Tranferring vinyl to digital - the utter beginner's thread

I'll take one of these please. To go.

 

I have experimented with this idea using Voxengo Curve EQ to match my hi-fi test LP needledrop of pink noise to a pink noise track generated by Audition. The result is an EQ curve that very closely matches that of the generated noise. However, I have also made a needledrop of my band's single, using an identical setup, which very closely matches the EQ of its digital vinyl master. If I apply the pink noise based EQ fix to this needledrop and compare it again to the digital vinyl master, the resulting curve is less accurate than before. Am I missing something here?

 

Maybe because the vinyl cutting engineer is quite likely to have made some EQ and stereo adjustments to get the best cut?

 

To the pink noise?

 

Nope, I guess I'm not really understanding what your question is? Pink noise is pink noise, but there are many things in most signal chains that are not flat, that could account for the discrepancy. I'm happy to trust that Len (The Cartridge Man) got the pink noise "right" on the Hi Fi test LP. Then it's just a matter of running that through your chain and looking at the end result on an analyser, and applying EQ at the end of the chain to get it back closer to a pink noise response, before you record the rip. You only have to do it once, and record the settings needed (or save a preset if in digital), and it will work on everything, as the chain remains the same. This is the traditional use of an "equaliser", and why it was named as such.

 

Yes, indeed. I did exactly as you describe above: I needledropped the pink noise, compared its EQ curve to that of a pink noise track I generated in Adobe Audition, then I applied the reverse EQ curve (calculated with Voxengo Curve EQ) to my pink noise needle drop so that it matched the EQ of my generated pink noise tone. The result of this is that when I needledropped the pink noise again and applied the reverse EQ curve the result is a very flat pink noise reponse. All good to go, you might think.

However: When I needledropped my band's single, I noticed that the EQ of the needledrop was incredibly close to the EQ of its digital vinyl master (see here). If I apply the aforementioned pink-noise corrected EQ curve to this needledrop the resulting EQ curve is less accurate when compared to its digital vinyl master than if I hadn't applied the "fix". So something must be wrong with the method described above.

 

Because your band's single is not pink noise, but actual music? No music is exactly like pink noise, if it was, it would sound like pink noise. Or because you are EQing it twice? Still not really sure what is happening. It works very well for me. Nothing wrong with the method as far as I am aware. Again, the vinyl cutting engineer may have applied processing to your track to get it to sit better on vinyl.

 

Surely the idea of doing this EQ alteration is to get a more accurate needledrop? It seems, however, that this isn't the case. You say it works very well for you, have you compared your resulting needledrop to digital vinyl masters as well?

 

Here's what I mean:

RED = Digital vinyl master of my band's single
YELLOW = Raw needledrop of corresponding single
GREEN = EQ'd using pink noise method



Clearly the untouched needledrop is more accurate than the re-eq'd one. Thus, the pink noise method to correct for system irregularities is flawed.

 

I'm not really following the line of discussion about pink noise, but I gave up using EQ-matching because, I can do just as good of a job of it by ear.

 

I don't get it either. The vinyl master (whether flat or EQd) and the digital master are of course going to be different, so it's like comparing apples with oranges. Whichever is closest to your original mix is completely beside the point when talking about EQing for the flattest response of a playback system.

I'm in the lucky position to have mastered quite a few releases which have ended up as digital downloads and on vinyl, so I've been able to compare things between the original mix from the client, my digital master, my digital master sans limiting which has been passed on to the lacquer cutting guy, and the final pressed vinyl. I'll say it again: the person who cuts the lacquer is very likely to make adjustments to level, stereo width, and EQ, in order to get the best cut. In my own tests, the vinyl sounds closest to my vinyl "pre-master", WITH the EQ set for the flattest pink noise response, but I'd never expect it to be exactly the same for the reasons mentioned.

And agree with Grant above, doing it by ear is just as good a method!

 

I think perhaps you are misunderstanding my point.

This is exactly what I've done. You can see the results in the graph I posted above. The graph clearly shows that the EQ curve of my untouched needledrop (in yellow) is a closer match to the song's digital vinyl master file (in red) than the EQ curve in green which is the needledrop "WITH the EQ set for the flattest pink noise response".

 

Yes, still misunderstanding you, and I obviously don't have an answer to your question, maybe someone else does? I am sure that sending pink noise through my chain and adjusting the EQ to get back to the same result at the end is the best way to achieve a flat response from my system, so over and out.

 

Stepping back into this conversation again because I'm trialling ClickRepair at the moment. I wonder if anyone cares to offer any advice regarding settings? I know this is a kind of "how long is a piece of string?" question but still... I've read many posts from lots of people suggesting everything from 4 or 5 up to 25/30. Some of the defaults are set to 50...

