Port calc_vnr_pred to Avona

[shuchitak]

calc_vnr_pred() needs to be ported to Avona. Once we have this, we can work on #255 and test keyword performance when passing VNR instead of VAD output to AGC.

[shuchitak]

I have it implemented here and need to add a unit test for it before starting #255

[shuchitak]

https://github.com/xmos/py_vnr/pull/39

[shuchitak]

I have added the calc_vnr_pred function in the IC module and a unit test testing avona implementation against the py_ic model in https://github.com/shuchitak/sw_avona-1/tree/feature/calc_vnr_pred

The current challenge is to get the IC and stage_b tests that make use of the CFFI framework passing. Now that vnr is a dependency of IC, even though everything still compiles on x86, extracting the predefs the way it is done here is tricky for vnr due to the number of header files in the tflite micro library. I need to think more about this, so I've decided to increase the size label and story points for this issue.

[shuchitak]

#293

[shuchitak]

Merged #293

The calc_vnr_pred() version in py_ic hash 3826a6919b6a54f48b8ad72a06cc34ea6d4474da is ported to sw_avona. Since py_ic is being actively developed, this function will change, and will need updating in Avona. Keeping this issue open to track further changes to py_ic calc_vnr_pred()

[shuchitak]

calc_vnr_pred() ported to Avona is compatible with py_ic v2.0.0. Closing this issue now.