# MiniMax Voice Clone and Voice Design > MiniMax's API clones a voice from a 10 second to 5 minute audio sample and designs voices from a text description. The resulting voice IDs work with its speech-2.8 text-to-speech models. - Canonical: https://www.anchorterminal.com/tools/minimax-voice-cloning - Markdown: https://www.anchorterminal.com/tools/minimax-voice-cloning.md (~6,850 tokens) - Slim: https://www.anchorterminal.com/tools/minimax-voice-cloning.min.md (~1,680 tokens, same facts, less prose, for token-sensitive contexts) - JSON: https://www.anchorterminal.com/tools/minimax-voice-cloning.json (this page as data, same URL with Accept: application/json) - Site index for agents: https://www.anchorterminal.com/llms.txt (full text: https://www.anchorterminal.com/llms-full.txt) - API: https://www.anchorterminal.com/api/v1/index.json - Updated: 2026-10-09 ## Overview **Grade D · 53.1/100 · rank #640 of 842 · #7 in Voice cloning & custom voices · not agent-ready · confidence medium** More from MiniMax, listed separately because each is its own product: [MiniMax Video API](https://www.anchorterminal.com/tools/minimax-video.md) (Video generation). ## Assessment Typed OpenAPI files, published prices of $1.50 a cloned voice and $3 a designed voice, and optional transcript and watermark checks. Cloning needs individual or enterprise account verification, the terms and privacy policy could not be read without a browser, and no statement on training with uploaded voice audio was found in the docs. ## Facts | Field | Value | | --- | --- | | Vendor | MiniMax (https://www.minimax.io) | | Kind | Model API | | Category | Voice cloning & custom voices (https://www.anchorterminal.com/categories/voice-cloning) | | Transport | HTTP, stdio | | Endpoint | `https://api.minimax.io` | | Auth | API key · Bearer API key created in the platform console under Account, API Keys. No scopes were found in the docs. Voice cloning also needs individual or enterprise verification of the account, and an unverified account gets error 2038. Mainland China accounts use `https://api.minimaxi.com`. | | Pricing | Pay per use ($1.50 / tx) · Pay as you go from a prepaid balance. Rapid voice cloning $1.50 a voice and voice design $3 a voice, each charged when the voice is first used for synthesis. Speech from the voice is $100 per million characters on speech-2.8-hd and $60 on speech-2.8-turbo. Previews are billed per character. Audio subscriptions from $5 a month include 10 to 800 voice slots. No free allowance for cloning is published (https://platform.minimax.io/docs/guides/pricing-paygo, checked 2026-10-08). | | x402 | No · No x402, MPP or L402 in the docs index, the pricing pages or the voice OpenAPI files. Prepaid balance top-up (checked 2026-10-08). | | Licence | Proprietary API under MiniMax's platform terms. The two official MCP servers are MIT | | Tools exposed | 8 | | Packages | pypi: `minimax-mcp`; npm: `minimax-mcp-js` | | Source | https://github.com/MiniMax-AI/MiniMax-MCP | | Docs | https://platform.minimax.io/docs/guides/speech-voice-clone | | llms.txt | https://platform.minimax.io/docs/llms.txt | | Last release | 2026-08-21 | | npm downloads / week | 377 | | PyPI downloads / week | 1,532 | | Sample length | One file of 10 seconds to 5 minutes, mp3, m4a or wav, up to 20 MB. An optional prompt clip under 8 seconds with a matching transcript improves similarity | | Instant vs professional | Rapid cloning only, returned in one synchronous call. No slower professional tier in the API | | Voice design | `POST /v1/voice_design` takes a description and a preview text of up to 500 characters and returns a `voice_id` with hex-encoded preview audio | | Consent and verification | Individual or enterprise verification of the account is required. No speaker consent field. Optional `text_validation` transcript check and optional `aigc_watermark` tone on the preview | | Voice ownership | Not established. The terms of service did not render for our reader. A voice can be used only by the account that created it (error 2042) | | Languages | Any language for the sample, 40 languages for synthesis | | Endpoints | `POST /v1/files/upload`, `POST /v1/voice_clone`, `POST /v1/voice_design`, `POST /v1/get_voice`, `POST /v1/delete_voice` | | Free tier | None published for cloning or voice design | | Rate limits | Voice cloning 60 requests a minute, voice design 20, T2A 60. Audio subscriptions set 10 to 800 requests a minute | | Data retention | An unused voice is deleted after 7 days and a used one is kept until deleted through the API. No retention period for uploaded audio was found | | MCP server | Official, stdio or SSE, `minimax-mcp` 0.0.19 on PyPI and `minimax-mcp-js` 0.0.18 on npm, 8 tools, 3 for voices | | Capabilities | voice.clone, voice.design, speech.tts | | Tags | official, hosted, usage-priced, prepaid, openapi, llms-txt, mcp, python, typescript, status-page | | JSON | https://www.anchorterminal.com/api/v1/tools/minimax-voice-cloning.json | ## Score breakdown (methodology v0.4, October 2026 research run) Assessed 2026-10-08 from public evidence against the published checklist (https://www.anchorterminal.com/benchmark/#checklist). Confidence: medium. Performance and Task success pending (no score, not in the total); the total is Σ(score × weight) ÷ 80 over the 7 assessed categories. "This run" is each category's share of the 100 points. | Category | Weight | This run | Score (0–100) | Points | | --- | --- | --- | --- | --- | | Reliability | 16% | 20 | 70 | 14.0 | | Performance | 10% | pending | pending | n/a | | Schema & documentation | 13% | 16.2 | 78 | 12.7 | | Agent ergonomics | 13% | 16.2 | 56 | 9.1 | | Security & auth | 14% | 17.5 | 39 | 6.8 | | Payments & pricing | 10% | 12.5 | 20 | 2.5 | | Task success | 10% | pending | pending | n/a | | Maintenance & community | 7% | 8.8 | 38 | 3.3 | | Transparency & trust (editorial 25, provenance 80) | 7% | 8.8 | 53 | 4.6 | | Negative events | up to −15 | up to −15 | none recorded | 0 | | **Total** | | | | **53.1 → D** | ### Why each score - Reliability 70: Atlassian Statuspage at status.minimax.io with a Text-to-Speech component and no separate component for cloning (20). The page showed incidents from 28 September to 8 October 2026, one of them on Text-to-Speech, 22 minutes of elevated errors on 28 September. The history page did not render and the feed held three days, so the rest of the 90 days was unread, minor only on what we saw (20). Rate limits are published, 60 requests a minute for voice cloning and 20 for voice design (15). Error 1002 and 2045 say to retry later and avoid sudden increases, with no Retry-After, backoff figures or idempotency key found (5 of 15). No SLA found in the docs, and the terms were unread (0). No beta or preview label on either endpoint (10). - Performance: Pending. Latency is measured per call by our probes, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until the first probe window closes. - Schema & documentation 78: OpenAPI 3.1 files for cloning, upload, voice design and voice management (25). llms.txt and a Markdown copy of every docs page (10). Field descriptions state purpose, limits and billing effects, with little on when to choose cloning over design (13 of 20). Enums for model, language and voice type, a 0 to 1 range on `accuracy` and length caps, while the `voice_id` rules sit in prose and the `file_id` description carries a sentence that belongs to `clone_prompt` (12 of 15). Request and response examples throughout and a table of 24 error codes, but only a 200 response is specified and the example response uses fields the schema does not define (10 of 15). Paths are versioned under `/v1`. The API release notes stop at 28 October 2025 and omit `text_validation` and `aigc_watermark` (8 of 15). - Agent ergonomics 56: The clone call returns a preview URL and billing counts, but voice design returns preview audio as hex inside the JSON body, and listing with `voice_type` set to `all` returns the 300+ system voices (15 of 25). `POST /v1/get_voice` filters by voice type and has no pagination (10 of 20). Errors come as numeric codes in `base_resp` with a short remedy for each, several of them specific to cloning, inside a response documented only as 200 (13 of 20). No idempotency key. A repeated clone with the same `voice_id` fails with 2039 instead of creating a second voice, the fee applies only on first synthesis, and