In today's competitive consumer audio market the Advanced Audio Coding (AAC) format has quickly become a must-have technology with its adoption on the Internet, in digital radio, digital television and home theatre. Compression using AAC retains high audio quality even at low bit rates. One reason for this effectiveness is the use of Huffman variable length coding to represent frequency domain information. However, this requires to perform the relatively complex task of Huffman decoding in the audio decoder, which is typically very sensitive to cost and processor speed requirements. Furthermore, encoders can sometimes create worst-case scenarios consisting of very long code words, even when unnecessary. Thus, one needs to optimize the Huffman decoding for these worst-case scenarios without giving up average performance. This paper discusses various methods for Huffman decoding, their inherent implementation tradeoffs on a DSP platform and proposes improvements that are specific to the Huffman codebooks used in AAC.
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.