Parameters for High Pass Filter and Noise Remover?

[Charlie Nash]

As my first Audacity Project I have recorded 2 old Ramsey Lewis Trio albums (purchased in the 60s), cutting out some songs that I never cared for and combining 2 albums into one really great album. It has a nice combination of loud passages as well as some very quiet ones. I borrowed from the experiences of Forum contributiors for my Workflow which is as follows:

Wash Vinyl
Capture in Audacity (save as “filename_Raw.au”)
Remove Subsonic rumble using Audacity High Pass Filter
Run Audacity Noise Remover

Export WAV file to be run thru ClickRepair
Run ClickRepair (Declick 35,PitchProt on, Decrackle off, wavelet)

Reopen in Audacity (and save “filename_ed_cr.au”)
Label tracks
Add fade-ins/outs
Make Amplitude adjustments
Optionally run Chris’ Dynamic Compressor (still undecided on this)
Export multiple WAV files
Enjoy

My questions are:

High Pass Filter: Rolloff -24db, default Filter .7071, Cutoff 20 Hz
1- Are these parameters generally okay? Would you make any changes to them?

Noise Remover:
Acquiring 1 sec quiet section from lead-in for Noise Profile (this is normally general vinyl noise, a low roar with some rumble)
Noise Reduction 12db, Smoothing 500Hz, Attack/decay 0.50 secs
2- Are these parameters generally okay? Would you make any changes to them?

3- Would you make any adjustments to the Workflow? If so, why?

Thanks in advance for all the great advice that I have been getting.
Charlie Nash

[steve]

I don't do a lot of vinyl copying, so hopefully others will provide more comprehensive responses, but the general rule with any kind of audio restoration is to keep the processing as gentle as possible.

For example, with rumble filtering I'd try running the high pass filter at 10 Hz 6dB per octave - if that's not enough, then go to 15Hz 12dB/octave and so on until any remaining rumble is sufficiently low to be acceptable (there are no hard and fast rules about what is "acceptable"). Note that when using a high-pass (or low-pass) filter, the filter frequency is defined as the frequency that the attenuation reaches a certain level (often -3dB) but the filter begins to have some effect before that frequency, so a 20Hz high pass filter will reduce the 30Hz range by a small amount, 20Hz by a larger amount, 10 Hz by a substantial amount...
Also, higher dB/octave settings can cause a bit of ringing on transients.

Having said that, the high-pass filter is fairly harmless to the sound quality unless it is massively overdone, so that's a good step to take first before other processing.

The Noise Removal effect in particular needs to be kept to a minimum as too strong a setting will create audible and unpleasant artefacts (usually metallic bubbly sounds) in the music.

When testing out the effects to find the optimum settings it is usually adequate (and much quicker) to just test some short sections. Ctrl+Z to undo.
I would usually leave Noise Removal 'till after other cleaning processes, and with luck it won't be needed at all, or at least only minimally. Noise Removal also performs best when it does not need to do very much work.

I would not normally use any dynamic compression unless I was specifically making a recording to play in the car. (as a musician I like the loud bits to be loud and the quiet bits to be quiet - the way it was intended).

[Trebor]

Noise Remover:
Acquiring 1 sec quiet section from lead-in for Noise Profile (this is normally general vinyl noise, a low roar with some rumble)
Noise Reduction 12db, Smoothing 500Hz, Attack/decay 0.50 secs
2- Are these parameters generally okay? Would you make any changes to them?

3- Would you make any adjustments to the Workflow? If so, why?

I can't see any advantage in having noise reduction fade in/out over half a second if it is being applied to the whole track.
If you set attack/decay to zero you will be spared a tiny bit of hiss at the beginning and end of the track where the NR fades in/out.

[waxcylinder]

See this workflow that I wrote a while back for the 1.3/2.0 manual: http://manual.audacityteam.org/index.ph ... gitization

I never bother with subsonic rumble removal or noise removal (I prefer to mess with the signal as little as possible)

I export a 32-bit WAV file for input to Clickrepair

I have very similar settings in ClickRepair to you - but my default De-click is set to 30 - and I use the Reverse setting

I use the Amplify effect and not the Normalize effect to brin the level up to -1.0dB (Amplify works equally on both channels, Normalize works on each channel independently and thus the stereo balance can be impaired).

I certainly do not use any form of compression - not even the magical Chris's - I prefer the levels and dynamics the way the engineers set them.

One key step you don't mention is backup - After I have exported my multiple 16-bit WAVs I make two backup copies to two separate external 1TB USB disks. Sometimes I also burn a CD - but usually I just import the WAVS into my iTunes library for conversion to AAC at 256 VBR.

