Voice isolation does a great job with noise suppression when the user is holding the phone in hand (facetime use case). But when the phone is about 4-feet away from the user, voice isolation quality substantially drops and we are seeing that it is better to not use it.
Our use case demands that user mounts phone on tripod and sites approximately 4 feet away from camera. In this case we are seeing worst performance from voice isolation, presumably because of heavy signal processing and lower original signal to begin with.