Your Vinyl Transfer Workflow (sharing best needledrop practices)*

Morning!

I finally have all the pieces lined up to take a more serious approach to transferring my vinyl to digital. My collection is modest but I do have a number of great LP's that are not (or will not) ever be on CD.

I would like to get a sense of what order folks are doing things and of course what tools and settings you favor. Here's my gear and what I believe will be my workflow (I am trying to tighten up a few areas).

Gear

• Technics SL-1210M5G turntable
• Audio-Technica 440MLa cartridge
• PS Audio GCPH phono stage
• Tascam DR-100 digital recorder / RME Multiface

Software

Wavelab 7
Soundforge 10
Audition 5.5
IzoTope RX2


Workflow (Capture @ 24/96)

• Record side 1
• Record side 2
• Open raw files in Soundforge 10
• Trim head/tail of each file
• Normalize each file to -0.03dbfs (peak)?
• Repair any obvious clicks and pops
• Split to separate tracks. (Still as 24/96)
• Save as individual .wav files
• Open each file in iZotope RX 2
• Perform manual pop removal with RX 2 Declicker
• Perform manual noise removal with RX 2 Denoiser
• Save all files
• Open files et in Soundforge 10
• Down-sample to 44.1khz using iZotope SRC
• Dither to 16 bit w/iZotope MBIT+ (Noise Shaping: Ultra Dither Amount: Normal)
• Save each file with “new” bit rate
• Convert to FLAC/MP3 as required

Questions I have are around the Normalize process and de-clicking. First Normalize - are you guys doing a "peak" normalize or more of an RMS normalize? I see a lot of folks talking up about the new -23dbfs RMS values being touted as the new target volume level to ensure the resultant files have a nice fat dynamic range and an RMS level that in the "pocket". I see others who who rather ensure all peaks are as high as possible (without going over 0.0dbFS...Thoughts?

And declicking? I see a lot of folks using the iZotope RX suite for SR conversion and dithering but hardly anyone using their declicking/denoising tools - most are favoring Click Repair. Is it that the first iteration of RX was so bad or is Click Repair really that good? I am now on the standard version of RX2 and it's said that the declicker and denoise has been improved in a big way.

Appreciate any comments on the workflow (especially my "order" of the tasks) and any of the questions.

Cheers!

VP

 

Nice post VP and a thread I will follow.

I too have been trying to optimize the transfer of Vinyl to digital only to find my recordings sound "veiled" in comparison to others. Not bad, but when I compare them to others, I bang my head against the wall and wonder WTF am I doing wrong.

Anyway, I to am curious about the Normalizing process and I hope to have this defined. 'm sure there will be others who can post about how to maximize the process.

-Scott

 

Hi VP,

Unless you seek to change the balance in level from song to song, I strongly suggest NOT making each file have the same max peak. If you do, you'll find the quiet songs to be as loud as the loudest songs - or worse, the quiet ones will sound louder than the loud songs.

First, I suggest the initial conversion to digital be done so the loudest peak does not exceed -6 dBFS. The reason is that every monolithic A-D converter I know of will exhibit its lowest distortion at that level. In my experience, with a 24-bit recording, you will get a cleaner result with -12 (or -20!) peaks than you will with -3 peaks. (Try it yourself, then match playback levels and hear it.) Final levels should always be adjusted after initial conversion to digital.

Next, when making those final level adjustments, I suggest using a max peak no greater than -0.3 (not -0.03) for the loudest peak. Then match the rest of the album, ideally by listening. Hint: use the lead vocal as a reference and make it consistent from track to track, so you don't have to change volume between tracks on playback.

In the old days, many engineers felt resolution is maximized by ensuring the loudest peaks hit 0. Some still believe this. Others know that in effect, 0 = clipping. The reason is that upon conversion back to analog, there will likely be intersample peaks with higher values than the individual samples. Many D-A converters will "stick" when they encounter these, resulting in possible clicks (which some people will not hear and/or some systems will not resolve - but they will be there).

Leaving some headroom avoids the problem and costs nothing in resolution. In other words there is benefit and no downside.

