Gnif's Voice Filter


Gnif's Voice Filter is specifically for those looking to improve their audio quality without having to spend a fortune on expensive microphones, recording equipment, or a studio. This filter was written out of a need to produce production quality recordings without having access to a production studio. Using a neural network specifically tuned to identify and remove sources of background noise a studio-like recording can be made from just about any recorded material in any environment.


Gnif's Voice Filter can be applied to live streams, production broadcasts, and post-recorded material. Latency is kept below the speech jamming threshold allowing for live performances, additionally, it is capable of eliminating most phasing, feedback, and echo issues when used along with multiple microphone sources on a stage.

Noise removed

Gnif's Voice Filter can remove just about every kind of background noise, it is especially effective at identifying and removing the following:

  • Fan/equipment noise
  • Keyboard and mouse noise
  • AC line hum
  • Wind blowing on the microphone
  • Street noise, cars, plains, hellicopters
  • Audience noise (clapping, cheering)
  • Auditorium echo
  • Feedback


As you may already be aware, neural network algorithms usually require a high-power computing platform to operate on, however as Gnif's Voice Filter is already trained and ready to go, it is extremely performant. A general low specification home desktop computer can process live audio at about 16x real time with very little memory usage.

Compatability & Limitations

Currently we have only released a version of Gnif's Voice Filter for Open Broadcast Studio (OBS) 64bit for Windows however commercial licensing is available upon request for integration into other products.

Gnif's Voice Filter is designed specifically for human speech and as such, it is not suitable for vocal work. Because the target is human speech the audio sample rate is fixed at 48,000Hz (48kHz) which is enough bandwidth to capture all details of human speech without requiring the neural network to be re-trained for other sample rates.

Try before you buy

The OBS plugin for 64-bit Windows is available with a 30 day evaluation period after which the cost of registration is only $30

NOTE: You must set OBS to 48kHz audio or the filter will be disabled.

Click here to get your copy today and get started.

Click here to purchase a license