MiniMax recently announced the official release of MiniMax Music 2.5. The core goal of this upgrade is to break the long-standing bottlenecks in "control" and "realism" in AI music, making Grammy-level professional creation accessible through technological iteration.

image.png

Compared to the previous generation, Music 2.5 has achieved breakthroughs in two dimensions:

  • Paragraph-level strong control (Direct the Detail): The model now supports full-paragraph tag control, precisely supporting 14 types of structural variations such as intro, bridge, and hook. Creators can control instrument configuration and emotional tension with tags, just like professional arrangers, rather than relying on luck for random generation.

  • Physical-level high fidelity (Define the Real): Through systematic optimization, Music 2.5 achieves continuous smooth vibrato, natural tremolo fluctuations, and multi-cavity resonance switching, significantly enhancing vocal vitality. In addition, the model has expanded over 100 instrument tone libraries and can automatically adjust mixing strategies based on style, ensuring that vocals and accompaniment remain clear even in dense scenes.

For the Chinese market, Music 2.5 has undergone deep customization, greatly optimizing the melodic logic of C-Pop and C-Rap, significantly reducing issues such as muffled words and unclear sounds. Even in bilingual scenarios, it maintains smooth transitions. Currently, MiniMax Music 2.5 has been launched simultaneously on its official website and API platform.

Product experience:

minimaxi.com/audio/music

API interface:

platform.minimaxi.com/docs/api-reference/music-generation

Key points:

  • 🎵 Full paragraph control: Supports 14 types of music structure tags, allowing AI creation to shift from "random generation" to "precise direction".

  • 🎤 Realistic vocal quality: Features delicate vibrato and tremolo expression, supporting natural collaboration and harmony levels in male and female duets.

  • 🎹 Professional-level mixing: Includes hundreds of instrument tones and has the ability to automatically adapt to styles, achieving studio-level audio delivery.

  • 🇨🇳 Deep Chinese market optimization: Systematically improves Chinese pronunciation and Chinese pop trends, adapting to various pop and rap scenarios.