Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

NoiseTorch-ng

github.com

  • Free plan
  • Linux
7.0 Club score Visit github.com

What NoiseTorch-ng is

NoiseTorch-ng is a free, open-source Linux application that reduces microphone noise in real time. It uses RNNoise and supports PulseAudio and PipeWire, creating a virtual “Filtered Microphone” that other applications can select. The project describes setup as a two-click process and distributes the app as a small, self-contained, statically linked binary. It names OBS, Mumble, and Discord as applications that can use the filtered microphone. The Voice Activation Threshold can be set as high as 95%, or to 0% to dampen noise without deactivating the microphone. An optional “Filtered Headphones” output lets applications be routed through output filtering. Microphone filtering adds 10 ms of inherent latency; output filtering is approximately 100 ms with PulseAudio. The project reports a suspected update-server and repository breach, says no malicious code was found, and says updates are retrieved from GitHub releases. GitHub lists no SECURITY.md file or published security advisories. The project wiki provides troubleshooting guidance, including debug-log collection instructions. The latest listed release is v0.12.2, dated June 17.

Who it suits

NoiseTorch-ng suits Linux users who want to reduce microphone noise in real time for recording or streaming. It may also suit users who want threshold controls or optional output filtering.

Good to know, and what to watch

  • Free plan for Linux.
  • Works with PulseAudio or PipeWire.
  • Creates a virtual filtered microphone.
  • Threshold can be set from 0% to 95%.
  • Wiki includes troubleshooting and debug-log guidance.
  • Microphone filtering adds 10 ms latency.
  • PulseAudio output filtering is approximately 100 ms.
  • Repository has no SECURITY.md or published advisories.

Free

Free

Free plan

Getting started

  1. Open the NoiseTorch-ng project website
  2. Download its small, self-contained binary
  3. Run the two-click virtual denoising microphone setup
  4. Select “Filtered Microphone” in a supported application
  5. Optionally route audio through “Filtered Headphones”

Limits to check

Microphone filtering introduces 10 ms of latency, and output filtering is approximately 100 ms with PulseAudio; echo cancellation is not included.

The Club's review

NoiseTorch-ng is a free Linux utility that cleans microphone audio before it reaches your applications. It suits people using PulseAudio or PipeWire for calls, recording, or streaming who want a virtual input rather than a subscription service. The strongest case is focused, low-cost voice cleanup; the main cautions are output latency, no echo cancellation, and the project’s security history.

Overview

NoiseTorch-ng uses RNNoise to suppress microphone noise in real time and presents the result as a virtual device. Its scope is deliberately narrow: it is an audio filter, not a meeting suite or a cross-platform product. Linux users who only need cleaner speech can get the essential function without a recurring bill, while anyone needing echo cancellation or broader operating-system support should look elsewhere.

Key features

The app creates a virtual microphone named “Filtered Microphone.” Select it in compatible software instead of your physical microphone; the project specifically names OBS, Mumble, and Discord. Recording and streaming workflows are supported, and the PulseAudio or PipeWire compatibility covers the two audio systems NoiseTorch-ng targets.

Setup is described as a two-click process for the virtual denoising microphone. Distribution is also lightweight: the application is a small, statically linked, self-contained binary, which avoids a larger installation footprint. Those choices suit users who want a focused utility rather than a heavily integrated desktop package.

RNNoise provides the suppression, while Voice Activation Threshold controls how aggressively the input responds. You can set it as high as 95% for a restrictive voice gate, or to 0% to dampen noise while keeping the microphone active. Microphone filtering adds 10 ms of inherent latency, a relatively small trade-off for conversational audio.

NoiseTorch-ng can also expose a virtual “Filtered Headphones” output for applications that need output filtering. With PulseAudio, that path introduces approximately 100 ms of latency, which can make timing-sensitive playback feel delayed. It is therefore more suitable for situations where cleaning output matters more than immediate synchronization. Echo cancellation is not included, so it will not solve speaker-to-microphone feedback on its own.

Support material is provided in the project wiki, including instructions for collecting debug logs. The latest listed release is v0.12.2, released June 17.

Pricing

NoiseTorch-ng has one plan: Free at free for Linux with PulseAudio or PipeWire. There are no paid tiers, usage quotas, seat limits, or renewal terms attached to this plan. It is the right fit for individuals and teams that can run Linux and need the complete stated feature set without subscription costs; there is no cheaper tier that trades away capabilities because the product is already free.

Platforms

NoiseTorch-ng runs on Linux and supports PulseAudio or PipeWire. There are no Windows, macOS, Android, iOS, or web editions in its platform offering, so mixed-device households and organizations need a separate solution for non-Linux users.

