New AI model keeps music recognizable when shifted in time or pitch

Original: Equivariant Music Transformer

Source: arxiv.org ↗