A note on generic splitting of quadratic forms