Your Vinyl Transfer Workflow (sharing best needledrop practices)*

Hah! When I started this journey back in 1998, my very first software was DCArt DiamondCut. It had a tube amp mod that I went crazy with, never mind that my first soundcard was a Creative Ensoniq PCI card. I put that tube emulator on everythang and thought it sounded good! I was recording my vinyl onto DAT, then transferring the result to the computer with that horrible card.

After two or three years, I stepped up to a Turtle Beach Santa Cruz card. The sound quality took a big jump! The last PCI card I had was the Creative E-mu card. Right now i'm using the Amp-Up Realtek audio that does 5.1 surround and DSD. But, it's only good for playback and not recording, which is why I still have to get that Scarlett 212. Yeah, I keep talking about it. I just gotta do it!

 

I have been using one of the two NR profiles that @Stefan posted several pages back for reducing or removing rumble. They are working great, sometimes by itself, and sometimes with some minor tweaking, as I can sometimes detect them removing some bass on some material. And, I finally worked out a de-hum setting that zaps that ground hum I had been getting without affecting the music.

Next week I plan to buy some used vinyl. I hope I find some good stuff.

 

Good luck on the vinyl shopping.
I'm using both of Stefan's denoise profiles. One for each pass. Seems to work. I'm just not using M/S.

 

I almost always tweak the NR profile for each record as no two sound exactly alike in terms of the noise to be removed. I've found that splitting the noise and tonal sliders has made a big difference (I almost never use any tonal denoising). I also vary the horizontal Artifact Control slider between full gating when doing a fade out to preserve as much high freqency info as possible to the opposite end for denoising in other portions with a lot of bass because there's less bass removal. As I've written before, listening to the output with just noise can really help tune the settings for the best results.

What are you using for de-hum? I've found using 150 for the filter Q works well and I never use Linear phase because it widens the Q too much and causes pre-ringing. On sharp percussives there can be post-ringing with or without linear phase but for a lot of material, especially 60 and 50Hz hum when there are not a lot of transients, the 150 Q works well without removing too much info.

 

I see I didn't word my post very well, but you get the point.

I haven't been using mid-side either, lately.

 

The frequency of the hum I have been getting is at exactly 120.0 Hz. I had to do a lot of sweeping to get there. My Q setting is at 616, and I usually limit the harmonics at 2. I never use the linear phase filter because that causes all kinds of undesirable artifacts, as you point out.

 

I took my needledrop of DSOTM and applied M/S and spectral de-noising to the last track on side B. When I put the two sides to L/R the bassline....stayed in the middle. Using RX7.

 

Just got done modding both my turntable and my receiver's preamp. This after simulating the entire circuit of the phono pre, the cartridge impedance, and reverse RIAA.

Yanked out the 220p of loading caps from the preamp so the capacitance is all cable, and increased the load resistance from 47k to 60k, as calculated would be optimal.

Here we get the realized RIAA frequency response of AT92E "recommended" 150pF/47k, and then the response of my 98pF (measured at the cable with cartridge unplugged) into 60k (bold):



We get extended HF response that meets the 27kHz cartridge spec. (this typical high-frequency boost and then attenuation in cartridges should be complementary to the ultrasonic mechanical resonance)

So how does it perform? It performs well for finding the faults in digital remasters...

Nice needledrop of "Don't Be Lonely" by Quarterflash from the Fast Times Soundtrack LP. Hey, I've also got that non-album track on CD when they finally released it on "Best of Quarterflash" - let's compare.



The digital re-release at top? Not only is it out-of-phase and a hair slower (my speed is speed corrected within 0.01%), but they give us less song!

I want to compare the frequency response, but I also notice another abnormality - they've been monkeying with the volume. The white highlighted section after the bridge they quite obviously reached for the fader and turned up the volume, jarring when you're already listing at 100dB SPL.

So we really have to just use the first half of the song. Comparing the frequency response, again it looks more like mastering than cartridge response here - a considerable mid-treble boost to the digital version.



So what do we learn? Not much, besides I didn't break my turntable with several disassemblies...

96k/24 flow:
- 0.642% digital speed correction (until I tweak the little knobbies on the bottom of the turntable again, measuring the number of samples between a perpendicular scratch)
- 4th order 27Hz Cheby highpass
- side channel 65Hz-and-below FFT brickwall (kills 90% out-of-phase hum picked up by cart coils)
- manual heal large clicks either heard or seen in spectral view
- click/pop eliminator at a rather high threshold
- noise reduction with profile of pop-free silent groove section.

 

That's strange as my copy of DSOTM already has the bass in the middle at the end of Eclipse! Could you post a before and after clip?

I'm wondering if you guys are overdoing the denoising on the Side channel as that will remove stereo separation in the bass frequencies if push too hard. Also, I normally only denoise quiet parts in the side channel as the music normally masks any noise the rest of the time. The only time I ever denoise the whole file from beginning to end is if it's a really bad pressing with a lot of noise. Then it's a choice of reducing stereo width in the lower frequencies or putting up with the noise.

 

For de-rumble I use the linear filters back and forth (2 passes -> & <-) from RX6 de-hum. Mid 24Hz, Q12. Side 42Hz, Q12. Q12 is almost a brickwall.

 

I don't de-rumble, I de-resonate 12Hz.
For some reason, long ago it was established that rumble happens more like 300Hz.

Semi-forgot that unique trick of both a forward and reverse pass of the same filter when time coherency is needed.