Who it's for

Choose NoiseTorch-ng if you use Linux and want real-time microphone noise suppression for calls, recordings, OBS streams, Mumble sessions, or Discord. It is especially appropriate when a virtual input and a small standalone binary are preferable to a subscription product. Consider another tool if you need echo cancellation, broad platform coverage, or output filtering without a noticeable delay.

Pros and cons

Pros

  • Free and open source: the full plan costs nothing, making ongoing deployment inexpensive.
  • Virtual microphone workflow: “Filtered Microphone” can be selected by named recording and communication applications.
  • PulseAudio and PipeWire support: it fits either of the Linux audio systems it targets.
  • Adjustable voice threshold: settings from 0% to 95% allow either continuous dampening or more aggressive voice gating.
  • Low input delay: 10 ms of inherent microphone-filtering latency is modest for live speech.

Cons

  • No echo cancellation: users with speaker bleed or feedback need another solution.
  • Slow filtered output: approximately 100 ms latency with PulseAudio can undermine synchronized playback.
  • Linux only: it cannot serve users on Windows, macOS, mobile, or web platforms.
  • Security concerns: the project reports a suspected update-server and repository breach, although it says no malicious code was found and that updates are retrieved from GitHub releases. GitHub shows no SECURITY.md file and no published security advisories.

Alternatives

For a wider market view, see AI Noise Cancellation Software and AI Speech Enhancement Software.

  • Kleervox is the better choice when you need Linux, Windows, or web access, acoustic echo cancellation, and a virtual microphone, accepting a one-hour-per-day cleaning limit on its free plan.
  • Noise Blocker suits Windows users seeking a free version or a 30-day business trial for up to 20 employees or computers.
  • Krisp AI Meeting Assistant is for users who need Android, iOS, macOS, Windows, or extension support and are willing to evaluate a seven-day premium trial.
  • NVIDIA ShadowPlay is a free Windows option for systems with a supported GeForce GPU and the required driver.
  • neep is aimed at Linux, macOS, and Windows users who prefer a paid service with unlimited use and no daily minutes cap on its annual individual plan.
  • DeskBroadcast is a macOS alternative with a free edition and a one-time Pro purchase.
  • Noise Gator is a free MIT-licensed standalone option for macOS and Windows.
  • Waves Curves AQ is for macOS or Windows users looking for a paid plugin ecosystem rather than a Linux virtual-microphone utility.

Verdict

NoiseTorch-ng is an easy recommendation for Linux users who need free, real-time microphone noise suppression through a virtual input. Its PulseAudio and PipeWire support, adjustable threshold, and low microphone latency cover the core call, recording, and streaming use cases well. Look elsewhere if echo cancellation, non-Linux clients, or low-latency output filtering is essential, and weigh the reported breach history before making it a standard deployment.

The verdict

NoiseTorch-ng offers Linux users microphone noise suppression with a virtual input and optional filtered output. Note the listed latency and the project’s reported suspected breach when considering it.

Questions people ask

Is NoiseTorch-ng free?

Yes. It has a free plan priced at no cost.

Which operating systems does it support?

The listed platform is Linux.

Which audio systems work with it?

It supports PulseAudio and PipeWire.

What applications can use the filtered microphone?

The project names OBS, Mumble, and Discord as applications where “Filtered Microphone” can be selected.

Does it provide echo cancellation?

No. Echo cancellation is listed as unavailable.

What the maker publishes

Audio systems
It supports PulseAudio or PipeWire. source
Binary
NoiseTorch-ng is distributed as a small, statically linked, self-contained binary. source
Integrations
The README names OBS, Mumble, and Discord as applications where the filtered microphone can be selected. source
Latency
Microphone filtering introduces 10 ms of inherent latency, while output filtering is approximately 100 ms with PulseAudio. source
Noise reduction
It uses RNNoise for noise suppression. source
Output filtering
Applications can be routed through a virtual “Filtered Headphones” output. source
Purpose
NoiseTorch-ng is an easy-to-use open-source application for Linux that suppresses microphone noise in real time. source
Release
The latest listed release is v0.12.2, released June 17. source
Security incident
The project reports a suspected update-server and repository breach, says no malicious code was found, and says updates are retrieved from GitHub releases. source
Security policy
GitHub reports that the repository has no SECURITY.md file and no published security advisories. source
Setup
The project describes a two-click setup for its virtual denoising microphone. source
Support
Troubleshooting guidance is provided in the project wiki, including instructions for collecting debug logs. source
Virtual microphone
The application creates a virtual microphone called “Filtered Microphone.” source
Voice threshold
Its Voice Activation Threshold can be set as high as 95%, or to 0% to dampen noise without deactivating the microphone. source

Checked against the maker's pages on 1 Oct 2026.