Code for the interactive APA-based acoustic echo canceller is now available

One year ago, we experimented with the use of an interactive genetic algorithm (IGA) for evolving the parameters of an APA-based acoustic echo canceller. The results were positive, with a vast majority of users that preferred the result of our IGA approach to that of a standard GA based on a classical error measure:

Results of the subjective comparison

This experiment was detailed in two conference proceedings: User-driven quality enhancement for audio signal processing and Interactive quality enhancement in acoustic echo cancellation.

We have now released the code of our simulations as open-source software on BitBucket: https://bitbucket.org/ispamm/iga-audio/. We believe that interactive approaches are extremely interesting in the context of audio signal processing, and we hope that our code can be a good starting point for additional research in this area. For any information on the publications or on the code, please feel free to contact me.