Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Detect tonal (hum) content in noise print capture and enable reduction on the tonal content separately of the noisy content (hiss) #18

Open
lucianodato opened this issue Aug 24, 2016 · 7 comments
Labels
dsp enhancement DSP improvements

Comments

@lucianodato
Copy link
Owner

Use some form of peak detection in the spectrum to find bins that are like hum and enable the user to reduce them separately (ala iZotope)

lucianodato referenced this issue in lucianodato/noise-repellent Aug 28, 2016
@lucianodato lucianodato added dsp enhancement DSP improvements and removed dsp enhancement DSP improvements labels Jul 3, 2017
@lucianodato
Copy link
Owner Author

One step further could be to implement a sinusoidal+residual model and even include some kind of resynthesis

@Efenstor
Copy link

Ah, this one would be really helpful to me now... I have to somehow clean the audio of the quadcopter hum, which is a mix of harmonic and white noise, but mostly it's harmonic. And what's even worse, I have to keep all the other ambient noises, including the mic self noise intact, for the sake of realism... Still, even now, Noise Repellent gives the best results when compared to the other dehum-type plugins which are giving "speaking through a fan" artifacts.

@FrnchFrgg
Copy link

@Efenstor Maybe for your use case you could use Noise Repellent but with a 3-band splitter (that is let the unprocessed audio pass through for most frequencies, but use the denoised signal in the most offending band). That would enable to benefit from the NR quality without removing the wanted noise outside of the band. You'll need plugin delay compensation like what Ardour 6 provides, so that the part of the audio that goes through NR is not late compared to the other part.

As for the harmonics, there is a "Notch filter" in Ardour (specific to ardour because it's written in Lua) that can remove a frequecy and its harmonics. Very effective (IIRC it was developped to handle GSM interferences, but I used successfully it to remove a kind of hiss produced by a defective phantom alimentation).

@FrnchFrgg
Copy link

(not saying that this wouldn't be useful in NR, quite the contrary, because sometimes the harmonics are not perfectly aligned)

@Efenstor
Copy link

Efenstor commented Jun 26, 2020

Thanks FrnchFrgg, I'll try it. In the case of the drone harmonics they are not perfectly aligned, because it's the mechanical noises, they span from 300 Hz up to 10-12 kHz, plus doing some glissandi when the drone moves. So it's a very tough noise to get rid of.

@lucianodato
Copy link
Owner Author

I have to get back to this plugin. Looking at it now, almost 3 years later looks awful hahaha. This feature is indeed a nice to have. Back then I remember being obsessed with papers talking about this type of possibilities.

@Efenstor
Copy link

Please return to it Luciano 🙏 ! There were so many amazing features suggested since then. And also, besides the terrible gwc and the rudimentary function in Audacity, it's the only denoiser for Linux and Ardour.

@lucianodato lucianodato transferred this issue from lucianodato/noise-repellent Apr 13, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
dsp enhancement DSP improvements
Projects
None yet
Development

No branches or pull requests

3 participants