On 02/10/2012 10:01, Don Cox wrote:
> On 02/10/2012, Jon Samuels wrote:
>
>> Tom,
>> Different transfer engineers have different philosophies. Some believe
>> any noise reduction is an unacceptable compromise by definition,
>> others denoise aggressively, in the belief that the buying public
>> can't stand surface noise, and that the engineer can EQ the denoised
>> results to make the final product sound as good as ever.
> Judging from many reviews on Amazon, the customers who complain about
> noise are those who have never before heard an old recording. A small
> minority of the buying public for CDs made from 78s.
>
>> My approach
>> is to remove as much noise as I can without audibly affecting the
>> signal.
> The big difficuly seems to be, not to avoid affecting the sounds of the
> instruments, but to preserve the acoustic of the studio. In the best
> transfers from single-microphone jazz recordings, you are aware that the
> piano is further away. In a less good transfer, it just sounds quiet.
>
>> No decent engineer/producer should accept digital artifacts.
>> That's just a given. You and I are on the same page here. Everything
>> should be done for optimal analog playback first. When digital
>> deticking is required, I prefer what I call manual deticking (what you
>> call re-writing the waveform), to real time or background deticking or
>> decrackling, though the time commitment can be huge. When the source
>> material is very noisy or in poor condition and can't be upgraded, I
>> decide the acceptable level of retained noise on a case-by-case basis,
>> erring on the side of leaving a little too much noise, as opposed to a
>> little too much signal loss. Other engineers can and do differ in
>> their approaches. Jon Samuels
> I found when doing some LP material that the "smoothing" function in the
> WAV Repair program was a good way to redraw the waveform, especially if
> repeated a few times (this can be a script).
>
> http://www.waverepair.com
>
> Regards
Well, my two bits' worth is this.

It should go without saying that you find the best possible source. I 
know from personal experience that Mosaic, for example, are very hot on 
this. Sometimes the best source is an old transfer of an original now 
destroyed, and then you have to do your best to repair the damage done 
by the garden shears and curve benders wielded by a guy who was doing 
his best when what we now have was science fiction.

Optimal transfer should also be a given. Correct stylus size, playback 
speed (although this is not as critical at this stage as it used to be), 
playback characteristic and centreing, which last is hugely important 
and so often neglected. Also, metal parts show a huge top lift over 
vinyl and a large one over shellac, owing to the hardness of the 
material, and this should be allowed for, now or later. I now go into 
digits at this stage for disc material, to give the CEDAR the best shot 
at sorting the muck out.

DSP must be carefully driven, as hard as necessary and no more. The acid 
test is whether your treatment sounds as if the recording had not been 
damaged in the first place. If not, it is better to leave comprehensible 
analogue faults for the ear to listen through than bamboozle ear and 
brain with space monkeys. Broadband de-hissing, in general, should be 
avoided. Most of the objectionable noise on disc is in the few kHz 
range, above most of the ambient clues to the recording environment, 
which you want to leave intact. CEDAR NR software allows you to remove 
this without damaging the ambience.

Bear in mind that not all DSP is not created equal. Some of it, even 
now, is ghastly. CEDAR costs, undoubtedly, but at least you get what you 
pay for.

The worst excesses usually result from somebody leaving CEDAR decrackle 
in "detect" mode, which is purely an aid to adjustment, not an operating 
mode. The effect is as has been described - no hiss at all, but it all 
sounds two feet under Shoeberry Ness.

And yes, John RT did use an M44, as do I for coarse groove. The relative 
humility of the device is far from being a limiting factor in this 
application. Also, many a click did I scrape under his tutelage - the 
technique was the best available at the time, although demanding and 
utterly tedious.

In an ideal world, restoration would be unnecessary. But this isn't and 
it is.