Your Vinyl Transfer Workflow (sharing best needledrop practices)*

You know a 32 converter is marketing right?
No different than Otari in putting Dolby HX Pro in their MX-80 back in 1987. MARKETING.
To be 32 bit the ADC would have to have a dynamic range of over 180 db.

A good 24 bit converter should yield 120 db of dynamic range. But 115 db is cool. 19 bits is all you get here.

Sweet spot you can't measure.

These values are all peak. Not RMS.
We are talking anout - 18 dbfs RMS.
Peaking at - 2 peak is not the same thing as peaking - 2 dbfs RMS.

 

Byron Lee's , 1961 Jamaica Ska album release is famous. It introduced the world to Ska and yet it is not on CD. Not the mono or stereo mix. I have a stereo copy of Jamaica Ska that Byron Lee gave to my Mother.

And I am sure it is in some fool's biography but my Mother never dated Byron Lee....or anyone else in the band. Nope.

 

With MOTU I record at 192K/24-bit. It's quite enough. 32-bit is fairy tale. Then I de-noise with a -106 dBFS gaussian white noise mask saved in Profiles to remove the ADC THD+N. What is removed is pure HF garbage that pollutes the original vinyl sound.

I think the trick to obtain maximum SNR in the recording chain is to set high the analog volume in MOTU (without red color in the input LCD meters) and further the digital sound to be reduced to -20 dBFS total RMS in the recording app. After post-processing it's easy to set the whole LP volume to -18 dBFS total RMS. If some peaks climb beyond -6 dBFS it's not a problem at all. The vinyl sound is already mixed and mastered by professionals.

 

It appears to be a good converter with an interesting design. You seem to know what you are doing.

 

They put all the tech information at the back of the album cover. It is bizzare. It was mixed to a Ampex half inch 3 track. Mixed hard left and right with nothing in the middle just like PPM and WTB.

Track 1: rhythm section . That is bass, drums and rhythm guitar. And rhythm horns. The ones in SKA that blare every quarter note.

Track 2: lead instruments. Piano, brass and reeds playing harmony parts.

Track 3: vocals.

The record is like 70 - 10 000 hz.

There are actually 4 separate versions/mixes of the hit single Sammy Dead (from LP)

1. The mono single mix
2. The mono mix from the album
3. The stereo mix from the album
4. The 1970's re-record.

If you ever wondered what it would sound like if Glen Miller did Ska than this is the record for you. I would put the whole thing up if I could. MEGA expired in a month and then wanted money for a membership. Is there anyway to put songs up on the forum without paying for some membership?

 

WeTransfer, Google Drive, OneDrive, Dropbox...

I would certainly like to hear that one.

 

We are told to limit audio posts to a short clip around 30 seconds or so to avoid legal issues for the forum (we certainly don't want to get the forum or our host into trouble).

Plus this thread is not the place for it! There's been more than enough threadcrapping and thread hijacking here lately. As per the title "Your Vinyl Transfer Workflow (sharing best needledrop practices)" is our topic here. That's it, that's all. Yes, we can always set certain individuals on ignore, but that doesn't work for those who choose to reply to the threadcraps and thus what is usually an invaluable resource for those looking to improve their needledrops gets bogged down with completely off-topic stuff.

There is a thread in the Long and Winding Thread section where folks post needledrop samples: Post Your Needledrops (Part 8). For non-needledrop samples, I guess it would be best for folks to create a thread dedicated to the sample in question. As for this thread, the only samples we normally post here are for comparisons of techniques, cartridges, denoising samples, etc. For example ghost rider has posted some excellent comparison samples of before and after denoising, etc.

 

Oh yeah, I totally forgot you guys in North America have far more draconian copyright laws to deal with. Here in Europe an album released in 1964 has been in the public domain for years.

 

I think this is in keeping with the thread topic: all that music should be in public domain here, too, but, as you called it, there are ways the content creators work around it and keep their works monetized. I'm not an entertainment attorney, but i'll bet the sudden rash of veteran artists selling their catalogs to companies has something to do with keeping that copyright control beyond the 50 years.

 

Sorry Grant, but what does whether or not releases are in the public domain have to do with improving needledropping workflow? It's certainly a topic for discussion but it deserves its own thread.

You've been around here long enough to know how a thread can slip into discussions that have nothing to do with the topic. This one has been a valuable resource for people trying to improve their needledrops, so let's try and keep it focused on the topic. There's nothing preventing anyone from creating a thread about public domain versus copyright, etc.

I'm glad Gary took the time to drop in and remind us to keep on topic.

