If you think about it, when you process ambisonics material through a reverberation algorithm, you don't have any control on how individual sounds from the recording are localized within the virtual space.
I'm brainstorming here and I might miss some critical detail, but hear my though.
Route the first (W) channel of the FOA (B-Format) material, to a reverb that can offer separate facilities for early and late reflections. Use only the late reflections as your material already has both naturally, so you only add the late diffused character of the virtual location and you don't mess with the localization of the real sources you already recorded and have the natural early reflections that help our brain localize them.
It's a balanced choice between keeping localization resolution and blending the virtual acoustic character you want in your compositions.
To further this philosophy, you can add mono sources on that composition by passing them through the IEM RoomEncoder, and then treat the result as it was coming from a recording, passing them through the same pipeline I describe above.
How about that?
|