Repository navigation
This release improves noise suppression in frames where near-end speech is present by attenuating individual frequency bins that only contain noise energy. The model size is larger for this release which allows the fine-grained targeting of individual bins.
Fixes: #6
Full Changelog: v1.2...v1.3