The impementation of recursive filters using fixed-point 2's-complement arithmetic for high-fidelity audio is discussed. It is now widely accepted that 16/32-bit DSP chips are inadequate for many audio applications. The engineer new to the design and realization of recursive filters may believe that the straightforward use of 24/48(+ overflow)-bit DSP chips will ensure sufficient quality for audio applications. Unfortunately this is not always the case. Filter topologies based on direct form I are recommended for audio. Many filter designs require the use of error feedback schemes to reduce noise to an acceptable level. Some extremely critical frequency response design require the use of double-precision coefficients. Sometimes it is appropriate to use other filter topologies to reduce the noise floor or to allow the implementation of extremely critical frequency response designs.
Click to purchase paper 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 $20 for non-members, $5 for AES members and is free for E-Library subscribers.