Example
- Original
- Enhanced
Summary
Currently there are not many wide-range, pre-trained, and effective noise suppressor programs for speech that can be used easily. NSNet2 is a deep learning artificial recurrent neural network used for background noise reduction in speech audio files. Microsoft (Research) originally released NSNet2 as an updated comparision baseline for their annual Deep Noise Suppression challenge, but it is inconvient to use and is not suited for real-time processing. NSNet2 just needed to be converted to a version that more people looking for additional audio filters for recorded speech could utilize. DNS‑NSNet2 is a user friendly conversion of the NSNet2.
Features
- Simple and Straightforward Program Usage
- Creates Equivalent Noise Suppressed Audio Wave (.wav) File from Input Wave File
- Much Lower Memory (RAM) Usage while Maintaining the Original NSNet2 Processing Speed
- No reliance on math or matrix software libraries for repeating calculations
- Code Modification is Friendly due to Explanations and Low Code Complexity