Your Vinyl Transfer Workflow (sharing best needledrop practices)*

This is the new work flow.
Check for Dc-offset cut 0-5hz if there is any. Mix -85db of pink noise in the entire file. @BrilliantBob Look at the 2nd and 3rd gaps pick the best quiet part for learned denoise scan. Follow @Stefan post 3843 as best I can. Learning each pass and adjusting the residual noise yellow line in RX for both passes. The reduction is around 4 for each trying to keep as low as possible. Topped off with Loudness peak-.3 and RMS -18.

I may not be able to hear any destruction caused by the 2 passes to the entire file. To me denoising before declicking make declicking much easier. Where this is an issue if I should determine this 2 pass denoise is destructive the best I can do is archive the rip and I would have to declick all over should I redo it. Doing declicking 1st I can archive a 32/96 wav file and it's always an easy fix or redo.
Here is a sample New Riders of the Purple Sage One I did last week and the other is the mixed pink noise and 2 pass denoise. If it's not destructive like i've been saying this is a game changer. The subsonics all look clean and bass is low appearing to be untouched. I'm wondering what someone with good hearing would think.
Dropbox - NRPS.flac - Simplify your life

 

I listened to the mono 1st and it sounded pretty good I would not feel like I was missing a thing. The stereo was cool too only filled the headphones differently kind of up and to the left maybe back a little and the whole sample stayed like that. I get why you are questioning the engineer's goal with this. I never understood why they even did fake stereo. I remember buying a used copy of Magical mystery tour and side B was horrible never listened to it.

 

Sounds good other than the intro on both needs some declicking in the right channel. since the music is panned to the left, there's nothing to mask the clicks so you can get away with selecting just that section and applying some heavy declicking. Try Random declicking with a strength of 7. That will leave it sounding a bit hissy but if you go too heavy handed with denoising, it'll sound unnatural. Hiss is not the enemy if it's unobtrusive.

 

Thanks, see I can't hear them. I have to crank it up from -18 to -2 way beyond my comfort zone. So stuff like that I'm OK if it stays. Sometimes I'll hear them through the speakers and I'll drop a marker and go back later in sound forge and remove them.

 

After a little advice from the sage engineers types here.. (better than asking my Mom)

I've been doing a little YouTube series lately on vinyl matchups - featuring needledrops of the same titles, on different vinyl pressings, trying to compare the SQ.
As I'm trying to minimse any interference in the analog chain, either during recording or post, I don't touch the volume levels.

Naturally between different pressings, there are differences occurring in volume levels in the output.
As the human brain tends to naturally favour the louder of two pieces of near identical music, is it fair to not normalise the two pressings for volume levels before publishing do you think?

My view has been, pressing A was at xdB, and pressing Y at xxdB, and that's how they are natrually presenting so that's what should be presented to the YouTube viewer.
But perhaps my recent correspondents are correct in that a fairer comparison would be to normalise the volume on both?

 

I definitely agree that you should try to adjust the level so that all samples have the same sound level. However, normalizing is normally used to target the maximum level of the highesg peak, which has nothing to do with perceived loudness. RMS levels (average levels) are what define perceived loudness. Youtube will automatically adjust audio so that it averages at a level of -14DB LUFS. That could clip some samples of a dynamic needledrop, so I recommend you target that level using a limiter. If you do then you have more control over how your audio will sound. If not, Youtube will control it.

 

Crackles in right channel: from 00:00:00.000 to 00:00:25.833 sec and from 00:01:31.344 to 00:01:57.948. Try de-click Single-band sens 4.0.

 

Thanks guys I got the crackles. My concern was the manor that I'm denoising. I don't hear any loss of bass and or detail but am clearly able to see noise being removed in RX.

 

@ghost rider I fixed your sample. Now it's sound very good on my audio system.

fixed.NRPS.flac

 

Understood, but as you know LUFS won't make each piece exactly the same loudness (you said the same).
I will get the overall peaks and troughs normliaed to an average of -14dB (and I'm already doing that for the entire clip, before it hits YouTube), but the individual songs may sound either less or more loud in parts still, doing that?
In other words, their perceived loudness to a listener, played against each other, one after the other, won't be helped by a limiter or aiming for a LUFs level.

 

Sorry, I should have been clearer. I was talking about peak normalization, which is usually what folks here are talking about when they use the term "normalization." Peak normalization will not necessarily result in two or more audio streams being perceived as being at the same loudness level because it just adjusts the audio level so that the peaks reach a target level. It doesn't look at the average level at all, so if you have one stream with really dynamic audio and another without a lot of dynamics, the latter will end up sounding much louses because it can be boosted a lot more before its peaks hit the target.

Loudness normalization is what you need for your vinyl matchups. This adjusts the average level to a target. If you target the same LUFS level using a properly designed tool, the perceived loudness of similar material will be very close. If a tool has a correct LUFS implementation, it measures loudness by taking into account a system developed from listening tests called K weighting, (basically certain frequencies contribute to our perception of loudness more than others). If you have different types of music with different mix densities, it doesn't work as well, but certainly for your purposes of comparing different pressings/masterings of the same recordings, it should work quite well. As for the limiter, I suggested it simply because vinyl playback tends to result in peaks that will likely exceed the maximum level of 0dBFS if you target -14 LUFS. I know some folks around here consider limiters inherently evil (and when heavily overused, they are definitely not nice) but to catch occasional peaks and keep them from clipping, limiters are fine.