WC

[whomper]

i would start the low cut filter at closer to 40cps corner freq
maybe higher depending on the signal that the vinyl actually had

if you are that picky i might run a commercial noise remover instead of the audacity. ditto declick etc.

wikipedia points to more info on restoration and lots of software eg
Sound Forge Version 9 includes a vinyl restoration plug-in and Mastering Effects Bundle, ..... CD Architect 5.2 software which allows (more control on burning cds) ...
or
IZotope RX: Complete Audio Restoration Suite. Ozone 4: Mastering effects system using ... ...

I would skip the fades and cut out the gaps with their noise - you can adjust this with a good burning program to put a second of silence between cuts.

burn a copy to cd after export wav !! burn two!!!!
and back it up to an outboard usb hd

vinyl is more likely to have (almost) enough compression already
but i would run chris's compressor anyway

30dB dynamic range is way plenty for normal people
you want to hear the low volume over the traffic outside
(or the hum of the refrigerator or whoosh of the hvac)
and you dont want the peaks waking the neighbors baby

if you are a golden_eared audiophile who has a soundproof studio to listen in then go for a 90dB plus range

[whomper]

http://www.delback.co.uk/lp-cdr.htm

some more related info
http://recordrestoration.com/

and another tutorial
http://www.wavecor.co.uk/tutorials/vinyltocd1.html

more info
http://www.bl.uk/reshelp/findhelprestyp ... ration.pdf

[whomper]

The order in which restoration processes are carried out makes a great deal of difference to the quality of the result.

The correct sequence is declick, decrackle, debuzz, and then dehiss.

This is because large clicks make it difficult for the de-crackle process to identify and remove the tiny clicks and crackles that constitute surface noise, buzz, and other such problems. Furthermore, if clicks are presented to any dehiss process they confuse it and create unmusical side-effects. Conversely, dehissing first will make it almost impossible to identify and remove clicks and scratches at a later time.

Decrackling should be the second process because small crackles will also cause problems for the dehisser. Similarly, you should perform de-buzzing (when necessary) at an appropriate point in the total restoration process. This will generally be after de-clicking and often after de-crackling. Consequently, dehissing should always be the final process in the restoration chain.

more at
http://www.cedaraudio.com/intro/intro.html

[billw58]

As you can see from the posts above, there are some subjective judgements involved. Inevitable when one's ears are the final arbiter of the result. Thus I feel justified in adding my 2 cents

I'd agree with whomper on the order of processing, but I add a de-rumble step at the beginning. Subsonic content can interfere with all subsequent processes and I believe it is important to get rid of it. Do a Plot Spectrum of the lead-in noise at maximum resolution (size = 16384): if there is any significant content below 20 Hz, get rid of it. In my case my turntable seems to introduce a significant component around 8 Hz. At the very least I run a high-pass filter at 15 Hz and 24 dB/octave.

These days I use Brian Davies' DeNoiseLF for this step. It uses a different process for dealing with rumble and vinyl roar (noise below 200 Hz). The standard de-noising methods do not cope well with noise below a few hundred hertz. It also applies a "sharp" 20 Hz highpass filter, and I trust Mr. Davies to design this filter for minimum intrusion into the audible band.

As for backups, my strategy is this. After capture I export a 16-bit AIF. I feed that to the various processing steps, and import the final processed audio back into Audacity for editing. I toss out everything except the original AIF of the capture - this is my backup. I can always run those AIFs through the process again if I need to. I can also go back to the original capture if I determine that I overdid or underdid a certain step.

-- Bill

[Charlie Nash]

WC

I export a 32-bit WAV file for input to Clickrepair

I have very similar settings in ClickRepair to you - but my default De-click is set to 30 - and I use the Reverse setting

I notice in one of your postings that you state that you use Click Repair in reverse. Have you found that this gives better results? Does it perhaps reduce the damage?
Do you know if other users do the same? Do you ever do it forward? or always in reverse? Do you ever do it both ways, perhaps at different settings?
Charlie

[steve]

Click's rise and fall very suddenly. Some sounds (for example percussion) will rise very suddenly, then decay more slowly. Some bright spark(s) on the internet realised that many "de-click" effects can be fooled by the sharp rise of some sounds such as percussion and process them as if they were clicks, thus (incorrectly) softening the attack of the sound. They thought "ah, now if I reversed the music before de-clicking, then the de-clicker would not be fooled because it would see a gentle rise of percussive notes (the slow decay in reverse), and will only see a sharp rise for real clicks. So it became a popular technique to reverse the audio before de-clicking. The difference of reversing the audio before processing is generally said to be marginal, but with some click repair effects it certainly does reduce the amount of damage that the effect does (less false-positives) to some noticeable degree. The benefit may be small (or with some de-click effects may make no difference), but it will not be worse than processing forward.

If I remember correctly, WC has reported that Brian Davies' ClickRepair now has a setting that will automatically de-click in reverse without needing to reverse the track manually reverse the track. I doubt that it will make a huge difference, but if you want to test it and see if there is a noticeable difference, then testing on music that has hard percussive sounds will probably be the best for revealing the difference (got any harpsichord recordings?)