HushMic
What it is
HushMic is free, open-source Linux software that creates a system-wide virtual microphone with real-time noise suppression. It uses the DPDFNet model to reduce keyboard sounds, fan noise and background chatter from microphone audio. Processing runs on the CPU, and the maker says audio remains on the machine rather than being uploaded. The project reports 80 ms of processing latency, in addition to PipeWire buffering. Users can switch among suppression, bypass, mute and turning off the virtual microphone through the app or command-line controls. A microphone test shows raw and cleaned audio side by side and records a 10-second sample for playback. Discord, TeamSpeak, OBS and browsers can select the virtual microphone as an input. Global shortcuts support mute, bypass, push to talk and push to mute, subject to desktop and portal support. HushMic requires Linux on x86-64 with PipeWire and WirePlumber; PulseAudio applications also need pipewire-pulse. Installation options include Debian or Ubuntu, Arch, Fedora, NixOS and AppImage.
For and against
For
- Runs audio processing locally on the CPU
- Reduces keyboard, fan and background chatter noise
- Works with Discord, TeamSpeak, OBS and browsers
- Provides raw-versus-cleaned microphone testing
- Available under MIT or Apache-2.0 licenses
Against
- Requires Linux on x86-64
- Needs PipeWire and WirePlumber
- Processing adds 80 ms latency plus buffering
- No Flathub listing yet
The verdict
HushMic provides local, system-wide noise suppression with testing tools and application compatibility. Its Linux and PipeWire requirements, along with CPU processing latency, define the setup considerations.
Who it suits
HushMic suits Linux users who want local microphone noise suppression for calls, recording or streaming. It is also relevant to users seeking open-source software with MIT or Apache-2.0 licensing.
On the record
- Audio processing
- The project states that processing adds 80 ms of latency and runs on the CPU.github.com · read 7 Oct 2026
- Controls
- Users can switch among suppression, bypass, mute, and off modes, including through command-line controls.github.com · read 7 Oct 2026
- CPU limits
- On CPUs without AVX2, the quality model can use about a third of a fast desktop core and more than a whole core on slow chips; on slow chips HushMic runs the light model on its own.github.com · read 5 Oct 2026
- Desktop shortcuts
- Global shortcuts support mute, bypass, push to talk, and push to mute, subject to desktop and portal support.github.com · read 7 Oct 2026
- Distribution
- The project lists installation options for Debian or Ubuntu, Arch Linux, Fedora, NixOS, and AppImage.github.com · read 7 Oct 2026
- Integrations
- HushMic creates a PipeWire virtual microphone that can be selected as input in Discord, TeamSpeak, OBS, or a browser.github.com · read 7 Oct 2026
- Latency
- The project reports 80 ms of processing latency, plus PipeWire's own buffering.github.com · read 5 Oct 2026
- License
- HushMic is available under either the MIT or Apache-2.0 license.github.com · read 7 Oct 2026
- Licensing
- The project is available under either the MIT or Apache-2.0 license, at the user's option.github.com · read 5 Oct 2026
- Local processing
- Audio processing runs on the CPU, and the project says audio is not uploaded.github.com · read 5 Oct 2026
- Maker
- The GitHub profile identifies the maker as Fovty (Johannes).github.com · read 7 Oct 2026
- Microphone test
- The mic test can show raw and cleaned audio side by side and record a 10-second sample for playback.github.com · read 5 Oct 2026
- Mic testing
- The mic test window can show raw and cleaned audio side by side and record a 10-second sample for playback.github.com · read 7 Oct 2026
- Noise removal
- It uses DPDFNet to remove keyboard noise, fan noise, and background chatter from microphone audio.github.com · read 7 Oct 2026
- Noise suppression
- It uses the DPDFNet model to reduce keyboard noise, fans, and background chatter from microphone audio.github.com · read 5 Oct 2026
- Notable limitation
- A Flatpak build is documented, but the maker says there is no Flathub listing yet and the host needs PipeWire 0.3.65 or newer.github.com · read 7 Oct 2026
- Performance options
- Users can choose a quality model or a lower-CPU light model, and HushMic can use a native engine or ONNX Runtime.github.com · read 7 Oct 2026
- Privacy
- The maker states that audio stays on the machine and nothing is uploaded.github.com · read 7 Oct 2026