 

I'm still trying to understand LUFS standard. But, I now read that even the industry is ignoring it.

One of the few things I still use Audition for is its excellent match volume function. I just set it to something like -18db or -14db, or match files to a reference, and it works amazingly well. It may choke on a highly compressed file or two, but I remain impressed. I've had no such luck with the one in Rx7 Standard. I'll probably just have to sit down and work with it some more.

 

That would be called RMS normalization, right?

LUFS was supposed to be an easy industry standard for various platforms such as streaming or CD. But, I think it fails because not only is it still complicated to grasp, but even the various streaming services have different standards of their own. Spotify, Apple, Tidal, and others, all have their own, and it makes engineers crazy trying to find the best settings for each one, or the optimum for all of them together.

Amazing how engineers labor over every minute detail of the sound that the average listener is totally oblivious to. It's so frustrating.

 

Actually Spotify, Tidal and YouTbe all now target -14LUFS and Apple -16LUFS, so they're not really their own standards just one different choice of level (Spotify used ot target -11LUFS). There's actually a very good write-up here: TC Electronic | Loudness Explained .

 

Then it's gotten better since I read about it.

 

Got it - thank you.

OK, might try that then.

 

Huge, no massive shout out to harby, BrilliantBob, Stefan, Grant, ghost rider, and many others for the discussion here in the last week. The sum of that has had a tremendous impact on my understanding on what to do to be more effective in my vinyl-to-digital dubs. And I have yet to absorb some of it. So thank you all!

 

By the way, what do you have for software? If you have Audition, do what Grant described. If you have a recent version of iZotope RX, just use the Loudness control available by just entering your LUFS target and a maximum peak. If you have multiple files to process, RX has a batch mode you could use.

If you have neither of these and want to try something free, you could always use foobar2000 combined with its Replaygain functions. This would be a multistage process. First you'd need to load your files and convert them to FLAC and then right click, choose Replaygain and then Scan per-file track gain. This assigns a Replygain adjustment number to the file but doesn't actually change the data. Then you would convert these files. this time in the conversion settings, choosing Processing and then applying Replaygain using the Track as source. This actually adjusts the track by raising or lowering the level by whatever value you assigned. For example, let's say you have a track that when scanned was given a Replaygain value of -3.5. When you apply that on the second conversion, it would actually turn down the track by 3.5dB. If you process multiple tracks this way and don't mess with the default Replaygain target, the resulting files should all be perceived to about the same loudness level. They won't be super loud as Replaygain by default averages around -18dB RMS. However, you'll notice that vocals and other midrange tones especially will be perceived to be at the same level.

Anyway, I know the Foobar2000 method sounds complicated if you're not familiar with the program, but it's actually quite easy to use once you play around with it a bit. Lots of cool and free extensions (they call them components) are available and it sounds good. for free sample rate conversion using a component based on a program called SoX, it's provides quality. foobar2000

 

I use Audacity to rip and to edit (which as I said, is little-to-no edits beyond crop out of starts and ends), then I take the WAVs output and once in the video editing software, I use a new progam I just came across called Auphonic to standardise the resulting vid/sound file to -14LUFs before exporting the whole thing to YouTube.

 

In my opinion, -14 LUFs is fine for a recording with a wide dynamic range. But for a consistently loud audio file (e.g. a thrash metal song), I find that value very conservative. I think -11 LUFs is the best loudness limit for that kind of music, and the value I would set for YouTube videos if I could change it.

 

OK, here's something I just don't understand: why do some people crop the silence at the tops and tails of files? If you want to play them without the silence, just program your playback software that way.

It's OK to crop the tops and tails for your own collection, but if you trade with someone else, it's quite annoying. It's bad enough that mastering engineers do it now.

Generally, I always leave at least .5 seconds at the start of music files, and two seconds at the end. If I make a comp where the intent is to have tight crossfades, then i'll crop all silence.

For me, it depends on how the limiting affects the sound of transients.

 

You didn't read my useage Grant (see halfway up this thread page to my original question).
I rip these files to use in sectional comparisons on YouTube, so I don't need full files, and I certainly don't need dead space at the lead in, or lead out/deadwax.

 

Me I'm mostly ok with -18 I try to just keep everything in the same ballpark. I no back when I was maxing the levels even compressing to get it louder. When I played the CDs in my trucks factory CD player they were distorted badly. Now I just use the volume knob when I want it loud. For the longest I have just normalize to -.3 seems easy enough. I'm still trying to figure out Loudness in RX if there's one for all setting. Right now I've been just setting the true peak to -.3 and RMS to -18 and it comes out looking right. I want some headroom. It's a work in progress for me.

 

I also don't understand the usage of YouTube for needle drops. Is it to share all your music with friends and or the world? I have never listened to any youtube music from anything better than computer audio. I guess it's around the same quality as MP3. I have listened to Pandora on the big system. I still like 24/96 for my needle drops I don't have a problem with file size. I know it's over kill and in a blind test I have to guess which is which. I have always believed it's better to capture and process it high. But I'm just stuck in my old ways.