I searched the forum, but couldn’t find what I was seeking. It’s possibly been answered however will post what I am looking to do and can this be done in Audacity.
I have found many posts about removing snaps and pops from old records by using various programs that, for the most part, simply suppress certain frequencies, etc., with varying degrees of success. I always return to Audacity to remove pops by simply expanding the wave form, then highlighting the snippet that contains the “pop” and either reducing it’s volume, thus preserving the timing, or, if it is like a micro second, I just edit it out. However, if I use the edit function, sometimes Audacity actually leaves a perceptible pop or click of it’s own. So, I normally go with just reducing the volume of that snippet and have had great success.
However, obviously, this can be a very time consuming process for a recording with a lot of scratches, etc. Which brings me to my question.
Years ago, a friend of mine mentioned how NASA is able to retrieve signals from far away explorers amidst the static of man made and natural space energy. As he understood what he heard, NASA takes two separate receptions of the same signal from two receiving stations, usually from opposite ends of the earth and then combines them. Then compares the two signals and anything that is not in both signals is removed. Thus, they get an almost clean original signal.
Using this idea, I have speculated that if you have two identical pressings of the same vinyl recording, which I have numerous double copies that used the exact same pressing matrix, you could record both pressings into Audacity, then have track two overlay track one and where ever the two tracks are identical, Audacity retains the recording and where ever one has something the other doesn’t, Audacity can remove it. That is to say, if we use color as a guide, recording one is in the color blue and recording two is in the color red. You overlay one on the other and you get purple. Any blue or red found would be removed leaving only purple.
Using this process, I would think you could obtain a near perfect rendition of the original recording.
Your only limitations to overcome would be the slight variations with your record player that might be induced into your Audacity recording, as we know a record player can be prone to slight speed changes due to voltage fluctuations, worn belts, etc. But, as long as you can adjust the wave form in Audacity so that you can line up the two recordings as precisely as possible, or if Audacity has the software capability of adjusting the two wave forms so that they match up exactly, then could one expect to get a clean copy, compared to the muddy or frequency suppressed sound that you get from the average “pop and click removal” software?
I’ve thought about the same thing… Most clicks & pops are short-duration and are unlikely to occur at the same time on two or more copies. My idea would be to choose the best (probably the quietest) recording moment-to-moment, rather than trying tying to remove the difference…
I’ve done some programming, but never any audio programming and it would take me about a year… I’m not that motivated… And since the records I digitized were too rare to be found digitally, it may be hard (or expensive) to get a 2nd vinyl copy.
Wave Repair offers several repair methods including copying from one channel to the other. That often works very well (even the same click is often offset in the channels, or worse in one channel than the other). You don’t notice the loss of stereo for a few milliseconds.
Wave Repair is manual (VERY time consuming(1)) but of course the left & right are automatically synchronized.
That’s not easy to do, especially with short-term clicks & pops. i.e. It’s super easy to make a “center-channel vocal remover” by subtracting left & right but there is no simple mathematical way to get the center only… It SEEMS like it would be easy but when you start playing around with L+R and L-R, there is no simple combination of simultaneous equations that keeps what’s common in the “center” and throws-out left-only and right-only. (With FFT you can do this kind of thing, but whereas regular addition and subtraction can be sample-by-sample, FFT needs a longer time-selection.)
That could be done automatically. (There is a tool called VocAlign.) It can’t be “digitally perfect” but I think it could be done well-enough. If you digitize the exact same recording twice at the exact same speed the digital will be different for two reasons - The analog will be sampled a different points along the analog waveform, and the (random) noise from the preamp will be different in both digital copies . (The noise won’t affect the timing).
(1) I was once cleaning-up a vinyl transfer and I had spent at least a couple of days on it when I found a CD (out of print). I was so happy, and it sounded so much better, that I didn’t mind that I had wasted so much time!
P.S.
AI is probably smart-enough and powerful enough to clean-up vinyl recordings… I suspect an “AI powered” application/tool will be available “soon”. … AI can create an Elvis song from nothing but a suggestion and it should be able to do all kinds of audio related things that are currently impossible.
Thanks Doug for the thoughts, I should add that I’m working with old 78 rpm records,
stereo records are a whole other step up and gets more complicated, which I understand. I am also encouraged as regards to locating a second copy of a record as there are a number of web sites devoted to recording and posting pre 40’s records that can be downloaded and used as a second copy, as they were also recorded from the same matrix. Obviously, a popular title, let’s say a Bing Crosby tune, would have been pressed many times over and a regular change out of the matrix would render two copies of the same tune with slight differences that would make it more difficult to obtain a clean copy. But, for a rarer recording that had a run in the hundreds or thousands, I would think my chances of obtaining an identical pressing is greater for the comparison to work.
If you have a stereo cartridge with a 78 stylus, you’ve already got two copies. 
But you’ll still have defects in both channels at the same time so it’s not as good as a completely different copy.
Actually, I have five Newcomb record players from the 50’s and 60’s which were work horses for schools and clubs. They were mono only and I use them to make my digital recordings. The needles are made for rough use and for the materials used to make 78’s prior to the 40’s. I have a stereo turntable dedicated to just LP records, wouldn’t use it for 78’s, even if the needle claims it can play 78’s, I feel old, non vinyl 78’s would just tear up the needle. But I see your thinking about a stereo recording of a mono 78, you’d have two channels to play with, and actually, I would suspect that a scratch would be more prominent on one side of the groove than on the other, depending on the direction from which the scratch was created. Still, this would just lead back to hours of painstaking work of editing which is what I was trying to get away from and might not give me the results I would hope for.
This topic was automatically closed after 30 days. New replies are no longer allowed.