Skip to content
Dubsheet

MiniMax Audio

Summary

MiniMax Audio is the vendor's browser studio for turning a script into speech with its Speech 2.8 models, voice cloning, voice design and music generation. The same speech models are sold separately through MiniMax's developer API.

From
Freethen $5 monthly · $4.50 billed annually, 10% less
Open source
No
API
Yes
Best for
Creators and developers wanting many languages and cloned voices at low cost

What we checked

Pages last read
Price on its site
Free planFrom its pricing page

MiniMax Audio price and terms

Terms

Free plan
YesYes
The terms in full

10,000 credits a month as a monthly gift, about twelve minutes of the HD model, three voice slots, standard generation speed and limited long text support; personal use only, with credit to the vendor.

Sign up without sales
YesYes

For AI agents

API
Yes
MiniMax Audio's pricing page

Pricing page, captured

minimax.io/audio/subscribe, checked

What we checked on MiniMax Audio

YesPartlyNo

What an hour of speech costsYes

Cost for 2 hours a month

$5.12 a month on Starter for 2 hours of 60 minutes. credits counted per character or per word are converted at the standard hour: 52,000 characters, 9,300 words

Starter $5 /month $5 Billed monthly

minimax.io/audio/subscribeChecked

What an hour of speech costsYes

Cost for 10 hours a month (heavy use)

$17.60 a month on Starter for 10 hours of 60 minutes. credits counted per character or per word are converted at the standard hour: 52,000 characters, 9,300 words

On the page: Plan cards, Monthly billing tab selected, Starter column

minimax.io/audio/subscribeChecked

What an hour of speech costsYes

Promotion and regular price

yearly shown monthly: the plan page opens on the Yearly tab, where Starter reads $4.5 a month billed yearly, while the Monthly tab shows the regular $5 a month.

$5 10% OFF $4.5 /month $54 Billed yearly

minimax.io/audio/subscribeChecked

What an hour of speech costsYes

What the price counts

Everything is counted in credits, and the vendor states one English letter equals one credit, so plans run from 10,000 credits free to 3 million on Pro.

In most cases, one English letter equals one credit, while one Chinese or Japanese character equals two credits.

minimax.io/audio/subscribeChecked

What an hour of speech costsYes

Voices that use more

The two speech model families cost different amounts: the vendor calls Turbo 40% more cost-effective than HD, and its own API rates are $60 against $100 per million characters.

HD is our highest quality model, ideal for professional content. Turbo is 40% more cost-effective and supports longer texts, perfect for drafting or…

minimax.io/audio/subscribeChecked

What an hour of speech costsYes

Past your allowance

Past the monthly allowance you buy credits as a top-up at a published rate, with a $5 minimum purchase and a two-month validity on what you buy.

Top-up Service: Priced at $30 per one million credits, this service offers one-time purchase of credits. A minimum purchase or recharge amount of $5…

minimax.io/audio/doc/paid-service-terms…Checked

What an hour of speech costsNo

Unused credits

Monthly plan credits are lost at the end of each calendar month, and separately bought top-up credits last only two months from purchase.

All monthly credits expire at the end of each calendar month.

minimax.io/audio/doc/paid-service-terms…Checked

What an hour of speech costsYes

Commercial use

Publishing the audio in monetized or business content starts on the $5 Starter plan; the free plan is restricted to personal use with credit to the vendor.

For commercial use (such as monetized content or business purposes), please subscribe to our starter plan and above.

minimax.io/audio/subscribeChecked

The voicesYes

Stock voices

The vendor states more than 300 system voices share its speech models, and the studio groups them in a browsable voice library with language and style labels.

The speech models provide speech synthesis, voice cloning, and voice design, supporting 40 languages and 300+ system voices, with synchronous or…

platform.minimax.io/docs/api-reference/…Checked

The voicesYes

Languages

A numbered list of 40 languages is published, from Chinese and Cantonese through Tamil and Afrikaans, and every paid plan shows the same 40+ languages and accents.

On the page: Speech Model section, '40 supported languages' table

platform.minimax.io/docs/api-reference/…Checked

The voicesNo

Re-voice your recording

The vendor enumerates four speech capabilities and re-voicing an existing recording is not among them; the studio offers a voice isolator, not a voice converter.

Four capabilities share the models above: 1. Synchronous speech synthesis (T2A) ... 2. Asynchronous long-text synthesis ... 3. Voice cloning ... 4.…

platform.minimax.io/docs/api-reference/…Checked

The voicesYes

Emotion, pace and emphasis

Speed, volume and pitch are adjustable, emotion can be set per passage, and pauses are written inline as markers measured in seconds.

Pause control: You can customize speech pauses by adding markers in the form <#x#>, where x is the pause duration in seconds. Valid range: [0.01,…

platform.minimax.io/docs/api-reference/…Checked

The voicesYes

Pronunciation fixes

A pronunciation dictionary is sent with the request, and single words can also be overridden inline with phonetic symbols, but the vendor documents no project-wide saved glossary.

Inline pronunciation: Wrap Mandarin Pinyin (with tone number 1-5) or IPA symbols or Cantonese Jyutping (with tone number 1-6) in half-width…

platform.minimax.io/docs/api-reference/…Checked

The voicesPartly

Long scripts and chapters

The plan table caps long text support on the free plan and makes it unlimited from Starter up, while the asynchronous interface takes up to a million characters in one go.

Asynchronous long-text synthesis: up to 1 million characters per request, ideal for entire books; supports sentence-level timestamps (subtitles).

platform.minimax.io/docs/api-reference/…Checked

The voicesYes

Audio you get

Downloads come as MP3, WAV, FLAC or raw PCM, with sample rates up to 44.1 kHz and MP3 bitrates up to 256 kbps.