the MCP tools carry no read-only or destructive annotations (10 of 20). Two required fields per call. No official SDK for speech, with MCP servers in Python and JavaScript and an official CLI that does not cover cloning (8 of 15). - Security & auth 39: Graded for voice cloning, with consent and misuse controls in place of the read-only line and training and retention of voice data in place of the prompt-injection line, as on the other listings in this category. Plain API keys from the console with no scopes found, and the docs say publicly leaked keys may be disabled (20 of 30). Cloning requires individual or enterprise verification of the account, input passes a content safety check, and two optional controls exist, a transcript match (`text_validation`) and a watermark tone (`aigc_watermark`, off by default). No field records the speaker's consent (8 of 20). The terms and privacy policy did not render for our reader, so training use is unknown. The docs state that unused voices are deleted after 7 days and a delete call exists (5 of 15). Responses carry `usage_characters` for reconciliation and voices can be listed, with no audit log found (6 of 15). No security.txt, disclosure policy, bug bounty or certification found, and an issue opened on the MCP repository on 25 August 2026 asks for a private disclosure contact (0 of 20). - Payments & pricing 20: No x402, MPP or L402 (0). Per-unit prices are public, $1.50 a cloned voice, $3 a designed voice and $60 to $100 per million characters of speech (20). No free allowance for cloning or design is published. The rate limits page mentions a free tier without defining it (0). A person signs up in a browser, funds a balance and completes verification before the first clone (0). - Task success: Pending. Task success needs the category task suites run through each tool, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until then. A data provider's data-quality score is published on its listing now and becomes half of this category when it's scored. - Maintenance & community 38: The newest dated change that touches this product is `minimax-mcp` 0.0.19 and `minimax-mcp-js` 0.0.18 on 21 August 2026, 48 days ago. The newest speech model, Speech 2.8, is dated 23 January 2026 (20 of 30). Two package releases for one change in the last 90 days, and no dated API entries (0). The API release notes stop at 28 October 2025. Support is by email and Discord, and the MCP repository lists 21 open issues, whose replies we could not read (6 of 15). No official SDK for speech. The MCP servers were updated in August 2026 and include `voice_clone` and `voice_design` (8 of 15). The Python server has tests and moved to version 2 of the MCP SDK in August, with no CI workflow in the repository (4 of 10). - Transparency & trust 53: Closed API under platform terms that exist but did not render for our reader, with MIT MCP servers (12 of 30). The privacy policy was unread. The docs say every interface is stateless and stores no user data, while also saying an activated voice is saved permanently, and they give no retention period for uploaded audio (5 of 30). The earlier text-to-voice API was marked legacy on 20 June 2025 with service continuing, and legacy speech models are labelled, with no written deprecation policy (8 of 20). No subprocessor list or data locations found. Mainland China accounts use a separate host (0 of 20). Fix list for a coding agent, everything this grade says the listing lacks, the biggest gain first (19 items): https://www.anchorterminal.com/fixes/minimax-voice-cloning.md (JSON https://www.anchorterminal.com/fixes/minimax-voice-cloning.json) ### What we couldn't check - unchecked: the terms of service, privacy policy and paid agreement at platform.minimax.io/protocol/ return no document text to a plain reader, so voice ownership, the licence taken over uploads, training use, retention, benchmarking rules and any SLA are unread - unchecked: the legal entity Nanonoble Pte. Ltd. is carried from the 30 September check on the MiniMax Video API listing and was not confirmed today - unchecked: status history before 28 September 2026. The history page did not render and the feed held three days - unchecked: GitHub star counts and maintainer replies on issues. The GitHub API refused us for rate limits - What individual or enterprise verification involves, and whether it applies to voice design as well as cloning - Which preview price for voice design is right, $30 per million characters in the API reference or $60 on the pricing page - What the free tier named on the rate limits page includes - How long uploaded sample files are kept, and how the statement that interfaces store no user data fits with permanently saved voices ### Sources - docs index (llms.txt): (seen 2026-10-08) - voice clone guide: (seen 2026-10-08) - voice clone reference and OpenAPI: (seen 2026-10-08) - voice design reference: (seen 2026-10-08) - get voice and delete voice references: (seen 2026-10-08) - API overview, speech section: (seen 2026-10-08) - rate limits: (seen 2026-10-08) - error codes: (seen 2026-10-08) - pay-as-you-go pricing: (seen 2026-10-08) - audio subscription pricing: (seen 2026-10-08) - API release notes: (seen 2026-10-08) - model release notes: (seen 2026-10-08) - API FAQ: (seen 2026-10-08) - status page: (seen 2026-10-08) - status feed: (seen 2026-10-08) - terms of service (no text without a browser): (seen 2026-10-08) - privacy policy (no text without a browser): (seen 2026-10-08) - official MCP server, Python: (seen 2026-10-08) - official MCP server, JavaScript: (seen 2026-10-08) - MCP repository issues: (seen 2026-10-08) - PyPI release history: (seen 2026-10-08) - npm release history: (seen 2026-10-08) - Speech 2.8 announcement: (seen 2026-10-08) - domain registration (RDAP): (seen 2026-10-08) ## Who's behind it (provenance 80/100, checked 2026-10-08) | Check | Finding | Points | | --- | --- | --- | | Legal entity named | Nanonoble Pte. Ltd. | 20/20 | | Domain age | minimax.io, registered 2021-03-27 (5 years) | 11/15 | | Endpoint on the vendor's domain | api.minimax.io | 15/15 | | Terms of service | published, but our reader couldn't read it | 7/10 | | Privacy policy | published, but our reader couldn't read it | 7/10 | | Status page | status.minimax.io | 10/10 | | Changelog | published | 10/10 | | security.txt | not found | 0/10 | The legal entity is carried from the 30 September 2026 check recorded on our MiniMax Video API listing. The terms page that would name it renders only in a browser and was not read on 8 October. The terms of service, privacy policy and paid agreement at platform.minimax.io/protocol/ return a page with no document text to a plain reader. RDAP for minimax.io gives a registration date of 2021-03-27 and a transfer on 2025-01-31. platform.minimax.io/.well-known/security.txt returns 404, and www.minimax.io/.well-known/security.txt redirects to the home page. The API release notes page has no entry after 28 October 2025. Model releases are dated on a separate page. ### Terms and privacy, as read A reading by a fixed set of rules, each answered with the vendor's own sentence. Not legal advice. **Terms of service** (https://platform.minimax.io/protocol/terms-of-service), read 2026-10-08. Our reader couldn't read it (the page has 0 words of text without a browser, so the document is drawn by script or sits elsewhere). **Privacy policy** (https://platform.minimax.io/protocol/privacy-policy), read 2026-10-08. Our reader couldn't read it (the page has 0 words of text without a browser, so the document is drawn by script or sits elsewhere). ## Live (updated 2026-10-09 10:42 UTC) - Right now: up, HTTP 404, 306 ms, checked 2026-10-09 10:42 UTC (get on `https://api.minimax.io`) - Uptime 24h 100.0% (33 probes) · 30 days 100.0% (33 probes) · p50 290 ms · p95 567 ms - Vendor status page: none, All Systems Operational - Always current: https://www.anchorterminal.com/api/v1/live/minimax-voice-cloning.json ## Probe metrics Not measured yet. Our benchmark probes haven't run, so there's no availability, latency or error rate from a run and Performance is pending. Live uptime, where we poll the endpoint, is under Live and doesn't change the score. ## Prices | Item | Price | Unit | Note | | --- | --- | --- | --- | | Rapid voice clone | $1.50 | per transaction | per voice, charged on first synthesis | | Voice design | $3 | per transaction | per voice, charged on first synthesis | | Speech from a clone, speech-2.8-hd | $100 | per 1M characters | | | Speech from a clone, speech-2.8-turbo | $60 | per 1M characters | | Across all listings: https://www.anchorterminal.com/prices/index.md ## Strengths - OpenAPI 3.1 files for clone, upload, design, list and delete, with llms.txt and Markdown copies of every docs page - A cloned voice costs $1.50 and a designed voice $3, charged on first synthesis and not at creation - Rate limits are published, 60 requests a minute for cloning and 20 for voice design - Optional `text_validation` rejects a sample whose speech does not match an expected transcript, and `aigc_watermark` adds a watermark tone to the preview - `POST /v1/delete_voice` removes a cloned or designed voice, and unused voices are deleted after 7 days ## Weaknesses - The terms of service and privacy policy render only in a browser, so voice ownership, training use and retention terms were not read - No field records the speaker's consent. The documented gate is individual or enterprise verification of the account - The voice design reference prices preview audio at $30 per million characters and the pricing page at $60 - The API release notes stop at 28 October 2025 and omit `text_validation` and `aigc_watermark` - No security.txt, bug bounty or certification found, and no SLA in the docs ## Before you call it (notes for agents) 1. Upload the sample to `POST /v1/files/upload` with `purpose=voice_clone`, then pass the integer `file_id` and a new `voice_id` of 8 to 256 characters to `POST /v1/voice_clone` 2. Synthesise once with T2A within 7 days or the voice is deleted. A preview inside the clone call does not count 3. Read `base_resp.status_code` on every response. Errors such as 2038 (no cloning permission) and 2039 (duplicate `voice_id`) arrive in the body 4. Voice design returns `trial_audio` as hex in the JSON body. Write it to a file and keep it out of the model context 5. Send `voice_type` to `POST /v1/get_voice`, because `all` also returns the 300+ system voices ## Connect First request: ```bash curl --location 'https://api.minimax.io/v1/voice_clone' \ --header "Authorization: Bearer $MINIMAX_API_KEY" \ --header 'Content-Type: application/json' \ --data '{"file_id": 123456789, "voice_id": "MiniMax001"}' ``` MCP client configuration: ```json { "mcpServers": { "MiniMax": { "args": [ "minimax-mcp", "-y" ], "command": "uvx", "env": { "MINIMAX_API_HOST": "https://api.minimax.io", "MINIMAX_API_KEY": "\u003ckey\u003e" } } } } ``` ## Similar tools Ranked by shared capabilities, then score. Same-category tools with no shared capability key are listed last. | Tool | Grade | Score | Rank | Shared capabilities | x402 | Markdown | | --- | --- | --- | --- | --- | --- | --- | | ElevenLabs Voice Cloning and Voice Design API | BB | 73.6 | 83 | voice.clone, voice.design, speech.tts | no | https://www.anchorterminal.com/tools/elevenlabs-voice-cloning.md | | Hume Octave Voice Design and Cloning + MCP | C | 54.6 | 611 | voice.clone, voice.design, speech.tts | no | https://www.anchorterminal.com/tools/hume-voice-cloning.md | | Fish Audio Voice Cloning API | D | 51.3 | 671 | voice.clone, voice.design, speech.tts | no | https://www.anchorterminal.com/tools/fish-audio-voice-cloning.md | | Speechify API Voice Cloning | BB | 74.1 | 74 | voice.clone, speech.tts | no | https://www.anchorterminal.com/tools/speechify-voice-cloning.md | | Fish Audio TTS API | C | 60.7 | 453 | speech.tts, voice.clone | no | https://www.anchorterminal.com/tools/fish-audio-tts.md | | Cartesia Voice Cloning API + MCP | C | 59.5 | 493 | voice.clone, speech.tts | no | https://www.anchorterminal.com/tools/cartesia-voice-cloning.md | ## Panel reviews (0) Reviewed by the Anchor panel (https://www.anchorterminal.com/reviewers/index.md): . Desk reviews, written from public documentation, pricing, terms, source and status history on 1 October 2026. No calls made. For a desk review, the outcome says whether the reviewer's questions could be answered from public material: success, partial or failure. How reviews work: https://www.anchorterminal.com/reviews/how-it-works.md ## Notable - A clone needs one mp3, m4a or wav sample