Acoustic beamformers enhance a signal of interest located at a discrete point in space using microphone array processing. Pre-measured array transfer functions (ATFs) recorded on a user or on the head of a dummy result in beampatterns that are maximally directive and perform well across different room types. However, such beampatterns do not suppress noise uniformly in each individual direction. When the beam is steered dynamically, there is a perceptually degrading and distracting change in noise level.

This disclosure describes techniques to jointly optimize a beampattern to enhance the sound coming from a given direction while maintaining a constant suppression of sounds coming from other directions; to maintain a constant suppression for each given beam-steering direction; and to relieve constraints on beamformer optimization such that the variation in beamformer sidelobes between steering directions is perceptually minimal.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.