MiniMax Voice

Voice DesignVoice CloningReusable Voice ID

MiniMax Voice on APIXO creates reusable custom voice IDs through two distinct workflows: text-described Voice Design and single-clip Voice Clone. Design includes a required voice preview, while Clone can return preview audio when preview text is supplied. The resulting voice ID can be reused for subsequent text-to-speech generation through the separate MiniMax Speech 2.8 route.

APIXO Modes

Design / Clone

Price

$0.50 / request

Result

Reusable Voice ID

Clone Input

One Audio URL

Availability

Live on APIXO

Loading workspace...

MiniMax Voice Model API Rate Structure

MiniMax Voice is billed per accepted custom voice creation request. Design and clone workflows use the same backend-authoritative APIXO price.

Voice Design

Billing basis

Accepted creation request

APIXO price

$0.50 / request

Voice Clone

Billing basis

Accepted creation request

APIXO price

$0.50 / request

Mode	Billing basis	APIXO price
Voice Design	Accepted creation request	$0.50 / request
Voice Clone	Accepted creation request	$0.50 / request

Common Cost Examples

Example: 3 MiniMax Voice design requests = 3 x $0.50 = $1.50.

Checking API access...

Integrate with AI

Copy a prompt for your AI assistant to install APIXO MCP, read the live schema, and wire this model into your app.

Endpoints

Create task

POSThttps://api.apixo.ai/api/v1/generateTask/minimax-voice

Poll status

GEThttps://api.apixo.ai/api/v1/statusTask/minimax-voice?taskId={taskId}

Request body

request_type

async

Use callback with callback_url for webhooks.

input

object

Mode, prompt, references, size, duration, and audio settings.

request_typerequired

Typeselect

Use async to poll statusTask, or callback to receive a webhook when the result is ready.async · callback

callback_urlconditional

Typetext

Public HTTPS URL for callback delivery.Only in request_type=callback

moderequired

Typeselect

Choose the generation backend version.design · clone

voice_id_prefixrequired

Typetext

Prefix used by MiniMax to generate the final reusable voice_id. Use letters and digits, at least 6 characters, starting with a letter.

promptconditional

Typestring

Required in design mode. Describe accent, tone, age, pacing, and use case.Only in mode=design

preview_textconditional

Typestring

Text used to preview the created voice. Design mode is limited to 500 characters.

need_noise_reductionconditional

Typeboolean

Enable noise reduction for clone reference audio.Only in mode=clone

need_volume_normalizationconditional

Typeboolean

Normalize reference audio volume before cloning.Only in mode=clone

accuracyconditional

Typeslider

Optional clone accuracy control from 0 to 1.Only in mode=clone

language_boostconditional

Typeselect

Optional language hint for clone recognition.auto · English · Chinese · Chinese,Yue · Arabic · Russian · Spanish · French · Portuguese · German · Turkish · Dutch · Ukrainian · Vietnamese · Indonesian · Japanese · Italian · Korean · Thai · Polish · Romanian · Greek · Czech · Finnish · HindiOnly in mode=clone

Parameter	Type	Required	Description
`request_type`	select	required	Use async to poll statusTask, or callback to receive a webhook when the result is ready.async · callback
`callback_url`	text	conditional	Public HTTPS URL for callback delivery.Only in request_type=callback
`mode`	select	required	Choose the generation backend version.design · clone
`voice_id_prefix`	text	required	Prefix used by MiniMax to generate the final reusable voice_id. Use letters and digits, at least 6 characters, starting with a letter.
`prompt`	string	conditional	Required in design mode. Describe accent, tone, age, pacing, and use case.Only in mode=design
`preview_text`	string	conditional	Text used to preview the created voice. Design mode is limited to 500 characters.
`need_noise_reduction`	boolean	conditional	Enable noise reduction for clone reference audio.Only in mode=clone
`need_volume_normalization`	boolean	conditional	Normalize reference audio volume before cloning.Only in mode=clone
`accuracy`	slider	conditional	Optional clone accuracy control from 0 to 1.Only in mode=clone
`language_boost`	select	conditional	Optional language hint for clone recognition.auto · English · Chinese · Chinese,Yue · Arabic · Russian · Spanish · French · Portuguese · German · Turkish · Dutch · Ukrainian · Vietnamese · Indonesian · Japanese · Italian · Korean · Thai · Polish · Romanian · Greek · Czech · Finnish · HindiOnly in mode=clone

File parameters (URL arrays)

audio_urls[mode=clone]Required in clone mode. Upload one public MP3, M4A, or WAV reference audio file.Only in mode=clone

Example (cURL)

curl -X POST "https://api.apixo.ai/api/v1/generateTask/minimax-voice" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "request_type": "async",
    "input": {
      "mode": "design",
      "voice_id_prefix": "apixod72644139",
      "prompt": "A warm, confident product narrator voice for SaaS demos. Clear American English articulation, natural pacing, friendly energy, polished but not overly corporate.",
      "preview_text": "Welcome to APIXO. This MiniMax Voice API example designs a reusable custom voice for product narration and customer-facing audio."
    }
  }'

Full API documentation

Capabilities

Two Paths to a Reusable MiniMax Voice

Text-Described Voice Design

Describe the desired accent, apparent age, tone, pacing, energy, and intended use without supplying source audio. MiniMax generates a custom voice and reads the provided preview text so the result can be evaluated.

