Computation of Generalized Mutual Information from Multichannel Audio Data
The authors present a new method to extract the mutual information for data from any number of channels from either a discrete or continuous system. This generalized mutual information allows for the estimation of the average number of redundant bits in a vector measurement. Thus it provides insight into the information shared between all channels of the data. It may be used as a measure for the success of blind signal separation with multichannel audio. Several multichannel audio signals are separated using various ICA methods and the mutual information of each signal is computed and interpreted. It is also implemented as a contrast function in ICA for a new method of blind signal separation.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is temporarily free for AES members.