Implementing an aduio effect processor with FPGA

OData support
Szántó Péter
Department of Measurement and Information Systems

Nowadays the variety of available audio effects on the market is huge.

There are integrated multieffects, single effect modules, etc.

All of these can take several physical forms.

Both analog and digital (HW - SW) implementations are existing.

There is a demand for increasing sampling frequency (96kHz or even 384kHz) as well as processing resolution (24 bits isn't uncommon) in order to achieve better sound quality.

Low latency (preferably below 10 ms) is another key requirement for real-time effect modules, in order to avoid perceptible artifacts.

This is especially important in setups with multiple effects connected in series.

Audio signal processing might be computationally intensive for some effects (e.g. Reverb).

Computing a cascaded effect chain on a shared computational resource (SW implemented effect on general purpose computer, of traditional DSP chip)

might violate maximal delay requirements resulting in unintended audible artifacts.

Leveraging the parallel computing power of FPGAs in suitable algorithms might be a good choice.

The FPGA market is continously growing, gaining more attention and increasing investment.

These factors together with evolving IC technology make it possible to increase the achievable frequency, the amount of both general FPGA logic and special resources

(e.g. SoC, DSP blocks). The price of the devices is decreasing, despite the increasing capabilities.

The aim of this work is to design a highly configurable IP based real-time multieffect system targeting FPGAs.

Configuration of the effect IPs can be done via SW.

After the configuration, the audio signal processing must be done by the effect IPs without the need for processor intervention.

All effect IP blocks are based on a common effect framework, making it easier to implement, scale and integrate the different effects.

The effect modules should be as platform independent as possible,

but there are tradeoffs between reusability and performance.

Some test platfrom specific parts are discussed as well.

The first section of the work is an introduction to the most common audio effects.

In later sections the selected effects are discussed in more detail. These are probably the most commonly used

Reverb, Distortion and Chorus.

The next section describes the designed multieffect system.

The last part is an evaluation of the design and discusses the possible improvements.


Please sign in to download the files of this thesis.