jneem/nnnoiseless
Recurrent neural network for audio noise reduction
What it solves
It provides a way to suppress noise from audio files, allowing users to clean up recordings by removing unwanted background sounds.
How it works
The project is a Rust port of the RNNoise C library. It uses a recurrent neural network (RNN) to identify and filter out noise. It can be used as a library in other Rust projects or as a standalone command-line tool that processes WAV or RAW PCM files.
Who it’s for
Developers looking for a high-performance audio noise suppression library in Rust, or users who need a simple command-line tool to remove noise from audio files.
Highlights
- Custom Models: Users can replace the built-in neural network with their own trained weights to target specific types of noise.
- C API: It can be installed as a library compatible with C, making it accessible to other languages.
- Flexible Loading: Supports loading weights from files or static bytes.
- Conversion Tool: Includes a script to convert existing RNNoise text-based weights into the binary format used by the project.
Related
- Project
- Project
- Project
- Project