Hope this helps.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

The turntable is a STD 305m with a SME III arm and Shure 97xe cartridge. Phono pre is from my Scott 299b integrated, one of the best sounding MM phono pres I've heard. I have to stuff a buffer amp between the "Tape out" and my I-Mac, the Yahama K-1020's line stage is used. There's a three meter cable to the I-Mac, I use the onboard processors. I record to Audacity at 44.1/32, set levels to peak around -3db and monitor at a low level. Usually, it's a side at a time. It's at this stage that I use my Click Repair software. I don't normalize until just before dividing into song tracks so dynamic difference is preserved. The process of breaking up the tracks results in downsampling to 44.1/16. Those AIFF files are then named and turned into an I-Tunes list. I burn the disc via I-Tunes.

The absolute audiophile ultimate? Nope. Often better sounding than commercial CDs? You betcha!

 

Good to know. I always maximize to -0.5 DB and try not to monkey with the dynamics any more than I have to.

 

+1 - Hence the questions I have had a great many discussions (some heated) regarding this topic - especially with regard to ballads or quieter tracks. Using a single peak value totally ruins the intended vibe or these tracks....hence I am thinking that normalizing to a uniform RMS level will be much more useful in the long run.

Yes - I stick to these level consistently...either when multitrack recording (I actually stick to -14dbFS there) or for digitizing. This week - I am just getting the hang of this new DR-100 and with it's direct balanced XLR connections to my PS Audio GCPH - I get a perfect -7dbFS level everytime...headroom is king!

Thanks again!

VP

 

Barry, could you expand on this -

Where in the process is this taking place. Do I load the file in say Izotope and increase gain to the -0.3 level?

Thanks.

-Scott

 

Barry makes some great, and really important, points about normalization.

IMO, most of the time it is best not to normalize each track separately. The original mastering engineer most likely put a lot of time, thought and listening into setting the level of each song. You will spoil the feel and flow of the album by normalizing each track individually. For example, you don't want "Songbird" to suddenly sound louder than "Go Your Own Way," which is exactly the result you would get if you individually normalized every track on Rumours to -0.3 dB.

Personally, I normalize each album side before splitting tracks. Most of the time the album sides will match fairly well, but you should always check to make sure vocal levels sound consistent on both sides after normalization, in case one side is more compressed than the other. This can happen when you have a record that used differently mastered stampers on each side, which is not uncommon.

Of course if you have a record that was poorly mastered and had bad level matching to begin with, you might want to play around with setting track volumes individually. Most of the time I'm just looking to preserve the original feel of the vinyl (minus really noticeable clicks an pops), so I tend to leave things alone.

Of course, it's all a matter of personal preference, since you're making the needledrop for yourself, you should do whatever you think sounds best.

 

Scott,

Ideally this takes place in your editor of choice. For my proposed workflow - this is around Step 5...

Cheers!

VP

 

Hi VP,

In my experience, the amount of "heat" in Internet audio discussions is inversely proportional to the confidence of the poster.

I prefer to manually do gain adjustments rather than commit to any automatic process (which I often refer to as "abnormalizing").

If I was going to adjust the levels on an already existing, already mastered album, unless I feel some corrections are in order, I'd use a single setting for the whole album - determined from the loudest track.

No need, in my opinion, for -14 (assuming you mean peaks). While better than -3, I usually aim for -6. However, on many recording sessions, where experience has shown me that real takes tend to end up louder than sound checks, I'll end up with files that peak somewhere around -15 to -12. Again, at 24-bits, it doesn't matter.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

I have roughly the same workflow. One key change is that I do my declicking BEFORE I normalize. The reason for this is that the clicks and pops can be quite loud compared to the music content and they skew the normalization results. The file is much quieter than it needs to be.

To Barry's point, I normalize on the entire album as a single file. My target is -.5 db. If I understand the process correctly, the software locates the highest peak in the file and calculates the difference between it and the target. It then amplifies the entire file by that difference. If my understanding is correct, then all of the material moves proportionally and the loudness relationship between loud and quiet songs is preserved.

My workflow also differs in terms of software used. I do my capture in Audition 3.0. Then I run a light declicking using Clickrepair. Then I use Audition for all other repairs, normalization, SRC, and cd burning.

 

Yes - I mean peak values of anywhere from -18 to -14. I find that if all the tracks are recorded in this range - when it comes time to mix - with the summing of most DAWs...keeping things in this range makes for a very nice comfy "in-the-pocket" mix with no overs, no faders hogging the show and a general overall mix peak of anywhere between -6 and -3...which is ideal for headroom.

I am finding anywhere from -8 to -6 is ideal for vinyl transfer @ 24bit.

VP

 

Hi Scott,

What I mean by ""Final levels should always be adjusted after initial conversion to digital" is that I'll always leave headroom during the initial digital recording (from any analog source: vinyl, tape or microphone).

