A singular matrix is a square matrix that does not have an inverse, characterized by a determinant of zero. This property implies that its rows or columns are linearly dependent, meaning they do not span the entire vector space.