Where a reversed pass of a filter might be interesting is in "reversing" the phase shift of a mastering processor used on the LP by duplicating it. Let's guess that an LP has a 40Hz 2nd-order highpass filter used in the lathe suite, and even lowpass ultrasonic filtering by design or limitation. You've got excellent perception or analysis skills, or some insight of what equipment might have actually been used. Do it again, backwards!



They could have been exceptionally clever in 1949, and decided that lathes needed to cut reversed tapes backwards, and then the same low and highpass filters are used on both cutting and playback. RIAA 2.0, except then it would make "live" preemphasis not reverse itself.

 

I'm coming to this discussion late, and forgive me but I don't have time to read through the whole thread.

My question is this: I'm considering, instead of using a hi quality phono pre amp (which I can't afford), I was thinking about cobbling together a simple amplifier with a flat frequency response and having the correct input impedance to match my cartridge.

I would then apply the RIAA correction in my recording and editing software (Goldwave)

Your thoughts please. If this has been discussed already, I'd appreciate if someone could post a link.

 

Yes, but not so much here. Here's 50 pages just discussing one "flat" phono preamp for MC. A Low Noise Balanced Input Moving Coil Preamp Using the ZTX851 - Pro Audio Design Forum or 25 pages about another Flat Phono Preamp Based on John's P10 and 2SK389 - Pro Audio Design Forum

One of the best software RIAA EQ is going to be using the RIAA preset in the free SoX command-line audio processing utility. It performs RIAA with biquad IIR filters of the correct time constant.

Low noise and level adjust is especially important, as you'll need to digitize to about -20dB to have enough headroom and noise floor for the 40dB electric gain and 20dB software gain at 20Hz.

 

Here are the results of my "de-rumble" method. Very small losses in low bass area. Not audible. Anyway, after "de-rumble" I give an equal-loudness contour to the entire needledrop (ISO 226-2003) for better sound in LF and HF. The final volume target is -18 dBFS total RMS and -18 LUFS Integrated BS.1770-2/3 loudness. It's a nice sound for ears. No fatigue on long run.

santana.caravanserai.test.wav

 

More transparent sample:
santana.caravanserai.test2.80dB.wav

 

Thanks!

 

Could you give us the link to where he is explaining how he does the denoising?

 

Both those samples sound good to me. Bob what's your work flow these days? As always the reduction "looks" really impressive. You are blackening every non music above 7k not know the song well I'm not sure if you removed and details but it sounds good to me.

 

What are we suppose to be listening for between the two samples when they're not the samples of the same section of the song?

 

Your Vinyl Transfer Workflow (sharing best needledrop practices)*
The entire thread is full of info you will find helpful. I get it like looking for a needle in a haystack. Starting on page 154 and then 161 Stefan's post 3843.
That is his set of 1st pass instructions. Later on he has revised info. This is a work in progress everything is constantly changing you could spend a few hours starting there reading forward you should get a good idea.

I could not find an earlier post where I has posted the same link a while back but I found the notebook I wrote it down in.

 

Here is an example of my most recent needle drop. Traffic When the Eagle Flies. This is a pretty clean record I love the overall sound. There are very few pops but a lot of those nearly inaudible clicks and general vinyl noise you find on these old pressings.

I finally am beginning to grasp what Stefan was saying about not denoising the whole file, at least not at the settings I have been using. I highlighted each gap and or quiet part on the gaps including the fade out and in and adjusting the setting to where I do not remove "ANY" bass. This is easy to check as he posted many times by output noise only. You will see the bass in the output and just keep undoing till you get the right setting.
Here is a sample of a quiet part with some nice bass. My older higher setting would reduce the bass here is did not, I am certain of. Where listening maybe I'll hear it looking at the output noise only you'll see it. Some people find a happy place where a little reduction is ok. I'm trying to go as conservative as possible.
Dropbox - Traffic.flac - Simplify your life
The screenshot just show have little reduction I could use without removing material.

 

I filter out any signal under -80 dBFS or -85 dBFS. With a flat noise profile mask. Eg. White Gaussian noise recorded at -80 dBFS RMS level. Reduction 40 dB. Two passes, forward and backward. Dual mono processing (each channel only), not stereo processing. This de-noise style pushes the noise down another 80 decibels from the chosen threshold and the free space turns black.

I am not convinced that I have removed original details. I think there is only noise, distortion, unwanted harmonics from hum, tonearm resonance, cartridge and cable crosstalk, EMI, RFI, interconnects noise, ADC noise, quantization noise, aliasing, hiss, hum, processing noise, etc. and audio equipment coloration. The master tapes SNR do not exceed 70 dB and so the best vinyl records. I think a transparent needledrop needs only 13-14 bit depth to record the original sound from vinyl.

Finally doesn't matter the method someone use. If the result sounds good, it's good.

You can see what is removed with -80 dB Gaussian mask. -89 dBFS total RMS / -68-69 dBFS max. RMS level. I don't use dither, I prefer to push the noise down under. The final post-processing step.

 

The second sample sounds cleaner because I applied a stronger noise reduction. I not intended to make a A/B debate. Just random examples.

 

Here's another sample of JJ Grey & Mofro this is a more conservative approach. This is with my newly replace TT and I don't mine the floor noise when I'm playing vinyl so I may not need to reduce it that much anyway. I just redid the drop I finished over the weekend this time I was more careful denoising mainly the gaps.
Dropbox - JJ Grey & mofro Brighter days.flac - Simplify your life

 

Extremely nice needle-drop. But, I believe the cut is Whitehouse instead of Brighter days. Very quiet with scary separation and air. I have this needledrop and will have to compare it to yours.