We address the task of separation of music and effects from dialogs in film or television soundtracks. This is of interest for film studios wanting to release films in new, previously unavailable languages when the original separated music and effects track is not available. For this purpose, we propose several methods for common signal extraction from a set of soundtracks in different languages, which are multichannel extensions of previous methods for center signal extraction from stereo signals. The proposed methods are simple, effective, and have an intuitive geometrical interpretation. Experiments show that the proposed methods improve the results provided by our previously proposed methods based on basic filtering techniques.