DeepFilterNet is a speech enhancement framework designed to suppress unwanted noise in full-band audio. It combines deep learning with digital signal processing to improve speech intelligibility while keeping computational requirements relatively low.
The software can be used to process recorded speech or to reduce background noise from a live microphone. Its architecture combines Rust components for signal processing with Python tools for running, evaluating and training neural networks.
DeepFilterNet is aimed at several different audiences. Users can enhance noisy recordings with a command-line application, while developers and researchers can work directly with its processing libraries, model implementations and training tools.
The framework also provides real-time integration suitable for Linux audio systems. This makes it useful for speech communication as well as offline processing and research.
This is free and open source software.
Key Features
- Neural deep filtering provides noise suppression for full-band speech audio at 48 kHz.
- Combines neural-network estimation with frequency-domain signal processing.
- Designed to provide speech enhancement with relatively low computational complexity.
- Command-line audio processing can enhance recorded WAV files.
- Supports processing multiple audio files through command-line workflows.
- Pretrained neural models provide speech enhancement without requiring users to train their own networks.
- DeepFilterNet model families support different approaches to speech enhancement.
- Optional post-filtering can apply additional attenuation in noisy passages.
- Delay compensation is available for processing workflows requiring timing alignment.
- LADSPA plugin enables real-time microphone noise suppression through PipeWire filter chains.
- Python interfaces allow the enhancement system to be incorporated into other applications.
- Training and evaluation components support speech enhancement research.
- Rust signal processing components handle operations including spectral analysis and reconstruction.
- Supports data loading, augmentation and model evaluation workflows.
- Cross-platform support – runs on Linux, macOS, and Windows.
Website: github.com/rikorose/deepfilternet
Support:
Developer: Rikorose
License: Apache License 2.0 and MIT License
DeepFilterNet is written in Python. Learn Python with our recommended free books and free tutorials.
Related Software
| Noise Suppression Tools | |
|---|---|
| NoiseTorch-ng | Creates a virtual microphone that suppresses noise |
| noise-suppression-for-voice | Noise suppression plugin based on RNNoise |
| Easy Effects | GTK4 audio manipulation which includes a range of tools |
| noise-repellent | Suite of lv2 plugins for noise reduction |
| DeepFilterNet | Low-complexity speech enhancement framework |
| BigLinux Microphone | Improve spoken audio captured from a microphone |
Read our verdict in the software roundup.
 Explore our carefully curated directory of recommended free and open source software, covering every major software category.The directory forms part of our extensive collection of articles for Linux enthusiasts. It includes hundreds of detailed reviews, together with free and open source alternatives to proprietary software from companies such as Google, Microsoft, Apple, Adobe, IBM, Cisco, Oracle, and Autodesk. LinuxLinks also covers interesting projects worth exploring, Linux-compatible hardware, free programming books and tutorials, and much more. Know a useful free and open source Linux application that we haven’t covered? Tell us about it using our submission form. |
Please read our Comment Policy before commenting.