Encoder implementations of MPEG Advanced Audio Coding and Layer-3 (mp3) on 32-bit or 16-bit fixed-point processors are challenging due to the fact that the usable word length is restricted to 32 bits if low processing power is required. This paper describes the modifications and optimizations that had to be applied to the algorithms of these audio encoders to make a true fixed-point implementation on a 32-bit or 16-bit device possible. The implementation had to be done without using floating-point emulations or even 64-bit values for the signal energies and thresholds in the psychoacoustic model of the encoder. At the same time high encoding quality and speed were required. Memory and processing power requirements on various platforms as well as results from a subjective listening test will be presented.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=12419
(117KB)
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.
