Okay, so I finally did it, I transcribed an LP to digital (aka a 'needle drop'). It was a slightly frustrating, although ultimately rewarding experience which I thought I would share with you here. I trawled around a little on this forum beforehand to see if I could find any useful tips but gave up pretty quickly. The advice on offer seemed to mostly fall into two camps: "buy XYZ USB turntable/phono stage" or "don't bother/buy the CD" and I had no intention of doing either. Here is a short description of what I did do and a couple of lessons learnt along the way.
My aim was to produce as high a quality digital transcription as possible, given my existing equipment, of one of my father's cherished LPs - Beethoven's Violin Concerto played by Alfredo Campoli with the RPO under John Pritchard, (HMV Concert Classic, SXLP 20043). I was prepared to invest a fair amount of time in achieving a good result as I saw this as a learning exercise. I only mention this to make the point that my approach and lessons may not be relevant to anyone whose goal is quantity (i.e. digitizing a given size of LP collection to mp3 in the shortest possible time) rather than quality.
Part of the impetus for embarking on this project was the sudden realization that I had all the necessary equipment on-hand and there were simply no excuses for not doing it. No 'necessary investments' either , oh well. In fact the biggest problem I had was that the audio equipment and the PC equipment were in different rooms! Whether to take Mohammed to the mountain or vice versa? For convenience I decided to move the record deck and preamp into my study which probably was the wrong call, see later.
The specific equipment I used was:
- VPI HW-16.5 record cleaning machine, MoFi 'Super Deep Cleaner' and VPI Record Cleaning Fluid
- Nordic Concept Artist/Breuer 8c/Dynavector XV-1s (deck/arm/cartridge)
- Boulder 1012 (phono/pre)
- Dell XPS tower PC (2006 vintage) running Windows 7
- Lynx L22 sound card, Windows Driver 2.0 Build 017 Release Candidate 3e
- Rudistor RPX-33 and Sennheiser HD800 for monitoring
- Audacity 1.2.6 and later 1.3.9 (beta) for recording and editing (FREE!)
Probably the only point I would like to make here is that the only equipment I deem absolutely necessary to make a good quality digital transcription of an LP is:
- A record cleaner
- A record deck and phono stage
- A PC/Mac with sound card
There is absolutely no need to get a USB-equipped record player or phono stage - your existing record player and phono stage will do the job just fine! Also, the advantage of using your existing equipment (both analog and digital) is that the resulting transfer should almost by definition be of the sound quality you are used to hearing. Almost all PCs have a built-in sound card and the Line Out of your phono stage (or Record Out of your preamp) can be connected to the Line In of the sound card (NOT the microphone input) with a simple RCA-to-3.5mm jack adapter cable.
IMHO, buying something new for this project (e.g. a USB-equipped phono stage) just introduces a big new sound-quality variable and risks disappointment. However if you insist on 'investing' I would suggest that you are likely to get better bang-for-the-buck SQ-wise spending money on a decent sound card (internal or external) than a USB-equipped phono stage. Let the flames begin!
I hooked up the Record Out of the Boulder to the Analog In of the L22 (luckily both run balanced so no special cables needed). The first challenge was to figure out how to configure the Lynx card to record from Analog Ins 1+2 while simultaneously monitoring on Analog Outs 1+2 (which drive the Rudi). That took a surprising amount of futzing, especially since the UI of the Lynx Mixer did not correspond to the instructions in the manual which appeared to be referring to an earlier software version. Finally I found the magic combination of buttons to enable 'hardware pass-through' so I didn't have to use Audacity's 'software pass-through' which consumes precious CPU resources.
Next I set both the Lynx card and Audacity to work at 24-bit 88.2 kHz and tried to set the recording level. This is where I encountered my first big problem. The level was very low - despite the fact that the 'Input Level' slider in Audacity was at maximum. What gives? The Boulder was undoubtely outputting a line level signal which is what the Lynx was expecting, so why so quiet? Checked in the Windows volume mixer, same there - set to maximum - yet when I played the LP the average level was down around -25dB/-30dB and none of the peaks exceeded -12dB which meant that the files really had to be cranked up on playback. Finally I realized what the problem was (long after I was finished, meaning I ended up doing everything twice). The Lynx has two different 'Trim' settings for the Analog Inputs and Outputs (set in the 'Adapter' pane of the Lynx Mixer): +4dBu ('professional balanced') and -10dBV ('consumer balanced and unbalanced'). I had never really noticed these before but they were set to +4dBu, presumably by default. Switching to -10dBV did the trick* and brought the maximum level to the point where the recording only clipped in about 3 places. Two of those were on clicks that I removed anyway and the third was a very marginal clip that wasn't really audible. Nevertheless, if I had infinite time I would have recorded everything a third time taking the level down by 3dB-or-so but it was 1 am by this time and bed was beckoning! Oh well, next time.