 

I was referring to posting samples. There is a thread for that, but they are also done here to show the results of their techniques.

I admit that I get sucked into posting off-topic matter. That's my failing. I'll try to control that.

 

We'll stage an intervention for you, Grant

 

I have a question concerning normalization with iZotope RX.

This is the situation:

Suppose you transfer a record, after recording is complete one has two files (Side 1 & Side 2).
Now one can normalize the two files and everything should be fine.

But

what should one do if one has splitted up the two files (Side 1 & Side 2) into the songs that are
on the album and then wants to normalize (all of) them?

 

If you have them still as one file per side, you can choose to normalize the whole thing to preserve what`s assumed to be the original intent, (at least of the cutting engineer) in terms of level, but once they're split into songs, if you want to preserve the level differences between songs, you need to calculate offsets, etc. Personally, I like normalizing to the same Integrated LFFS level (if your version of RX has the Loudness (RX 7 and possibly earlier, I can`t recall) or Loudness Control (RX 8) module. A good level to aim for is -18 LKFS providing your True Peaks within a song don't exceed 0. LKFS is based on scientific listening tests on human perception of loudness. I find that this method ends up with vocals (for example) that sound consistent from one track to another, as if the same person was performing live. It doesn't at all ruin the flow of louder songs versus softer songs as some would assume. Try it and see how you like it.

 

Thank you!

My version of iZotope has the "Loudness module".
Could you tell me which numbers shall I choose
for "True peak" and for "Integrated loudness" when I work with single songs?

 

Try -18 for Integrated Loudness and -1 for True Peak. If you see that this reduces a lot of peaks, that means your recording is quite dynamic. You might then undo the Loudness and try -20 for integrated loudness.

 

Today, The Verge published a peak-behind-the-scenes look at how the Internet Archive digitizes 78rpm records. Fun fact: 78s were mostly made from shellac, i.e., beetle resin, and were the brittle predecessors to the LP (microgroove) era.

Here’s how the Internet Archive digitizes 78rpm records ​

Also, buried at the bottom of the article is a link to the whole collection: https://archive.org/details/georgeblood

 

I'm still on RX 7 so I have to do all of my intricate normalization with Ozone 9 which is very complicated to use until you get the hang of it. But, for quick and dirty results, I use Audition CS5, and it works quite well.

 

Which "Target peak level [dBFS]" should one choose for normalizing one file per side using iZotope RX?

 

Well if you're normalizing to peak values, the critical point is to keep it below 0 dBFS but since you're using RX 8, you should be sure you're keeping it below the "True Peak" value as this is what the reconstructed analog waveform peaks will hit during playback.

This being said, I've often found differences in levels between the two sides of an album, so for the sake of having a consistent listening volume, I would not normalize to a peak level because if one side has lower peaks than the other, the perceived loudness level for each side may differ quite a bit.

Instead, normalize each side to the same Integrated [LKFS] value. Here's a picture of the Loudness Control screen from a quick Google I did for RX 8 help (I'm at my workplace and don't have RX 8 here). The -20 Integrated LKFS is not a bad choice for really dynamic audio as it gives you (in this case with the -2 True Peak) 18dB of crest factor. The EBU R128 recommendation for maximum True Peak is -1 dBTP, so I would use that. -18 Integrated LKFS is a good target for slightly less dynamic material. I did over 20 needledrops this week when I was working from home and only about 5 of them had peaks that required me to target an Integrated LKFS below -18dB. For those, I targered -20dB and it was fine.

 

Hey guys I'm playing records again..

Bes from Music Direct came by yesterday and set up my replacement TT. It looks and sounds beautiful. I figured out a way to safely (for my shoulder surgery) play records. The one thing I can't do is take the cover on and off, so the cover will be staying off for a while.

The first needle drop I worked on in a fresh rip of Pink Floyd the Wall. I didn't have quite the weirdness I had 6 weeks ago but I determined working in M/S may in rare cases cause issues. To recap the issue. At the end of side B there is a bass line totally in the left channel. Using RX8 advanced once in M/S if I denoise only in the side channel the bass line migrates to both channels equally. If I denoise both channels in M/S then it returns mostly correct but some residual bass remains in the right channel.

Since one of my biggest concern is to not make changes to the original engineer's intent. I went through a phase of doing that.

