New Feature:

Our classic noise reduction algorithms remove broadband background noise and hum in audio files with slowly varying backgrounds.
We released the first beta version of the new dynamic noise reduction algorithms now, which work much better with fast-changing and complex noises. Listen to the audio examples below, they demonstrate some of the new features and use cases!

Glitch While Streaming by Michael Dziedzic.

How to try out the Beta Denoiser

At the moment, only users with access to our advanced algorithm parameters can try the beta noise reduction algorithms (free users: please just ask for access).

To try them out, create a (singletrack) production in our web interface, toggle the Advanced Algorithm Parameters and change the Denoising Method to one of these options:

Dynamic:
This keeps speech and music signals, but removes everything else in your audio.
Speech Isolation:
Only keeps speech and removes everything else, including music, from your audio.

Warning:
These algorithms might delete signals from your audio, like certain sound effects, because they are classified as noise!
We will provide other settings in the future to specify exactly which signals should be kept or removed - in the meantime, please double-check the results in case you process very complex productions.

Real-World Audio Examples and Use Cases

Here are real-world audio examples from our users to demonstrate some features of the new beta denoising algorithms. More real-world examples will be available after the beta phase.
We recommend listening with headphones, so that you can hear all the details!

1. Extreme Breathing, like Darth Vader

In this example (Freak Show 136), one speaker is breathing directly into the microphone, which creates a Darth Vader-like effect:

Noisy:
Denoised:

2. Recording in a Van, Noise Reduction Amount

This is a recording in a van, with changing engine and wind noises in the background:

Noisy:
Denoised:

You can also use the parameter Noise Reduction Amount, to not remove all the ambiance/noise but just improve the speech intelligibility:

-3dB Denoise:
-9dB Denoise:
-15dB Denoise:

3. Keep Music, Static Background Noise

Our Dynamic denoising method keeps speech and music signals, but removes everything else in your audio (Speech Isolation would remove music as well).
In this example (HeadRightOut Podcast), the intro music and fade out are preserved but static background noises (>30s in the audio sample) are removed:

Noisy:
Denoised:

Noise Reduction Test Files

As additional examples, we added some short files from test datasets, to demonstrate further features and use cases of the new denoiser.

1. Non-Stationary Noises

Male speaker on the left (stereo audio), non-stationary noises in the center:
Noisy: Denoised:

Male speaker, computer game shooting in the background:
Noisy: Denoised:

Female speaker on the left (stereo audio), drilling machine in the background:
Noisy: Denoised:

Female speaker, animals (sheep, birds) in the background:
Noisy: Denoised:

2. Stationary Noises

Female speaker on the left (stereo audio), broadband noise:
Noisy: Denoised:

Male speaker on the left (stereo audio), hum in the background:
Noisy: Denoised:

Male speaker, babble noise in the background:
Noisy: Denoised:

Female speaker, low-frequency hum background:
Noisy: Denoised:

3. Periodic / Music-like Noises

Male speaker with background music (denoising with Speech Isolation method):
Noisy: Denoised:

Female speaker and periodic beeps:
Noisy: Denoised:

Feedback

Please try out our new denoising algorithms and send us some feedback (directly in the production interface or via email) if something works well or if you find any problems.
We are currently tuning the algorithms to more practical use cases. For that any feedback helps a lot!

Stay tuned for further updates and more new audio restoration algorithms :)




Similar entries