An audio reproduction method comprises providing equal loudness response data defining a dependency between sound pressure level and frequency, resulting in sounds of equal loudness, including for sounds of a notional negative loudness; generating a required spectral weighting profile from the response data to represent a difference in frequency response between a first audio transducer and a second, different, audio transducer; applying a spectral weighting to an audio signal in dependence upon the derived frequency response; and reproducing the spectrally weighted audio signal using the second audio transducer at an audio reproduction level. In some examples, the generating step comprises: mapping a difference between the first audio transducer and the second audio transducer to a mapped difference in audio level; deriving a frequency response dependent upon a difference between: the response data applicable to the audio reproduction level; and the response data applicable to the audio reproduction level as modified by the mapped difference in audio level.