D
DeepFilterNet
Description
DeepFilterNet is an open noise suppression framework for speech, built for full-band 48 kHz audio and low computing cost. It can be run as a ready binary (deep-filter) on a WAV file, as a Python package (pip install deepfilternet) with a command-line tool and a Python API, or as a real-time plugin that cleans a microphone on Linux. Pretrained models of three generations are included, DeepFilterNet2 is the default. The binary currently handles 48 kHz WAV files. It works on Linux, macOS and Windows, a GPU is optional. The code is dual-licensed: MIT or Apache 2.0, at your choice. A demo is available on Hugging Face.
Pros and cons
Pros
- Free choice of MIT or Apache 2.0 license
- Works on Linux, macOS and Windows
- GPU is optional
- Ready binary, Python package and live microphone plugin
Cons
- The ready binary handles only 48 kHz WAV files
- Live microphone plugin is for Linux
- Supported spoken languages are not stated
Key facts
- License
- MIT или Apache 2.0 (на выбор)
- Commercial use
- Allowed
- Developer
- DeepFilterNet authors
- Checked on
- Oct 20, 2026
Editorial score
How strongly the AI Expertise editors recommend this open model for each of its tasks right now, on a 0–10 scale. The score combines output quality, capabilities, ease of launch, license openness and freshness.