Single-Clip Voice Clone

Clone mode derives a reusable voice identity from exactly one public reference-audio URL. The recording acts as the speaker source rather than as speech content for conversion or dubbing.

Reusable Voice Asset

Both workflows return a MiniMax voice ID instead of only a finished audio file. That identifier can be stored and selected in compatible downstream synthesis requests when a recurring narrator or character is needed.

Preview Before Synthesis

Voice Design requires preview text and returns audio for evaluating the generated voice. Clone mode can also produce preview audio when optional preview text is supplied, allowing teams to review the result before wider synthesis use.

Clone Cleanup Controls

APIXO exposes optional noise reduction and volume normalization for clone references. These controls can improve source preparation when a permitted recording contains inconsistent level or manageable background noise.

Recognition Guidance

Clone mode includes a language hint and an adjustable accuracy control. The hint assists recognition of the reference speech, while accuracy lets applications tune the requested cloning behavior within APIXO’s supported range.

Creation Modes

Design an Original Voice or Clone an Authorized Speaker

Voice Design from a Written Brief

Provide a voice description, preview text, and valid identifier prefix. This mode is suited to fictional characters and brand narrators because it creates a voice from requested traits without requiring a real speaker recording.

No Audio Input

Voice Clone from One Recording

Supply one public MP3, M4A, or WAV URL from a speaker you are authorized to reproduce. Optional cleanup, recognition, and accuracy controls prepare the source before APIXO returns the reusable cloned voice ID. Supply optional preview text when an audio sample is needed for validation.

Reference Audio

Specifications

MiniMax Voice Inputs on APIXO

Each mode has separate requirements and should be implemented independently.

Description + Preview

Design Input

1–500 Characters

Design Preview

Exactly One URL

Clone Input

MP3 / M4A / WAV

Clone Formats

0–1

Clone Accuracy

Letter-First · 6+ Chars

Voice ID Prefix

Use Cases

Build Reusable Voices for Production Systems

Brand Audio

Original Brand Narrators

Design a voice around traits such as warmth, authority, pace, age, and accent, then validate it with a representative preview. The resulting ID can anchor product tours, explainers, and campaign narration without copying a real individual.

Authorized Talent

Licensed Speaker Workflows

Create a reusable voice from a clean recording supplied under appropriate consent and usage terms. This can extend an approved narrator or performer across localized scripts and recurring content while keeping source authorization explicit.

Interactive Media

Recurring Character Voices

Generate distinct voices for game characters, learning companions, digital presenters, or fictional assistants. Store the returned identifiers with character metadata, then use the separate synthesis route whenever new dialogue is required.

Accessibility

Consistent Reading Voices

Establish a familiar custom voice for educational material, interface guidance, or accessible reading experiences. Review pronunciation and emotional delivery in downstream synthesis, particularly when scripts change language or contain specialized terminology.

Notes & FAQ

Voice Creation, Activation, and Rights

Important Notes

MiniMax Voice creates reusable voice assets; it does not expose final text-to-speech narration controls.

Design mode requires a voice description, preview text, and valid voice ID prefix.

Clone mode requires exactly one publicly reachable MP3, M4A, or WAV reference URL.

A newly created voice that is never used for later synthesis may become unavailable after seven days.

APIXO recommends beginning status checks about ten seconds after creation, but actual processing time can vary.

Clone only voices you are authorized to process and reproduce; successful creation does not establish consent or commercial rights.

Frequently Asked Questions

Design creates an original custom voice from a written description and preview script. Clone instead derives a voice from one reference recording. The two modes have different required inputs but both return a reusable MiniMax voice ID.

No. The route creates and previews a reusable voice asset. Pass its voice ID to APIXO’s separate MiniMax Speech 2.8 route when you need to synthesize complete narration, dialogue, or application speech.

APIXO charges $0.50 for each accepted Voice Design or Voice Clone creation request. Later text-to-speech generation is a separate operation and follows the character-based pricing of the selected MiniMax Speech 2.8 tier.

Use a clear, authorized recording focused on one speaker. Music, overlapping voices, long silence, strong echo, clipping, and heavy background noise can weaken the usable speaker signal even when the file is technically accepted.

Clone mode provides language hints for supported recognition contexts, and compatible MiniMax synthesis models may reuse custom voice IDs across languages. Accent, pronunciation, identity, and emotional character can drift, so cross-language results require speaker-aware review.

Explore Other Models

Discover more AI models for your next creative workflow

View all models

MiniMax

MiniMax Speech 2.8

MiniMax Speech 2.8 is available on APIXO for Audio. The model page brings examples, creation controls, pricing, and results into one focused workspace.

AudioNew

$0.06/1000 chars

suno

Suno V5

Suno V5 is available on APIXO for Audio. The model page brings examples, creation controls, pricing, and results into one focused workspace.

AudioNew

$0.12/audio