Hello audio-mates,

I have got recently an audio of an opera performance (symphony orchestra, choir, soloists audio tracks), but the audio at all channels buried by two types of noises:
1. background somewhat pink (rather than) white noise with peaks at 50, 100, 200 and 400 Hz
2. clicks, stomping and clatter.

What denoise plugin (except Izotope rx) would you recommend? Please provide pros and cons.

Thank you very much in advance!

I am using Steinberg Spectralayers for such tasks. It works for me, but I never really compared it to anything, execpt some old (Waves?) denoiser vst.

Spectralayers or Acon Digital Acoustica.

Almost by accident stumbled on the Noise plug in within Reaper.

Was shocked at the good results I got.

It's free to try.

You're refering to ReaFIR I assume? Very useful as subtractive EQ. I've had good luck with it for hum, HVAC noise, etc. But for coughs and clatters it's hard to beat Izotope Rx (or CEDAR if you're really rich...). You could try spectral editing in Reaper but I have had limited success with erasing coughs using it.

ReaFIR yes indeed. Yeah I think you are correct it does well at the continuous noises (HVAC is no.1).

I used Cedar like a very long time ago on dialogue.

Oddly tried a demo of Izotope last year and it didn't work well at all ( the computer they gave me was awful and flaky, work not Izotope.)

Quite often I Weiss EQ at daft Qs for the odd noise but I'd like something less labour intense.

@ Alcandre
I'm not sure why you are asking for things other than RX, but I think it's the best tool for the job. You need the Standard or Advanced versions to get the two modules that will most directly address the issues in your recording. Spectral Denoise will be your choice for the steady-state noise, and Spectral Repair will be most useful for manual repair of your obtrusive thumps and such. There is a learning curve to using these tools, so don't expect a quick fix without talking the time to learn them.

RX has enabled me to make dramatic improvements to recordings where noise would otherwise severely detract from the enjoyment of the performance. One example: I was given a recording to try and rescue that was choir with chamber orchestra and organ. During the concert, the first violinist had a persistent loud cough, mostly during the quiet sections of the piece. I was able to eliminate nearly 100 obtrusive coughs using Spectral Repair. Also, while the choir was holding their next to last chord, the organist came in a measure too early. Zap! Like it never even happened.

Why don't you want to use RX? It is the best for fixing these types of problems.

i assume you're referring to subgroups/stems?

before a denoiser, i'd use a simple expander (which is NOT a gate on a mild setting!); to re-establish some sense of space, i'm occasionally using artificial efx (reverb), even on the full mix if necessary (for which i'm using a quantec) - for de-noising (on stems, not on a full mix), i'm still using cedar...

I only like the spectral denoiser of RX. If used for a noise reduction up to 4 or 5 dB it doesn't introduce too many artifacts.

I use Acon Digital denoise 2 and declick 2 and I am very satisfied. They are available in a bundle called Restoration Suite 2. The prices are very affordable and the algorithms do a great job.

Before I bought them I was using the denoise algorithm in audacity but the lack of realtime processing was slowing me down a lot. Also Acon Digital's algorithm somehow does a better job.

I can sometimes push it to 7 or 8 dB without artifacts, but at that point it takes the life out of the recording so I only go that high in extreme cases.

If the obtrusive noise is all in one frequency region you can be much more transparent and avoid most artifacts by setting according to the attached screenshot.

You'll notice that the Reduction Amount is set to ZERO, but Reduction Curve is enabled. I also like to use Smoothing. The corner frequency of that curve is set at the point just above where the rumble or other steady noise is most problematic.

What this means of course is that Spectral Denoise is only affecting that region with the most obtrusive noise, leaving alone (in this case) the midrange and treble. As always, adjust by ear and make sure you are also listening with Output Noise Only ticked to make sure no music is being grabbed.

to me, RX is the miracle worker. With the noise removal, I have gotten out steady-state sounds like furnaces and waterfalls. Even the 'walla' of a crowded restaurant. As for intermittent sounds, many of the automatic modules can do just what I need... and if they can't, I can dig into the spectral tools.

Thanks for that tip. Wasn't aware of that possibility. Much appreciated!

The story of denoising technology, particularly when it eventually hybridizes/joins with AI in the coming months and years, is an always-evolving one.

By way of a recent example, Rx and CEDAR were insufficient to isolate the various voices (dialogue) from large film studio ambience, erratic boom miking and music instruments overpowering the Nagra mono document tape…which became the principal audio reference document of the Peter Jackson “Beatles/Get Back” 6 hour documentary, assembled during New Zealand’s covid lockdown period.

There are before/after video samples (in the linked interview below) of the original mono source tapes….as well as the post-processing extractive and mixing miracles that were achieved, using tools invented specifically for realizing the director’s vision.

Necessity is the mother of invention !

https://youtu.be/HN7evYFUWts?si=Jms1cY0OfiaupH2p