A method and apparatus to capture sound produced by an audio device including a sound source and a loudspeaker. The apparatus includes a microphone support coupleable to the loudspeaker and a microphone mounted on the support, the microphone being locatable within a volume surrounded by a diaphragm of the loudspeaker and being arranged to detect pressure waves caused by movement of a vibrating element of the loudspeaker. Aspects and embodiments provide an element to capture sounds from a loudspeaker in a manner which is compact and which ameliorates disadvantages associated with alternative arrangements since a greater core sound to ambient sound ratio may be captured.