Audio Restoration: Removing noise and hum

[mfransen]

Hello,

I have a capture from an VCR (philips N1500) with decent quality audio but it contains noise and a video hum. Is it possible to remove this to get a clean audio track without getting some "bubbly" voices/instruments?

In the attachment is one of the 8 tracks I want to restore. The amount of noise is fairly the same, the amount of hum differs a little bit.

Thank you in advance.

Kinds Regards,

McMurdyy

[redwudz]

Try the freeware audio editor Audacity. If the hum is a consistent frequency, you may be able to remove it. But that will also remove the same frequency from the rest of the audio track.

Just 'noise' or static is much harder to remove as it can contain different frequencies. Less filtering is usually better to keep from
distorting the audio too much.

[johnmeyer]

The head switching noise is tricky to remove. If it were hum, as you suggested, it would be trivial to remove.

Removing head noise from your sample is especially difficult because the audio you provided doesn't have a quiet section that has just the noise. The quiet section at the beginning is noise-free. In the ideal situation you want a quiet section with just noise because it gives the iZotope RX Advanced algorithms the "noiseprint" of what needs to be removed. I used Decrackle, and then found a section that had nothing but the organ. Since that instrument has pure tones and harmonics, I temporarily removed those to give me a reasonably quiet section that had just the noise. I've attached the result. I tried to keep the ambience and reverb of the original, something that can easily get crushed when removing noise. As with any such attempt at improving the audio, it is a tradeoff.

So, Decrackle followed by Denoise.

[mfransen]

Thanks for you try It sounds a little better, but i hoped to get rid of more of the crackeling...
I do have a WAV of the whole episode (it's a TV episode of 50 minutes) with all the songs and the noise and quiet sections maybe. Would it be wise to post that?

[johnmeyer]

I'm afraid that is as good as I know how to do. Someone else will need to try to help.

[mfransen]

Okay maybe someone else likes to try it.
The whole episode is posted here: https://files.videohelp.com/u/198806/Hansje%20en%20het%20welbehagen.wav
The sample i posted in the first post is at the end of the program.

[pandy]

First: In future consider to use flac or similar lossless compression to reduce file size (175.2MB after

@ffmpeg -v quiet -stats -i yourfile -vn -c:a flac -compression_level 12 yourfile.flac"

vs 496MB) before upload.

Seem your source is highly compressed (bell/gong sound is totally clipped - harmonics from clipping are easily audible)- not sure if this is issue for recording or source nature but don't expect miracles. Most important - you can't apply blindly single noise profile to whole file as source was probably created from few different time lines - it is easily perceived during listening - consider to apply noise profile on smaller parts of recording (try to search for local noise profile - perhaps some waterfall spectral view will help to analyse this parts - it was easily visible and audible to me).
There is nothing except noise and harmonics over i assume 14kHz, also bellow 55Hz - i would filter those areas completely - normal cases usable VCR spectrum rarely goes over 9 - 10kHz, hum can be reduced by using group of notch filters located at 50Hz and its harmonics (so 50, 100, 150, 200, 250 ... 500Hz) - it will be good to measure 50Hz real base frequency or you must define quite wide notch filter - not sure what is your preferences.
After all result should be quite OK but without miracles.

[joeandmarg0]

Run the audio through WaveCor,it is free,this may be able to isolate the hum and remove it.You can always read the excellent operators manual before deciding.
Joe

[lordsmurf]

The "full" audio appears to be a comp piece, and each section has different problems/fixes. It must be split up to be processed.

The capture was bad, levels were too high on the first section. You can see flat distortion at 20 seconds, and throughout the audio. Having to filter distortion always hurts some. Best to capture properly from the start. The too-high levels also tends to exaggerate noise, which is a problem here.

The tracking was bad. You may need to manually realign the VCR for best capture. The Philips N1500 VCR is also a piece of crap consumer deck, and should have never been used for quality transfer. This is the sort of crummy work you get from a low-end deck. The only thing the N1500 is good for is as a museum piece.

All of this is correctable in Sound Forge. I was in the mood to do it, until I saw that the single file was many sources crammed together.