 

I think it was @quicksrt I've been following on settings lately with descent results, you can always adjust
Method-Simple -Declick 15 Decrackle off Reverse un-checked
Then do it again with reverse checked

 

15 is an ideal starting place. I do the reverse pass first.

File - batch setup- add files - select all tracks needed - send to click repair - start.

DeClick 15 is very mild, and two passes really gets the job done. You can see the red lines showing what has been found and corrected, the second pass in opposite directing really gets things missed in first pass.

 

Oops! Thanks for correction

 

For that second pass, uncheck 'reverse" box before selecting the file - batch pull down. The first thing you do is uncheck or check that box after opening the program.

 

Be very careful with clickrepair. Even as low as 5, it can clip the bottom end off of weird analog bass synths and other instruments that generate sawtooth waveforms, smoothing or smudging the sound in a very odd way. I've had it do this to Pink Floyd's Wish You Were Here (some parts of Welcome to the Machine, etc.). If you zoom in on a click in Audacity, Goldwave or whatever, they usually look like a sawtooth peak with a sharp attack and a really sharp decay. This is, mathematically, what Clickrepair looks for. Stack a bunch of them up and they're not a click after all, but a buzz or, at lower frequencies, a hum, and it can be intentional.

If you want to monitor it carefully, set the output to "noise" while it runs. This way you'll catch unusual passages that it's messing up on, because you'll hear it buzzing instead of a pop or a click here or there. Sure, it will take 100 times longer to process, but you won't end up with a nasty surprise after it's done working. It takes much longer because it plays back in real time. With monitoring off, it just blazes through the audio file as fast as your computer will allow.

Over time, I've gotten a good feel for what it's going to eat, and what I can let it process at warp 9 without worrying about the output, but this just comes from experience and knowing the "shape of the sounds" for that album really well.

One last tip before this gets too heavy. I generally rip and process each side of a record as a single file. However, if Clickrepair isn't playing nice, I'll cut the songs up and feed them into CR individually, or cut the one song out it's having a problem with, repair by hand, and splice it back into the repaired side, THEN go cut it up into individual tracks.

Oh and I have a pretty nice macro set up I run against both sides at once when I'm happy with the CR results and re-import those into Audacity. I'll throw out the settings for that if anyone wants them. They're referred to as "chains" in Audacity, and I have 3 things chained together for final production.

 

Just wanted to add my settings I've been using lately (with cleaned records) and have had much success with:

Pitch protection: on, Reverse: on, DeClick: 6 (I'll go as high as 15 on terrible records), DeCrackle: off, Automatic: All. Method is set to x2.

If you're going to decrackle some really poor vinyl (or just plain bad pressings) it's best to do it in a separate session so you don't accidentally kill an otherwise good declick session. Declick first, save a copy of that, then decrackle.

 

I record into Audacity most the time. I tried using their click removal option and it messed with the audio too much. For me, I keep the recording as close to the signal as possible. I got out of the habit of applying any Normalization or other filters too. If needed I do any eq and gain adjustments, I do it on a mixer before it goes into the audio interface.

If the record plays with some pops and crackles I can live with it. the recording can have them too. they are merely a backup for convenience

 

The built in tool isn't very reliable, that's why we're talking about using the program called ClickRepair. Audacity gives you a few different ways to fix by hand if you're inclined (for major clicks especially). You can use Edit > Remove Special > Silence Audio for very small pops, or if you want to get subatomic with it, there's the Effect > Repair option which actually does a really good job by flattening the peak and smoothing the attack and decay.

I just might get motivated enough to make a Youtube video on this subject eventually, even though I don't have the best grasp on what it can do, I know what I can do with it so far and the results it gives.

 

appreciate your feedback. I briefly looked into ClickRepair (its linked in the Audacity wiki) but figured its just more processing that could potentially alter the sound. Unless something clips I dont really mess with it. Next time I'll try the Repair option though.

 

ClickRepair brand tick removal has the least potential to mess with the sound where one does not want it too. It's designed to only alter the parts it sees as clicks, leaving the rest alone.

 

I did the SQ version of Wish You Were Here at 24/96 and saw ClickRepair do nothing odd to the track Welcome to the Machine. I will try and recreate your scenario with the stereo mix.

The only program material I have so far seen issues with are loud horn sections. Chicago live in Japan, and a TJ Brass LP.

When you see solid red sections in ClickRepair as repaired, then you know to stop program and delete that result. There should be vertical red lines and not solid red blocks or corrected material.