of 10 seconds to 5 minutes and up to 20 MB, plus an optional prompt clip under 8 seconds with its transcript (source: ) - Cloned and designed voices are temporary until first used in speech synthesis, and are deleted after 168 hours if unused (source: ) - The API overview says voice cloning requires individual or enterprise verification (source: ) - `text_validation` sends the sample to speech recognition and rejects the request with code 1043 when similarity to the expected transcript falls below `accuracy`, 0.7 by default (source: ) - The official MCP servers have 8 tools after the music tool was retired on 20 August 2026, including `voice_clone`, `voice_design` and `list_voices` (source: ) - The voice design reference states a preview fee of $30 per million characters, and the pay-as-you-go page states $60 (source: ) ## Compare - [Cartesia Voice Cloning API + MCP vs MiniMax Voice Clone and Voice Design](https://www.anchorterminal.com/compare/cartesia-voice-cloning-vs-minimax-voice-cloning.md): C 59.5 vs D 53.1 - [ElevenLabs Voice Cloning and Voice Design API vs MiniMax Voice Clone and Voice Design](https://www.anchorterminal.com/compare/elevenlabs-voice-cloning-vs-minimax-voice-cloning.md): BB 73.6 vs D 53.1 - [Fish Audio Voice Cloning API vs MiniMax Voice Clone and Voice Design](https://www.anchorterminal.com/compare/fish-audio-voice-cloning-vs-minimax-voice-cloning.md): D 51.3 vs D 53.1 - [Hume Octave Voice Design and Cloning + MCP vs MiniMax Voice Clone and Voice Design](https://www.anchorterminal.com/compare/hume-voice-cloning-vs-minimax-voice-cloning.md): C 54.6 vs D 53.1 - [MiniMax Voice Clone and Voice Design vs Murf Voice Cloning API](https://www.anchorterminal.com/compare/minimax-voice-cloning-vs-murf-voice-cloning.md): D 53.1 vs D 46.1 - [MiniMax Voice Clone and Voice Design vs PlayHT Voice Cloning API](https://www.anchorterminal.com/compare/minimax-voice-cloning-vs-playht-voice-cloning.md): D 53.1 vs F 3.9 - [MiniMax Voice Clone and Voice Design vs Resemble AI Voice Cloning API](https://www.anchorterminal.com/compare/minimax-voice-cloning-vs-resemble-ai-voice-cloning.md): D 53.1 vs C 54 - [MiniMax Voice Clone and Voice Design vs Soniox Voice Cloning](https://www.anchorterminal.com/compare/minimax-voice-cloning-vs-soniox-voice-cloning.md): D 53.1 vs C 58.6 - [MiniMax Voice Clone and Voice Design vs Speechify API Voice Cloning](https://www.anchorterminal.com/compare/minimax-voice-cloning-vs-speechify-voice-cloning.md): D 53.1 vs BB 74.1 - [MiniMax Voice Clone and Voice Design vs Ultravox Voice Cloning](https://www.anchorterminal.com/compare/minimax-voice-cloning-vs-ultravox-voice-cloning.md): D 53.1 vs E 40.8 ## Verify this listing For the vendor. The badge or a plain link to this page verifies the listing, from a page on minimax.io or one of its subdomains, or the README of github.com/MiniMax-AI/MiniMax-MCP. It shows the listing is the vendor's and that the vendor knows it's here, and it never changes a grade, rank or review. The vendor sends the page's address to `POST https://www.anchorterminal.com/api/v1/verify` as `{"slug": "minimax-voice-cloning", "url": "…"}`, or calls the `verify_listing` tool at https://www.anchorterminal.com/mcp. We fetch the page once, then again every week; two failed checks in a row and the verification lapses, and a later pass restores it. What we check: https://www.anchorterminal.com/builders/index.md#verify HTML badge: ```html MiniMax Voice Clone and Voice Design on Anchor Terminal ``` Markdown badge, for a README: ```markdown [![MiniMax Voice Clone and Voice Design on Anchor Terminal](https://www.anchorterminal.com/badges/minimax-voice-cloning.svg)](https://www.anchorterminal.com/tools/minimax-voice-cloning) ``` Plain link: ```html MiniMax Voice Clone and Voice Design on Anchor Terminal ``` ## Share this listing For the vendor. Sharing assets for social media, two PNGs of 1200 × 630 that say MiniMax Voice Clone and Voice Design is listed on Anchor Terminal, with the vendor's logo and this page's address and no grade or score. - Dark: https://www.anchorterminal.com/assets/share/minimax-voice-cloning-dark.png - Light: https://www.anchorterminal.com/assets/share/minimax-voice-cloning-light.png