RNNoise
RNNoise is a free, open-source noise suppression library based on a recurrent neural network.
What RNNoise is
RNNoise is a free, open-source noise suppression library based on a recurrent neural network. The project is intended for use as a library and includes a command-line demo for processing audio. The demo works with mono, 16-bit raw PCM at 48 kHz and uses raw input and output files rather than WAV files. Building it follows an autotools sequence involving autogen.sh, configure, and make. For CPU optimization, the project recommends AVX2 and describes AVX2 or SSE4.1 runtime CPU detection as build options. RNNoise includes a default model and can load a custom model through its API from a file, filename, or memory buffer. An alternative smaller model with increased sparsity is also described. Training models requires clean speech and noise data sampled at 48 kHz in 16-bit PCM format. The repository identifies the license as BSD-3-Clause and lists Linux as the platform. Its README characterizes the GitHub repository as a convenience copy and points to GitLab for the latest source. Xiph directs people to mailing lists or IRC channels for contact.
Good
- Free and open source under BSD-3-Clause
- Includes a command-line demo
- Supports custom models through its API
- Offers AVX2 and SSE4.1 detection build options
Watch out
- Listed platform is Linux
- Demo uses raw PCM, not WAV files
- Training data must be 48 kHz, 16-bit PCM
Who it suits
RNNoise suits developers who want a Linux noise suppression library or command-line demo and can work with raw PCM audio. It may also suit those who need to load or train models using the specified audio format.
Verdict
RNNoise provides a library-oriented noise suppression option with a demo and model-loading choices. Its raw PCM input requirements and Linux listing are key considerations.
Plans and prices
From RNNoise's own pricing page, checked 4 Oct 2001.
| Plan | Price | What it gives you |
|---|---|---|
| RNNoise | Free | BSD-3-Clause license · no paid plans stated |
The facts
Each one read from RNNoise's own pages, with the date it was checked.
What it does
- Build
- The README gives an autotools build sequence using autogen.sh, configure, and make.checked 4 Oct 2001
- CPU optimization
- The project recommends AVX2 and describes AVX2 or SSE4.1 runtime CPU detection as build options.checked 4 Oct 2001
- Distribution
- The README says the GitHub repository is a convenience copy and identifies GitLab as the location of the latest source.checked 4 Oct 2001
- Input format
- The demo processes mono 16-bit raw PCM at 48 kHz and its input and output are raw files, not WAV.checked 4 Oct 2001
- License
- The repository identifies its license as BSD-3-Clause.checked 4 Oct 2001
- Models
- RNNoise can use its included default model or load a custom model from a file, filename, or memory buffer through its API.checked 4 Oct 2001
- Purpose
- RNNoise is a noise suppression library based on a recurrent neural network.checked 4 Oct 2001
- Security
- The opened project pages state a software license but do not describe security certifications or compliance.checked 4 Oct 2001
- Smaller model
- The README describes an alternative little model with increased sparsity.checked 4 Oct 2001
- Support
- Xiph directs people to its mailing lists or IRC channels for contacting the foundation and community.checked 4 Oct 2001
- Training
- Model training requires clean speech and noise data sampled at 48 kHz in 16-bit PCM format.checked 4 Oct 2001
- Use
- The project is intended to be used as a library and includes a command-line demo.checked 4 Oct 2001
Alternatives to RNNoise
- 1
Krisp AI Meeting Assistant
AndroidiOSExtensionmacOS7.8From $8 a month - 2
DeepFilterNet
Free planOpen sourceAPILinuxmacOSSelf-hosted7.5Free plan - 3
DeskBroadcast
Free planmacOS7.3$79, paid once - 4
Kleervox
Free planWebLinuxWindows6.9Free plan - 5
Waves Curves AQ
macOSWindows6.8From $20.83 a month - 6
Noise Blocker
Free planWindows6.6$30, paid once