Artificial Stereo Extension Based on Gaussian Mixture Model
In this paper an artificial stereo extension method is proposed to provide the stereophonic sound. The proposed method employs a minimum mean squared error (MMSE) estimator based on a Gaussian mixture model (GMM) to produce stereo signals from a mono signal. The performance of the proposed stereo extension method is evaluated using a multiple stimuli with a hidden reference and anchor (MUSHRA) test and compared with that of the parametric stereo method. It is shown from the test that the mean opinion score of the signals extended by the proposed stereo extension method is around 5% higher than that of the conventional stereo extension method based on inter-channel coherence (ICC).
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.