White spaces are portions of the TV spectrum that are allocated but not used locally. If accurately detected, white spaces offer a valuable new opportunity for high speed wireless communications. We propose a new method for white space detection that allows a node to act locally, based on a centrally constructed model, and at low cost, while detecting more spectrum opportunities than best known approaches. We leverage two ideas: first, we demonstrate that low-cost spectrum monitoring hardware can offer good enough detection capabilities. Second, we develop a model that combines locallymeasured signal features and location to more efficiently detect white space availability. We incorporate these ideas into the design, implementation, and evaluation of a complete system we call Waldo. We deploy Waldo on laptop in Atlanta metropolitan area in the US covering 700 km2 showing that using signal features in addition to location can improve detection accuracy by up to 10x for some channels. We also deploy Waldo on an Android smartphone, demonstrating the feasibility of real-time white space detection with efficient use of smartphone resources.