To present three-dimensional sound spaces with high accuracy, not only reproduction techniques but also those to capture sound space information are important. Because of the rapid recent progress of digital electronics, sound space measurement and reproduction can benefit from using numerous channels. In this paper, we introduce the approaches of our current research using multi-channel arrays of microphones and loudspeakers. About measurements, two types of microphone array systems with numerous microphones are examined: a surrounding-microphone array with a total of 157 microphones installed onto the ceiling and four walls of a room, and a spherical microphone array with more than one hundred digital microphones. The former is designed for the characterization of sound events by attributes such as the source position, source signal, directivity, etc. By extracting these attributes, we intend to enable the modification of sound space information according to the listener's preferences; namely, an 'editable sound field system.' The latter type of array is to capture sound space information at a certain point in space with high spatial resolution. Reproduction methods using numerous loudspeakers are also assessed. One promising method to re-create the captured information is the 'Ambisonics reproduction.' However, a crucial disadvantage of this technique is its limited ideal listening area. Therefore, we proposed a new method for Ambisonic decoding that can enlarge the listening area significantly while maintaining low construction errors. Results of evaluation using an irregular surrounding 157-loudspeaker array demonstrate that the proposed method significantly outperformers the conventional one.