Once the recording exists in digital form, I'll make the final level adjustment digitally, using the "Adjust Gain" or "Change Gain" function of the software I'm working with. This should be done at a bare minimum of 24-bits. (My favorite software processes internally with 80-bit data paths. Something like Reaper will do it at 64-bits. Most consumer software works internally at 32-bits BUT tends to process at the word length of the source file.)

I do final level adjustment using editing/mastering software. As mentioned above, this is done after the program has already been digitized.

I much prefer to do this manually, using a gain change process for each track, rather than a "normalization" process.

Hope this helps.

Best regards,
Barry
www.soundkeeperrecordings.com
www.barrydiamentaudio.com

 

The Cleaning Process

• Clean with "Pre-Cleaner" and steam then vaccuum
• Place in ultrasonic tank with water/sporicidin mix for 20 minutes (LP's rotate on a spindle)
• Clean with solution and steam and vaccuum

The recording process

• Place turntable on platter and Zerostat
• Place outer ring clamp and lower Terminator arm
• Record in Audition at 192khz
• Adjust sample rate of each side to 48khz and save

The automated post-processing

• Run thru Click Repair at 15 | x2 | Reverse
• Batch Process in Adobe Audition that readjusts sample rate to 192khz and inverts the phase (The E-MU 1212 inverts phase)
• Resample in Izotope RX2 to 44/32

The manual processing

• Open in Audition and join both sides
• Normalize to .7, will do the whole album as one unless there is a big level change from one side to the next (mostly due to one side being way longer and cut softer)
• Manually remove clicks and do between song processing and place song markers

Finalizing

• Dither in Izotope RX2
• Open in Audition and using Markers split file to songs
• Flac the files in Traders Little Helper

 

Here's what I use/do:

Record deck, (Michell Orbe/Michell Techno arm/Van Den Hul MC 2) into a Trichord Dino+ phono stage into a Bryston BP 25 using Vand Den Hul interconnects, then into a Korg MR-1000 using some high end Atlas cables that I fitted some Neutik 1/4" jacks on at the Korg end.

I record as DSD files with peaks around -6dB and then transfer to external hard drives, I then either convert and burn to CD or convert to 24/192 files using Korg Audiogate software. I don't do any messing about or editing, most of my records are in excellent condition so almost no pops or clicks.

I do transfers for labels so I do get to hear what can be done in a studio if you are spending hundreds of Pounds and getting things mastered for release and the results are very impressive, I wouldn't know they were needledrops if I hadn't done them.

 

Denon direct-drive table
Denon DL-110 cart.
Cambridge 540P pre-amp
Tascam CD Recorder
Wave Pad (convert .cda to .wav)
CD Wave (split tracks)
Roxio to burn finished cd-r

 

Technics SL-1200M3D table
A-T95E cart
Mackie Big Knob pre fed to Mackie 1604 mixer
direct channel out to Digi002R interface
iMac Computer
Software: Bias Peak recording at 44.1, 24-bit
Manual click or pop removal
If necessary, auto click removal with Bias Soundsoap Pro
In extreme cases only, Apple AU Peak Limiter (I know, sacrilege...)
Normalize entire side to 99% (0.1db, I never peak at zero)
Split into individual files and rip with Apple Waveburner, or manually create new files and drag into iTunes if that's all I'm doing it for.
That works for me!

 

How much time does this take for all of you for one LP, out of curiosity? I've been looking forward to getting a Tascam digital recorder and starting to transfer some of my LPs to digital but am just not sure I'll have the time in the next year or so due to being a new father. Perhaps I should wait until the kids are a bit older...

 

Music Hall 5.0 turntable
Audio Technica 150MLX cartridge
Cambridge Audio 640p Phono preamp
E-mu 0404 PCI soundcard
Adobe Audition CS5.5 (4)
Click Repair
iZotope RX
Sony Sound Forge 10 w/MBIT+


My current method:

CLEAN

I warsh record with Nitty Gritty record solution and rinse with water. I dry the record with a lint-free cloth, then blow-dry with cool air with a hair dryer. (I don't know why this little step makes a positive difference, but it does).

I then ensure the turntable mat and stylus is clean.

RECORD

I record the vinyl, both sides in one go, in Audition 4 at 32-bit/44.1k or 32-bit/96k, whichever I decide will be best for the situation.

