Summary form only given. A complex sound source consists of several partial sound sources that all contribute to the total sound pressure. A method to separate these partial sound sources into separate time histories is based on inverse filtering of reciprocally measured transfer functions. The transfer functions are measured reciprocally between a number of fictitious monopoles on each partial source and measurement points distributed around the sound source. The method is divided into 5 steps: recording of sound pressure signals, measurement of transfer function, calculation of source strength matrix, calculation of filters and filtering of the recorded sound pressure signals. Correct estimations of the transfer functions are critical for inverse methods to work satisfactory. Normally the transfer functions in this case of studies are calculated as H1 because of the noise contribution to the responses. However, it has been suggested that inverse methods could benefit of using H2 instead. The objective of this investigation is to analyze the effect of selecting either H1 or H2 for the calculation of the transfer functions in case of auralization of separated time histories. For the experiments a complex sound source consisting of two separate cylinder heads with valve covers have been used. Each cylinder head with valve cover was treated as a partial source. The two partial sources were excited with two uncorrelated signals that could be controlled individually. By that, listening tests could be used to verify the authenticity of the separated source signals depending of the transfer functions were estimated as H1 or H2