Text this: Can all variations within the unified mask-based beamformer framework achieve identical peak extraction performance?