iranroman
Conversion toolbox from MIC to FOA, possibly Binaural
Right now, we have several databases/simulated databases that have data in 4-channel MIC format, but need conversion to FOA. The conversion should be relatively straightforward:
A helper function that adds foa data from existing mic data in the room .pkl files would be extremely helpful for future work. Additional support that takes an HRTF file and generates a binaural output would be useful, but may be a separate issue.
I did some basic experiments with the above method, and the results were worse in terms of performance (~20% performance drop across all metrics). Ideally we can integrate the work from the original code used to compute the FOA data in TAU-SRIR, but the code is a lot of signal processing written in MatLab
For the record, the need for these encoders and our work in progress toward them has been highlighted again in #59
@beasteers, you mentioned that maybe you would be interested in helping with this issue. I recommend looking into the array2sh plugin from SPARTA.