Abstract
This paper proposes a new aliasing cancelation algorithm for the transition between non-aliased coding and transform coding with time domain aliasing cancelation (TDAC). It is effectively utilized for unified speech and audio coding (USAC) that was recently standardized by the Moving Picture Experts Group (MPEG). Since the USAC combines two coding methods with totally different structures, a special processing called forward aliasing cancelation (FAC) is needed at the transition region. Unlike the FAC algorithm embedded in the current standard, the proposed algorithm does not require additional bits to encode aliasing cancelation terms because it appropriately utilizes adjacent decoded samples. Consequently, around 5% of total bits are saved at 16- and 24-kbps operating modes in speech-like signals. The proposed algorithm is sophisticatedly integrated on the decoding module of the USAC common encoder (JAME) for performance verification, which follows the standard process exactly. Both objective and subjective experimental results confirm the feasibility of the proposed algorithm, especially for contents that require a high percentage of mode switching.
Highlights
Unified speech and audio coding (USAC; ISO/IEC 23003-3) standardized in early 2012 shows the best performance for speech, music, and mixed type of input signals [1]
2.2 Forward aliasing cancelation algorithm Since the USAC consists of two different types of coding methods, it is very important to handle the transition frame where the encoding mode is switched from frequency domain (FD) codec to time domain (TD) codec or vice versa
This paper proposes a new aliasing cancelation algorithm that does not need any additional bits while successfully removing the aliasing parts
Summary
Unified speech and audio coding (USAC; ISO/IEC 23003-3) standardized in early 2012 shows the best performance for speech, music, and mixed type of input signals [1]. In case of encoding signals with mixed characteristics, one of two coding modes is chosen depending on the characteristic of input contents This approach improves the quality of all types of contents, many problems occur at transition frames where mode switching is needed between entirely. The scheme introduced time domain weightings applicable as a post processing in the decoder in order to remove a look-ahead delay inevitable for a window transition from the long window to the short window. This is considered as an aliasing cancellation signal described in this paper.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have