I have been following Barry's suggestion of not exceeding -6 db on recording.

I correct the phase. See my next post.

RESTORATION

After recording in AA4, I remove any outstanding clicks manually, then declick with either AA4 or with Click Repair.

I have gone back to using AA's NR for reducing any surface noise. If I have hard-core issues that can't be resolved with AA, I use RX.

I remove any DC offset in AA and normalize the entire file at once to -.3db.

If there are any outstanding peaks, I will determine how important they are. If I don't think they will harm the transient's sound, I will use a hard limiter to clip them off and unity gain the entire album at once.

I also follow Barry's advice of using the vocals to normalize if I have to make level adjustments.

I also try to ensure that the channels are balanced if they are off.

The reason for taking the peaks up to 98% is to ensure that the file will not clip on SRC/dithering.

I top & tail the files and do any little NR, whatever on the fades if needed (usually for bad vinyl). I preserve the original gaps, if any, but typically create gaps at the album side transition to feel and taste.

I then do a save, and then chop up the file into individual tracks in AA. Save 'em.

CONVERT

I convert the files to redbook in Sound Forge with MBIT+ and save them to FLAC. After that, I tag the files, add artwork, ect in Media Monkey.

ARCHIVE

I create mp3 versions of the FLAC files with the LAME encoder at 320kbps with Switch file converter.

I archive to various hard drives and server.

 

System

• Ortofon 2M Black/Project Xpression II with acrylic platter and Speedbox
• Yaqin all-tube phono-stage with (2x12AX7 RIAA curve, 2x12AU7 pre-amp)
• E-MU 1212M PCI

Cleaning

• Spin-Clean

Recording

• Record each side in Reaper at 24/96

Automatic declicking

• Run each file through Clickrepair (x2 processing, Reverse on). Declick settings vary according to condition of record: 10, 18, or 25 are my usual choices.

Manual declicking and post-processing
Open each file in RX2 Advanced for:

• Manual declicking & clean-up in spectral view
• Clean-up between tracks
• lowpass filtering (I remove everything above 30kHz)
• Placing markers between tracks
• Save file

Level adjustment & File splitting
Open resulting files in Reaper

• Import markers
• Use Reaper's normalize to boost each full album side to 0dBFS (I have the Reaper output mixer set to -0.75 for rendering so this takes care of a global buffer for overshoots. If an album has strange levels across a side (I notice quite a few albums reduce level as the side gets closer to the center), I'll split the files in Reaper at each market and override the level adjustments manually. For music with vocals, I adjust by ear so that vocal levels seem to be roughly the same for all tracks.
• Use Autorender (plugin for Reaper) to split files & save songs as 24/96 flac files. If there is any across-the-board EQ, etc., I'll apply it while rendering the split files in Reaper.

Tagging

• Use MP3Tag (freeware) to tag the flac files.

Playback
I almost create CDs anymore since at home I play the hires files from HD through my 1212M and in my car or elsewhere, I use my Sansa Clip+ (Rockboxed) to play 256k, forced stereo MP3s. To create these from my 24/96 archive files, I usually just use Foobar2000 with the SoX resampler component to creat 16/48kHz MP3s (which are supported by my Clip+). If it's something really critical or I'm burning a CD, I'll use RX2's resampler with Mbit+ Ultra dithering.

 

One crucial step I think you are missing is to do a DC offset removal from both channels before normalizing to 0 dBFS.

 

My 0404 does too. Why?

 

Thank you! I had that in there - but read some other stuff that had folks not doing this....but I will add it back in!

VP

 

123 successful (I thought) needle drops in.... And I have no idea of DC offset means. Please explain.

Thanks!

 

My work-flow is very simple.

Turntable: Sansui SR-2050C
Cart: Ortofon OM10
Amp/Preamp: Sansui G-7700

From the tape out on the Sansui I plug an RCA to Headphone jack wire into my 2010 Macbook Pro.

I have the Macbook Pro set to capture and play back at 24/96 from the Analog in/out

The program I use is Audacity. I record each song separately as a new stereo track and export them individually as a lossless Wav. file. I do not preform any de-click or de-noise processing. The input level in Audacity is set to .8 as it gives plenty of headroom for my cartridge. This setting may be lowered when recording other sources (Cassette, VHS, Open Reel, ect.) I do not adjust the input level for each album or song because if it is to be louder than the last then that is how I will keep it unless there is clipping.

Simple, clean and quick. That is how I get it done.