Method, device and system for generating a light control parameter for controlling light (20), e.g. ambient light, in response to an audio signal (A). The audio signal (A) is analyzed in order to extract a temporal evolution of semantic content of the audio signal (A). A light control parameter is then generated according to the temporal evolution of semantic content of the audio signal (A), either on-line or off-line. This enables generating light that is perceptually aligned with the audio signal, e.g. music. In a preferred embodiment spatial properties of the audio signal (A) are extracted, and separate light control parameters are generated for controlling light sources spatially positioned according to the spatial properties of the audio signal, e.g. for stereo or surround sound systems where light sources are positioned near each loudspeaker. Such embodiment enables spatial correspondence between light and audio. In another preferred embodiment semantically distinct temporal sections of a sequence of the audio signal (A) are determined, e.g. sections of a piece of music, and wherein the light control parameter is generated to reflect these sections of the audio signal (A). Such embodiment enables temporal changes in light that follow perceived changes in e.g. a piece of music. Applications of the described method is e.g. within home entertainment systems, theatres, jukeboxes, bars, restaurants, cinemas and discotheques.