Brief

Researchers have developed LiveBand, a novel system capable of generating high-fidelity music accompaniments in real-time to live audio input. This system adheres to strict causal constraints, meaning it produces music without needing to look ahead into future audio frames. LiveBand utilizes a causal transformer generator trained within the latent space of an audio autoencoder, employing adversarial supervision. The system demonstrates improved performance over existing methods in audio quality, beat alignment, and mix adherence, while operating efficiently on consumer hardware. AI