abab-music-1

abab-music-1 is a large-scale, multimodal AI music generation model developed by MiniMax. designed for speed, fIDElity, and creative flexibility, it enables users to generate complete musical works—including pure instrumentals, acapella vocals, and fully produced tracks with Synchronized accompaniment and singing—in a single inference step. Unlike modular tools requiring separate stem generation or post-processing, abab-music-1 handles composition, arrangement, and perfORMance synthesis holistically, significantly reducing iteration time for creators.

What it does

abab-music-1 transforms descriptive Prompts—such as genre, mood, instrumentation, tempo, or even reference audio snippets—into coherent, stylistically consistent musical ouTPUts. It supports both text-to-music and audio-conditioned generation, allowing users to seed ideas with melodies, hums, or existing CLIPs. The model natively produces stereo audio at production-ready quality, with support for variable durations and dynamic structure control (e.g., verse–chorus transitions).

Key capabilities

• End-to-end music synthesis: Generates full compositions—including hArmonies, rhythm SECtions, lead instruments, and expressive vocals—without chAIning multiple models.
• multimodal input handling: Accepts natural language descriptions and raw audio inputs to guide style, timbre, and phrASIng.
• Vocal + accompaniment co-generation: Produces synchronized singing and backing tracks in one pass, preserving timing, key, and EMOtional intent.
• Efficient architecture: LeveRAGes Linear Attention and Mixture-of-Experts (MoE) design to scale efficiently across long-form sequences while maintaining Low Latency.
• Production-ready output: Delivers stereo WAV files suitable for editing, licensing, or direct integration into video, games, or podcasts.

Who it is for

abab-music-1 serves a broad spectrum of users: Professional composers seeking rAPId mockups; indie filmmakers needing custom score drafts; content creators requiring unique background music for videos or streams; game developers prototyping adaptive audio; and hobbyists exploring music-making without formal training. Its accessibility makes it especially valuable for non-musicians who rely on intuitive prompting rather than DAW expertise.

How to get started

IndiVidual users can access abab-music-1 via the Hailuo Music web interface. Developers and Enterprises integrate it through the MiniMax open platform API, which offers SDKs, documentation, and scalable Inference endpoints. No local installation is required—generation occurs entirely in the cloud. Input formats include plain text prompts and optional audio references; output is delivered as downloadable audio files.

Pricing and access caveat

Pricing may vary by usage tier—individuals often receive free credits or trial quotas, while commercial users typically subscribe to credit-based or volume-based plans. API access requires registration and may involve rate liMITs or APProval for high-throughput applications. As of now, no desktop application or offline mode is publicly available; all functionality depends on MiniMax’s Cloud Infrastructure. Users should verify current terms directly on the official platform, as offerings may evolve.