2000 
Create Speech and Songs With MiniMax Audio
With MiniMax Audio, you can turn a script into speech or your lyrics into a full song. Type the text, choose one of 300+ voices and you'll get a ready voice-over. If you have lyrics instead, Music 3.0 builds a complete song of up to five minutes from them, so one tool covers both the voice and the soundtrack.
On Cabina.AI, the MiniMax AI audio generator works in the same chat as the other top AI models. You don't need a separate MiniMax subscription, because one Cabina.AI plan lets you try every model on the platform. You can start on the Free plan and check the token cost of each version before you spend anything.
Try MiniMax Audio Free on Cabina.AIWhat You Can Make With MiniMax Audio
Text-to-Speech From Any Script
Paste any script, and the MiniMax TTS model turns it into spoken audio. You pick a voice from the full built-in library, so each project can get a sound of its own.
Sample voice: Expressive Narrator (British, Male, Audiobook)
Speed, Volume and Pitch
Sliders let you set speed, volume and pitch until the speech matches the way you hear it in your head. For the audio file itself, you can also choose the sample rate and the bitrate, which start at 44.1 kHz and 256 kbps.
Voice Cloning From a Short Sample
MiniMax AI voice cloning needs only a 10-second sample. Record yourself once, and every new script you type can come out in your own voice, ready for narration or a voice-over.
Songs From Your Lyrics
Describe the style you want and add your lyrics. MiniMax AI music generation then returns a complete song with your words sung as vocals, so you can go from a text file to a finished track.
Verse, Chorus and Bridge Tags
Tags such as verse, chorus and bridge tell the MiniMax AI music generation tool where each part of the song goes. With Music 2.6, your prompt can also set the BPM, the key and how the mood builds.
Covers in a New Genre
Upload a track to get a MiniMax music cover in a new genre or with other instruments. The melody stays the same while everything around it changes, so a piano ballad can come back as a rock song.
What to Prepare Before You Generate
For a Cloned Voice
Before you start MiniMax TTS voice cloning, check that your recording fits these limits:
- File type:MP3, M4A or WAV
- Length:anywhere from 10 seconds to 5 minutes
- Size:20 MB or less
For a Song
The MiniMax AI music generator works from two pieces of text:
- Style prompt:up to 2,000 characters about genre, mood and instruments
- Lyrics:up to 3,500 characters, with section tags if you want them
If you want an instrumental, skip the lyrics. A style prompt alone is enough.
MiniMax Audio Pricing by Version and Plan
MiniMax TTS pricing is set in tokens per message, and the number of generations grows with your plan.
| Version | Tokens per message | Free $0.00 / month 50 tokens | AI $4.99 / month 450 tokens | Agent $19.99 / month 1,000 tokens | Agent Pro $49.99 / month 2,100 tokens |
|---|---|---|---|---|---|
| Speech | |||||
| Speech 2.5 Turbo | 17.76 | 5 | 25 | 56 | 118 |
| Speech 2.5 HD | 1.60 | 62 | 281 | 625 | 1,312 |
| Music | |||||
| For songs, MiniMax Audio pricing follows the same pattern. Music 3.0 shows a dash instead of a generation count on every plan. | |||||
| Music 2.6 | 15.00 | 6 | 30 | 66 | 140 |
| Music 3.0 | 2,400.00 | - | - | - | - |
If you budget by the month, your MiniMax subscription cost equals the price of the plan you pick.
Use MiniMax Audio Next to Other AI Models
Multi-Format mode keeps text, images, video and audio in one chat. You can sketch an idea with Gemini, turn it into a picture with Flux, animate it with Runway and then let the MiniMax Audio model voice it. Every step shares the same context, so you never lose the thread of the project.
Compare Mode shows answers from two AI models side by side. Send the same song idea to both, keep the better lyrics and only then generate the track. If you want to hear other voices, ElevenLabs and OpenAI text-to-speech are available too.
MiniMax Audio FAQ
Is MiniMax Audio Free to Use?
Yes, you can start on the Free plan without paying, so MiniMax pricing only matters once you outgrow its limits. The free tier has daily and overall limits on some models, and any paid plan removes them. Pay-as-You-Go lifts the limits too, for 100 days after your last payment. You're never charged for a request that gets no answer.
How Long Can a MiniMax Song Be?
A song made with Music 3.0 can run up to five minutes. That leaves room for an intro, several verses and choruses, a solo and an outro, which you mark with tags right in the lyrics.
Which Is Better for Voice Cloning, MiniMax or ElevenLabs?
It depends on the voice, the language and the sound you're after, so neither tool wins every test. With MiniMax Audio, AI voice cloning starts from one short recording. Since ElevenLabs is available here too, you can voice the same script with each and keep the result you like.
Can I Use the MiniMax Voice Cloning Tool on Someone Else's Voice?
Only with permission. Under MiniMax terms, you need full legal rights to any voice you submit, or approval from whoever holds those rights. Cloning your own voice is fine, since you keep the rights to it. For a friend, a client or an actor, get their consent first.
Which Languages Does MiniMax Audio Support?
MiniMax speech works in 40 languages and dialects. The list includes English, Spanish, French, Russian, Arabic, Hindi, Japanese, Korean, Chinese and Cantonese.
Try MiniMax Audio on Cabina.AI
Start free and turn your first script or set of lyrics into audio.
Start Free on Cabina.AI