I'm now using the entire audio processing module from WebRTC as opposed to individual DSP algorithms pulled from there before. Seems to work better this way.
