Your Vinyl Transfer Workflow (sharing best needledrop practices)*

With the assumend 60dB dynamic range of an l.p., you need 10 good bits.
If you have a 16 bit converter, subtract 6 bits, then you are at -36dBFS (36dB below full scale, that is, the clipping point at 0dBFS). Most if not all 16 bit A/D converters have 96dB dynamic range, that is the theoretical limit for 16bits.

Every so-called 24 bits converters have "only" around 120dB of dynamic range, that are 20 good bits. (There's only very little A/D converters that reach the theoretical 144dB of dynamic range in a real world measurement.)
(The lowest bits convert noise only, noise from the converter chip itself, or from the analog circuits before the conversion.)
So, for our 24 bits converter, we have a noise floor peaking at -120dBFS, your l.p. has around 60dB dynamics, so you should not peak below the -60dBFS mark.

However, in a real world situation, you never would like to go with that low levels. You will aim at around -10dBFS or at -20dBFS, and you are fine. No headaches about clipping, no headaches about "lost dynamics".
The rule of thumb is, if you can clearly hear the surface noises of the l.p. in the needle drop (if you crank up the volume, of course), then the noise floor of the A/D conversion obviously was much lower than the noise floor of your l.p.

Best regards

 

Some good points. Plus that 120dB of range is only for some of the better devices out there and is not always measured in real world situations. Some gear touted as being pro or semi-pro might be only rated at 100-110db of dynamic range. Plus there are factors such as thermal noise from chips and other components that may affect the real values.

In theory, to make maximum use of whatever bit depth a person really has, he or she should aim to have the loudest peaks fall anywhere above -6dB. Since each bit represents about 6.02dB, if a recording is peaking only at say -7dB, it's wasting a bit. If it's peaking at -13dB or lower, at least 2 bits are lost, etc., etc.

However, in practice, just normalizing afterwards to say -1dBFS should be fine. The issue with needledrops of course is that if the level is too low, a big boost will be boosting surface noise as well but that's not really the fault of the recording device. It's a product of the cartridge/phono stage outputs.

I know that as long as one normalizes so that the highest peak reaqches any value below 0dB is technically ok, but if you're going to be converting to a lossy format such as MP3 or AAC, the encoding/decoding process produces overshoots that I've seen as high as about 0.7 or 0.8dB, so I make it a practice to normalize to 1.0dB (As for those who believe normalization is bad, no it's not, if done properly).

 

Sure,
you are 'wasting' bits when peaking at -10dB not at -6dB, but, these are not good bits. But bits that just encode surface noise of the record, or thermal noise of the phono pre-amp or the A/D converter.

If you normalize after the recording, will not hurt dynamics of the needledrop, because all of the 10 to 11 good bits are kept.

Best regards

 

Actually no, these are the "good bits" a.k.a "most significant bits." The noise floor defined by the equipment and the format remains at the same level down in the leat significant bits. If you record low and then normalize to boost the level up to an acceptable value, the noise floor gets raised. Of course, that particular noise floor is at such a low level that you'd have to have complete silence between songs and the level cranked up well beyond normal listening level to actually hear it. The vinyl playback system noise floor is much higher and one I hate having to boost if I can avoid it. I always try to get the best signal possible coming into my recording rig without any clipping and that way, there's less boosting to do later. I've used text records to try and calibrate my system so that signals on the test record at 0VU record at -20dBFS, which is what I've always read to be the conversion (0VU=-20dBFS).

I was just googling for the something related and found this excellent explanation by someone over on the gearslutz forum:

 

I have yet to hear an audible difference between hitting peaks of say, -20dBs to louder records hitting -4dBs on my recorder. And that range probably covers every single record I own. So I normally just set my Tascam up on a louder (hotter) record so that it never clips and then forget it. Setting it up for each record to hit -4dBs is a waste of time, IMHO.

 

Stefan, I am sorry,

but you are wrong.
Think of the noise-floor of the source program itself, the record.
The record's dynamic range is only 60dB, that is 10 good bit, no more.

You can put these 10 good bits on the top, that is, the most significant bit of the 10 good record bits, equals the most significant bit of the 16bit A/D converter. Then, every bit below the 10th bit from top, that is the 6 least significnat bits of the 16 bit A/D converter, are only 'bad' bits, containing nothing like noise.

Now, place the 10 good bits at the lower end, then the least significant bit of the 10 record bits is equal to the least significant bit of the 16 bit A/D converter. The 6 most significant bits of the A/D converter are 'empty'.

No matter what you do, there's only 10 good bits. No matter, where you place them or align them.

If you boost the volume and normalize to 0dBFS, you don't get mor good bits, you just shift the good bits to the top end.
That's all.
Best regards

 

One question to you guys: How do you go about wavering surface noise. It is not steady but comes and goes. What is the best way to get rid of it?

 

I buy another copy of the LP if it is across an entire side

 

I would gladly do it but this is rare record hardly reachable.

 

If an ultra-sonic bath doesn't do it you are done. As far as I know there isn't a program to filter that out. Heat damage is a bitch! I've had so many rare lps that are heat damaged. Makes me want to cry!

 

What about spectral repair? I think Izotope RX has the feature of spectral replacing specific parts of recording that go across all frequencies.

 

I've not used that. Sounds interesting.

 

Maybe somebody knowing this app could share the experience in this.

 

I can manually treat those types of issues with my DAW but it takes a LONG time, and I only do it for rarities. What LP is it?

 

George Harrison Somewhere In England initial UK press with 4 different tracks on it.

 

It works quite well if you can configure it properly. It doesn't always work, though. What I use it most for is getting rid of bad glitches.

 

Post a sample and I will see if RX6 and my limited skills can help.

 

OK. I will prepare sample and post it tomorrow.

 

Can you post a sample where the wavering surface noise is present without music?

 

Can anybody point to a tutorial on how to manually declick/depop a song using iZotope or any other high-caliber specialized software? Thanks!

 

Back in the needle drop thread Stefan did a pretty good job of detailing how to declick in RX2 I checked and the images are gone but the text is still there. Starting at post 1060.
Post Your Needledrops (Part Four)

 

OK. Here is the sample: Dropbox - SIE Sample.aif. I extended the break before the song starts so that you can have more of no music surface noise audible.

 

Well I see your issue the best I can say is you can make it less irritating. Here is what little I can do. How are you recording or processing your records? Everything above 18kHz is gone completely silent.

SIE Sample tweaked.aif (19.49MB) - SendSpace.com

 

I hear that wavering noise in your sample. I've heard it on records I own. Like you I tried cleaning to no avail. Don't know what causes it.

I tried cleaning it up with Magix Audio and Music Cleaning Lab Premium. I wasn't able to eliminate the noise.

https://www.dropbox.com/s/th9bdx4zdfm76qy/SIE SAMPLE 2.aif?dl=0

 

Well this is rather sad story behind it. I bought this record for quite a price and learnt only after it’s on this noisy vinyl giving me the wavering effect as s bonus. I got upset and returned the record but ripped it before. Then my hard disk failed and my WAV file was gone. So I was left with AAC file on my iPhone and what I sent here is 96/24 upsampled AAC version. Now I think I should have kept this record despite this noise...