone for an audiobook

## 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.

- **VoiceStudio (VoiceStudio maintainers)**: A local desktop app with MCP over stdio or http://localhost:3900/mcp/; OMNIVOICE_MCP_OUTPUT_MODE=files keeps audio out of context (default resources returns base64). Models not declared commercial raise model_licence_required until accepted in the Model Catalogue UI. Speed bar unpublished.

- not this job: Cartesia Voice Cloning API + MCP (Cartesia AI, Inc.): Hosted API only; no local model or app.
- not this job: ElevenLabs Voice Cloning and Voice Design API (Eleven Labs Inc.): Hosted API only; the MCP server calls the hosted API and needs credits.
- not this job: Fish Audio Voice Cloning API (Hanabi AI Inc.): This listing is the hosted API and MCP; the open fish-speech model is a separate project.
- not this job: Gradium Voice Cloning and Voice Design (Gradium SAS): Hosted API only.
- 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: Inworld Instant Voice Cloning (Theai, Inc. dba Inworld AI): Hosted API only; the CLI and its MCP call the hosted service.
- not this job: MiniMax Voice Clone and Voice Design (Nanonoble Pte. Ltd.): Hosted API only.
- not this job: Murf Voice Cloning API (Murf Inc.): Hosted Enterprise API only; the GitHub SDK and MCP repositories call the hosted service.
- not this job: Resemble AI Voice Cloning API (Resemble AI, Inc.): This listing is the hosted API; Resemble's open Chatterbox model is a separate project.
- not this job: Smallest.ai Instant Voice Clone (Smallest, Inc.): Hosted API only.
- not this job: Soniox Voice Cloning (Soniox Inc.): Hosted API only.
- not this job: Speechify API Voice Cloning (Speechify, Inc.): Hosted API only.
- not this job: Ultravox Voice Cloning (Theai, Inc. dba Inworld AI): Hosted Ultravox Realtime only.

## What stops agents finishing

- default model is non-commercial and returns a licence error until accepted in the app (VoiceStudio)
- context window caps output length (acc-19)
- the cloning model needs a GPU the user does not have
- a person must accept the model licence in a desktop UI before the first API call

## Around it

Before: file-storage.read-file. After: text-to-speech.long-form-narration, text-to-speech.self-hosted-local.

## Where this comes from

Weight 16 of 100 in Voice cloning & custom voices: 8 of 34 accounts: VoiceStudio issues (Android cloning stalls, one take at a time), open-weights clones that change speaker, speed up or stop after 17 seconds, and people cloning their own or a relative's voice locally for audiobooks and podcasts. Most accounts are on models that are not listings (Qwen3-TTS, Neutts-air, VibeVoice), recorded because the agent's job is the same

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