One of the key components in current multichannel upmixing techniques is identification and extraction of ambience from original stereo recordings. This paper describes correlation-based ambience extraction within a time-frequency analysis-synthesis framework. Two new estimators for the time- and frequencydependent amount of ambience in the input channels are analytically derived. These estimators are discussed in relationship to two other algorithms from the literature and evaluated with simulations. It is also shown that the time constant used in a recursive correlation computation is an important factor in determining the performance of the algorithms. Short-time correlation estimates are typically biased such that the amount of ambience is underestimated.
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 free for AES members and E-Library subscribers.