MuLaCover
An AI cover and music remixing model that supports reference audio, lyrics, MIDI, and controllable musical styles
Note: Please open in the PC Client. Go to download >>>
Features
Supported Languages
System Requirements
25GB+ storage recommended.
macOS 15+: M-series chips required, 32GB RAM recommended.
Windows 10/11 64-bit: 16GB RAM recommended. NVIDIA GPU with 12GB+ VRAM required.
Note: For NVIDIA GPUs, install a newer driver.Screenshots
Introduction
License: The official model weights are licensed under CC BY-NC 4.0 and are limited to non-commercial use; commercial use requires additional authorization from MuLa Labs.
Overview
MuLaCover is an AI music generation model developed by MuLa Labs, the music and audio research team under Vera Praxis, designed primarily for AI cover songs, music adaptation, and remixing.
Rather than being a simple text-to-music system, it can combine reference audio, lyrics, melody MIDI, chord MIDI, and style prompts to generate new versions of existing music with different styles and arrangements.
Key Features
- AI Cover: Generate new versions based on existing songs.
- Style Transformation: Control genres, instruments, moods, and other musical characteristics.
- Lyric Control: Generate music based on new lyrics while using existing musical references.
- MIDI Conditioning: Use melody and chord MIDI as inputs for music generation.
- Musical Structure Control: Use symbolic information such as melody and harmony to guide generation, rather than relying solely on text prompts.
Underlying Technology
MuLaCover is a roughly 4B-parameter music generation model that uses symbolic musical control, incorporating melody, harmony, text conditions, and reference audio into the generation process.
Its complete workflow also integrates components such as HeartCodec, Qwen3-Embedding-0.6B, YourMT3, and ChordNet.
Use Cases
MuLaCover is suitable for musicians, AI music creators, and researchers working on:
- AI cover songs
- Music remixing
- Style transformation
- Arrangement exploration
- Music generation research
- Rapid music demos from MIDI