An electromagnetic transducer, such as an audio speaker, having a voicecoil disposed within a magnetic gap between a pair of magnetic arrays, e.g., Halbach arrays, is disclosed. In an example, the paired Halbach arrays include vertically-poled magnets to direct magnetic flux across the magnetic gap orthogonal to electrical current carried by a planar winding of the voicecoil. Accordingly, a Lorentz force may drive an oscillational mass, e.g., a speaker diaphragm, in a longitudinal direction orthogonal to the magnetic flux and the electrical current to generate vibration or sound. Other embodiments are also described and claimed.