Specifies the format of the generated audio. Supported values: [mp3, pcm, flac, wav, pcmu_raw, pcmu_wav, opus].

platform.minimax.io/docs/api-reference/…Checked

Cloning a voiceYes

Instant cloning

Ten seconds of recording is enough to clone a voice, and even the free plan carries three voice slots for cloned or designed voices.

Experience instant voice magic with just 10 seconds of audio input!

minimax.io/audioChecked

Cloning a voiceNo

Professional cloning

The published voice management rate card lists exactly two ways to make a voice, rapid cloning and voice design, with no longer-recording high-fidelity training tier.

On the page: Audio section, Voice Management table

platform.minimax.io/docs/pricing/overvi…Checked

Cloning a voicePartly

Consent of the voice’s owner

The vendor requires identity verification before a voice can be cloned, but states no rule that the voice's owner must agree, beyond a general ban on impersonation.

Voice cloning: upload the audio to clone to get a file_id (optionally upload sample audio to improve quality), then call the cloning API to produce a…

platform.minimax.io/docs/api-reference/…Checked

Dubbing and translationNo

Video dubbing

The product makes speech from text and clones voices; the vendor's own list of speech capabilities contains no video upload or translated re-voicing step.

On the page: Speech Model section, API Overview list

platform.minimax.io/docs/api-reference/…Checked

Dubbing and translationPartly

Captions and subtitles

Speech can be returned with sentence-level or word-level timestamps as a subtitle file, and the separate transcription service exports SRT and VTT, but translated subtitles are not offered.

Controls whether subtitles are enabled. Default is false. Available for models: speech-2.8-hd, speech-2.8-turbo, speech-2.6-hd, speech-2.6-turbo

platform.minimax.io/docs/api-reference/…Checked

Working with itYes

API

The same speech models are sold through a developer platform with HTTP and WebSocket synthesis, cloning and voice design, on its own pay-as-you-go balance.

Synchronous speech synthesis (T2A): real-time text-to-speech, up to 10,000 characters per request; 300+ system and cloned voices, adjustable volume /…

platform.minimax.io/docs/api-reference/…Checked

Working with itYes

API price

The developer rate card is public: $100 per million characters for the HD model and $60 for Turbo, with cloned and designed voices charged once each.

On the page: Audio section, Text-to-Speech table

platform.minimax.io/docs/pricing/overvi…Checked

Working with itYes

Platforms

web: the studio runs in the browser at minimax.io/audio, and the terms it is sold under cover websites and software applications without naming a desktop or phone app for it.

The following terms and conditions ("Terms of Use") apply to your access to and use of our websites and software applications, including any content,…

minimax.io/audio/doc/terms-of-service.h…Checked

Working with itYes

Works inside

Official Model Context Protocol servers put speech synthesis and voice cloning inside AI coding and agent tools, though no plug-in for Canva, Adobe or WordPress is documented.

MiniMax provides official Model Context Protocol (MCP) server implementations ... Both support speech synthesis, voice cloning, video generation, and…

platform.minimax.io/docs/api-reference/…Checked

Labelling and rightsYes

Who owns the audio

The vendor disclaims ownership of what you generate, but you also grant it a perpetual worldwide license over that content which survives termination.

We do not claim ownership of User Contributions or User Generated Content.

minimax.io/audio/doc/terms-of-service.h…Checked

Trying it, starting and getting helpYes

Free plan or trial

A standing free plan gives 10,000 credits a month as a gift, roughly twelve minutes of the HD model and three voice slots, with slower generation.

Free $0 /month 10k Credits Monthly Gift ~12 mins of HD model ~33 Songs generated 3 Voice Slots

minimax.io/audio/subscribeChecked

Trying it, starting and getting helpPartly

What the free plan leaves on your audio

The free plan's audio carries no watermark the vendor describes, but it is personal use only and you are asked to credit the vendor when you post it.

For personal use (like social media posts or non-profit presentations), you're welcome to use the audio freely - just remember to include our logo or…

minimax.io/audio/subscribeChecked

Trying it, starting and getting helpYes

Still sold

live: plans are on sale today with the new Speech 2.8 and Music 3.0 models headlined on the studio and the plan page.

Create Lifelike Audio with a Great Plan! Choose your plan, Top Up Credits, or Use Code

minimax.io/audio/subscribeChecked

Trying it, starting and getting helpYes

Sign up without a sales call

Every published plan is bought online by card through the vendor's payment processor, with tax handled at checkout and no sales call.

The name, address, and VAT or Tax ID displayed on your invoice are based on the details in your Stripe payment information.

minimax.io/audio/doc/paid-service-terms…Checked

Trying it, starting and getting helpYes

Contract and cancelling

Subscriptions are monthly or yearly and can be canceled at any point in the cycle online, but fees already paid are not refunded.

You may cancel a recurring subscription at any time, but if you cancel before the end of a subscription period, we will not refund any subscription…

minimax.io/audio/subscribeChecked

Trying it, starting and getting helpPartly

Support

Support is by email only for the studio, with a separate address and a Discord for the developer platform; no hours or chat are published.

If you have any questions regarding your bill, please contact us at contact.audio@minimax.io

minimax.io/audio/subscribeChecked

MiniMax Audio alternatives

All 9 alternatives

Is this yours?

Claim this listing for free to see who reads it and mark it as yours. You confirm it with an email at minimax.io.

Report an error if a fact is wrong.