I'm back ... if anyone's thinking of having a second honeymoon, don't leave it till you get to our age!

Ok, the story so far: I tried the trial version of DePopper and it's not bad, but not quite up to what I need. GoldWave seems to do it pretty well. What I've found the best approach is ...

Record in stereo

Copy the run-in noise to the clipboard, then select the whole file and click on Noise reduction

Set the parameters to Use clipboard, FFT=12, Overlap=95, Scale=100 and run it

Click on Pop/Click and set the parameter to 1000 and run it

Select the left channel and click Save2 to save it to a mono file, "left"

Do the same with the right channel, saving it as "left"

Save it again as "mono" (either chanel -- it's just to create a receiving file)

Close all the current files and open "left", "right" and "mono"

Click on f(x) (the expression evaluator) and type in (wave2(n)+wave3(n))/2, where file #1 is "mono", file #2 is "left" and file #3 is "right"
(Save this as a new group/expression for use another time)

Run the expression: it will calculate the average (arithmetic mean, if you're fussy) of the two channels and put the result in file "mono"


What we end up with is (almost) the finished product. Some of the noise is on one channel only, or occasionally some of it is in anti-phase, so any remaining differences ("noise", by definition) will reduce or even cancel out in the averaging process. If there are any remaining clicks/pops after all this, they can be hunted down manually and removed by selecting a smallish surrounding area and running pop/click with a lower setting of 750 or 500. (This is very clever: it doesn't just smooth out the jump, it "guestimates" what the signal should have been and replaces the jump with that; it sounds very good.) If the dynamic range is a bit extreme, it can be expanded/compressed.

So far so good ... it only takes about 30-40 minutes to clean up a whole disc! 5,000 hours is about 200+ days ... but at one or two discs a night, that's ... a lot of nights! I'll have to see if I can run GW in command line mode or something.

So, the last problem (which really ought to be the first job) is to redo the equalisation from the modern RIEEE curve to the appropriate one for the recording. Hmm ...

Steve