So I am now trying Stefans 2pass but not in M/S. I used a reduction level of 4 for both passes. My initial impression is this is good. It reduces the resonance and other floor noise smoothly while not reducing any bass. I used that part of the Wall and a higher reduction did trim off some of the low frequencies. I know I had much success using M/S but it's my fear of the unknown that has me stepping back from it, for now at least. It does make sense, especially when removing material from one channel that in certain situations it may not convert back the same. It's easy for me to get stuck on stupid doing drop after drop only to find out I'm harming the music.

By the same token I started using normalize and waveform stats to level my drops to about -18 RMS LUFS. It's pretty easy to do and much faster than loudness. I like doing it this way because a dynamic record might get the peaks clipped using loudness, if you are not careful.

The screenshot is a good example of how if you are not careful and you start leveling the channels for the whole file you could make changes to the original engineers intent. This was Golden Earring 8 miles High and there is a lot channel separation certain instruments mainly in one channel or another. Looking at the Stats you might thing I need to raise the right channel but when I listen to the louder tracks the right channel is peaking higher than the left so I left it alone. The most I do generally is bump one channel or the other up between .25 to 1.0 unless I determine it is just a crappy pressing.

 

Glad to hear you're back in the game. However, one thing I feel the need to point out is that the peaking has virtually nothing to do with the balance of the channels (which is what you seem to indicate with your text). The only thing you need to do with regards to peaks is make sure they don't go higher than zero (True peaks that is). They have nothing to do with balance in a stereo record. Depending on the recording, the transients may or may not be centered so the peaks won't necessarily even be close. Most of the time, they'll probably be within a dB or so of each other but not always.

You need to listen to the results after trying an RMS balance to determine if it sounds balanced. Sometimes an adjustment is required. For example, yesterday I was needledropping a UK pressing of the Beatles Hard Days Night LP. Those old stereo mixes are often difficult to balance because of how they are mixed but I decided to try balancing anyway. I found that it worked for most tracks but in some cases, the lead vocal wasn't centered, so I boosted the appropriate channel in each case 1dB at a time until each track sounded good. Another good way of doing this is to use the Mixing module's Mix to Mono preset and preview wby turning the bypass on and off to see if the vocal jumps to one side when you switch to stereo by hitting the bypass. However, for the most part, I find that at least 90% of the time balancing the Total RMS values really provides the best sound.

As for the original engineer's intent, well that's an individual choice. As I've written before in this thread and elsewhere, I've come across several examples of albums that suffered from careless processing somewhere in the chain of work done between the initial recording, mixing, mastering, cutting to vinyl. channel misbalances, heavy 60Hz hum (or 50Hz for European pressings), goosed treble and emasculated bass on 80's LPs, etc. This brings up the question of which engineer? The mix engineer, the mastering engineer or the cutting engineer (and again I use the term "engineer" loosely as none of these people may actually be truly certified "engineers"). Take the example of Roxy Music's Avalon that I mentioned before in this thread, when the tracks on one side are nearly 3dB unbalanced to one side (one of the tracks was actually 3.5dB louder on the right) and with a similar misbalance on the other side, I certainly question the competence of the person responsible. If this was the "original engineer's intent" then he needs to change his career because it's terrible work. Avalon is a sublime recording that sounds stunning when properly balanced across a stereo field. It's definitely not some old early 60's stereo mix before they knew what they were doing with stereo. I have a Canadian pressing of Toto IV that has a similar problem with the first side although not quite as drastic. So really I don't give a rat's rearend about some anonymous studio worker's original intent, which many not at all correspond to the original mixing person's original intent, or to the desire of the original artist.

Besides, I do this hobby to produce sound that I enjoy. I don't do this as a job or to sell a product so I really don't care.

Bill, I'm not trying to be critical, I just don't believe in some mythical, untouchable original intent of one of the "engineers" involved in the production of the recordings we listen to. After all, If that were the case, then we should stop complaining about nasty brickwall masters from the past 20 years or so because of course they were some engineer's original intent!

 

You pretty much made my point. I lack the ability articulate it. Like you said balancing the RMS mostly gets it done. With the Golden Earring it would have wrecked the imaging.

I like the way I use normalize to get to -18 RMS LUFS

Most records I have sound pretty good and I get many engineers did and do crappy work. I try to fix those to.

I get you and Grant , Bob have better ears and skills. I remember before I found this forum I was running everything through Ozone thinking it sounds cool. Thinking back it was like back when Home Theater was new all those music modes Hall, Stadium,

 

I still have problems posting with my phone. Posted it before I was done.

My new work flow I record manual declick then I denoise. So when I denoise the Pink Floyd only to the side channel that bass line stayed equally in both channels. I have no idea why you have different results.