The channel based 3D audio can be compressed to a down-mix signal with side information. In this paper the inter-channel transfer functions (ITF) are estimated through training over fitting convolutional neural networks (CNN) on a specific frame. Perfectly reconstructing the original channel and keeping the spatial cues the same is set as the target of the estimation. By taking this approach, more accurate spatial cues are maintained. The subjective evaluation experiments were carried out on stereo signals to evaluate the proposed method.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=19852
(349KB)
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 Join the AES. If you need to check your member status, login to the Member Portal.
