Save a voice you can use again

Create a reusable voice from a clean sample, name it clearly, and use it for later MiniMax speech takes.

A recurring narrator gives a series a familiar sound. Instead of picking a different voice for every script, create a reusable one from a recording you have permission to use.

The quality of that short source matters more than a clever preview sentence.

Record one speaker clearly

Open Audio and choose Your voices. The current cloning workflow uses a 10–30 second MP3 or PCM WAV sample.

Choose a recording with one speaker, natural delivery, and little background noise. Skip music, overlapping voices, heavy room echo, and clipped audio. Listen to the entire sample before uploading.

Use only a voice you are authorized to clone and confirm the permission required by the page.

Give the voice a useful name

Choose a library name you can recognize later, such as “Workshop narrator / calm,” rather than “Voice 2.” Add an optional preview script that includes a few words you actually expect to use.

Try:

Welcome back. Today we are turning one clear idea into an image, then giving that image a little movement.

Review the displayed price before confirming. One clone includes its preview; later speech generation is a separate paid action.

Listen before making it your default

Check the preview for recognizable character, clean pronunciation, and a natural pace. If the voice sounds unstable, inspect the source sample before repeating the cloning step.

A useful saved voice should work beyond one flattering sentence. Try a representative short script when you are ready to create a real take.

Reuse it in the compatible workflow

Return to Text to speech, select a compatible MiniMax model, and choose the saved voice where available. Review the new script and its own price before generating.

A saved MiniMax voice is not a universal voice setting for every model or tool. Keep the destination compatible, and download the accepted takes you want to retain.

Your first reusable voice: a clear sample, a recognizable library name, and a preview that sounds like the narrator your project needs.

Next: write a stronger spoken script.