Minimax Music 2.5

Full-length song generation with vocals, lyrics, and rich instrumentation from text guidance.

Model Overview

Minimax Music 2.5 is a text-guided music generation model that creates complete songs with vocal and instrumental structure. It supports lyric-driven composition and style/mood prompting for production-ready draft outputs.

Best At

• Generating complete songs from lyrics and style prompts.
• Producing vocal-forward tracks with coherent arrangement sections.
• Iterating quickly on genre, mood, and storytelling direction.

Limitations / Not Good At

• Output quality is highly dependent on lyric quality and stylistic specificity.
• Very niche production techniques may still require manual DAW post-processing.
• It is not designed for speech synthesis, transcription, or sound-effect-only workflows.

Ideal Use Cases

• Song ideation and rapid demo generation.
• Draft background tracks for videos, short films, and social content.
• Exploring multiple lyrical and style directions before full production.

Input & Output Format

• Input: lyrics (required), optional style prompt, and output controls (audio_format, bitrate, sample_rate).
• Output: Generated song audio file URI.

Performance Notes

• Best results come from structured lyrics and explicit style guidance.
• Audio format and quality parameters can be tuned for downstream editing or delivery.