* Incidentally, the 'Trim' setting has an equivalent effect on the Analog Out meaning I now need to dial in a fair bit less volume on the Rudi to get the equivalent volume level in the cans.
I saved the Audacity 'project' and listened to the resulting recording. Not bad, pretty damn good in fact! There was some minor vinyl roar and crackle at the start of each side, and of course some master tape hiss from the 1962 recording, but the audio was beautifully clear and clean right across the frequency band and with subjectively excellent micro- and macro-dynamics. In fact, the resolution was so good that I could clearly hear a very low level of mains hum I had never noticed before - from the original recording!!
Next job was to remove the clicks. I had run the record through the VPI twice before starting, following Stephen's recommended 'two-fluid regime' (see his blog - sorry I couldn't find the link, the 'Search' function doesn't seem to cover the blogs - weird!) and the result was very little surface noise indeed - some very minor crackles in a couple of places and only about four 'proper' i.e. scratch-type clicks in total. Kudos to the old man for taking good care of his vinyl - this record is older than I am! Anyway, Audacity has a Click Removal tool for this very purpose, however just like the old 'Unsharp Mask' in Photoshop it takes some practice to get the very best results. After much experimentation I found the default settings of 90 (threshold) and 20 (width) to be well chosen, eventually finding that I only needed to increase the width for the few really large clicks and decrease the threshold for some really small ones. Generally this tool rendered the clicks completely inaudible (nice job Craig DeForest!), however in a few cases there were audible artifacts (at or below the level of the programme, so pretty unobtrusive) that I was unable to banish, seemingly regardless how much I futzed with the parameters. If I start doing this more I may try a more advanced noise removal tool like Golden Wave or Sound Soap, but I suspect there is more mileage to be had from 'Click Removal', if I can only figure out how to extract it.
The result was a pretty clean file, clearly identifiable as an analog recording (I made no attempt to remove the master-tape hiss) and even as an LP transcription - but the latter only really in a few places. I did notice that there was a very slightly higher level of vinyl roar than I am used to, but I initially dismissed this as a subjective effect of monitoring through headphones (I almost always listen to vinyl through the speakers on the bigrig). However when I was done I did some checking and realized that the noise floor was indeed noticeably higher with the record deck on the desk in my study than on its usual rack in the living room. I will not claim to have eliminated all the variables here but as far as I have been able to ascertain the difference is down to the support. Compared to my Grand Prix Audio Monaco four-shelf stand (with Apex footers) the noise floor was about 3-6dB higher when I moved the turntable to my IKEA study desk - despite the fact it was standing on its dedicated Base Technology constrained layer isolation plinth. I was quite astonished at the magnitude of the difference - quite an endorsement of the Grand Prix Audio rack!
After trimming the lead-in and lead-out grooves (but not too much - they establish an important aural 'context'), adding a short fade in and fade out, and splitting side 2 into two tracks (using Audicity's neat 'Label' feature) I tried to export the resulting project to FLAC. This is where I encountered my second problem: the export just gave a useless error message, something like "Exporting to FLAC failed". Great, thanks! I looked in the help and searched for the error message on the Audacity Wiki. Nothing. Damn! After much frustration and poking around I chanced upon the solution in the release notes of the still-in-beta Audacity 1.3.9 (up to that point I had been using the 'latest stable version' 1.2.6). Bingo! There it was in the release notes: "Export to FLAC - now working in Version 1.3.3". Jeez! After upgrading Audacity and opening the project in 1.3.9 the export worked fine (once I'd figured out how to set the parameters correctly - I ended up with a 24/44.1 file first time I tried it).
Eventually I ended up with three 24/88.2 FLAC files with the correct metadata (Audacity fires up a metadata editor on Export - again, pretty handy) however for some reason they wouldn't play in foobar, giving some unhelpful error message about not being able to set an endpoint. By this stage I was beginning to feel the same way! In order to see if foobar was the problem I resampled the files to 16/44.1 using dBpoweramp Music Converter. The conversion worked fine and the resulting files played both in foobar and on the Sonos indicating that the problem might be related to the high sample rate. I initially suspected a bug in the hi-rez FLAC export routine of the still-in-beta Audacity but this turned out not to be the case as the problem disappeared when I upgraded the Lynx device driver from Build 017 RC3 to RC3e. Turns out foobar was just having a problem opening a 24/88.2 stream to the soundcard, well why the heck didn't it just say so!
All-in-all I am very happy with the results I obtained here. I feel I have achieved what I set out to and learnt a great deal in the process - although I wouldn't say I have any more scars on my mighty audiophile derri