Available now

MiniMax Music 3: text to music

By MiniMax

MiniMax Music 3 sings your lyrics over an arrangement you describe, in tracks of up to 300 seconds.

Model ID
Maker
MiniMax

Input

Fields marked * are required. Review the price before generating.

Music description: style, mood, vocals, instrumentation and arrangement. For precise control use a Structured Caption with global metadata (genre, BPM, key, emotional progression), vocal details, and a section-by-section arrangement.

Example prompt. Edit it or write your own.

Upper bound on the generated audio length in seconds. The model may stop earlier; the actual duration is returned in the output.

Additional settings

Random seed for reproducibility. If not provided, a random seed will be used.

Number of flow-matching Euler steps per 8-second denoising chunk. More steps improve quality at the cost of speed.

Classifier-free guidance scale of the flow-matching stage.

Estimated total · 60 s$0.19

You sign up first. Your settings are kept.

About this model

Overview

Describe the style, mood, vocals, instruments and arrangement, then add the lyrics with structure tags such as [verse], [pre-chorus], [chorus] and [bridge]. MiniMax Music 3 returns a sung track; set a maximum length of up to 300 seconds, and the song may end before it.

For tighter control, write the description as a structured caption: genre, BPM, key and how the emotion should build, then the vocal details and the arrangement section by section. A seed repeats a result, and more steps trade time for detail.

What you can build

  • Full songs with your own lyrics
  • Theme songs with a planned arrangement
  • Instrumental breaks and solos inside a sung track
  • Several takes of one song with different seeds

Prompting tips

  • Put every structure tag on its own line; text on the same line as a tag is dropped.
  • Give BPM and key in the description when the song must match other material.
  • Use [instrumental] or [solo] in the lyrics where you want a section without singing.

What to check

Listen to the complete audio and check pronunciation, lyrics, timing and sound quality.

Inputs

  • Prompt (required): Music description: style, mood, vocals, instrumentation and arrangement. For precise control use a Structured Caption with global metadata (genre, BPM, key, emotional progression), vocal details, and a section-by-section arrangement.
  • Lyrics (required): The lyrics to sing. Structure tags such as [intro], [verse], [pre-chorus], [chorus], [post-chorus], [bridge], [instrumental], [solo] and [outro] must each be on their own line; text on the same line as a leading tag is dropped by the model's input contract.
  • Duration (optional): Upper bound on the generated audio length in seconds. The model may stop earlier; the actual duration is returned in the output. From 1 to 300.
  • Seed (optional): Random seed for reproducibility. If not provided, a random seed will be used.
  • Number of Inference Steps (optional): Number of flow-matching Euler steps per 8-second denoising chunk. More steps improve quality at the cost of speed. From 1 to 100.
  • Guidance Scale (optional): Classifier-free guidance scale of the flow-matching stage. From 0 to 20.

Output and limits

  • Length: 1 second to 5 minutes.
  • Sung or spoken words can come out unclear or differ from your text.
  • Exact length, tempo and key are not guaranteed; listen to the whole result before you use it.
  • Prompts that name real artists or protected songs can be refused.
  • Some prompts or outputs can be refused by safety checks.

Price guide

  • per second: $0.0032 / second
  • The total for your settings is rounded up to the next cent.