Your Vinyl Transfer Workflow (sharing best needledrop practices)*

You're summing to mono twice, once in hardware and once in software. Why at all, and why twice?

When I digitise a mono record I never sum, considering that if there are slight phase differences between the two channels they might cancel each other out, effectively working as a low pass filter.

The way I go about it is I first record in stereo because that's just how the hardware works, but then only keep one of the channels, depending which one has fewer clicks. The other channel gets discarded completely. Then I create a single channel file from that because there's no need to create a file with two identical audio channels (unless you want to create a CD from it, but that's a separate issue).

 

I record mono discs in stereo, mix to M/S, discard the side channel, keep the middle and work from that. This method gives good results and I get rid of the clicks in the side channel quickly.

 

This is a good way to do it. If you really want to go whole hog, you can jump back and forth between the channels if one or the other has a louder click or extraneous noise at a given point. But that can be a lot of work and can drive you crazy if you start.

 

I am back at digitizing my records(not all records are made of vinyl so calling a record vinyl isn't always true). Neither do I needledrop. I never just drop a needle on a record. Anyway, I use styli and not needles in the equipment I use for recording. I use needles in my Victrola acoustic phonograph, and at that I use jewel-tipped needles. I get a little irritated about some of the terms people use for records, "vinyls" being the worst. But that's just me, to each their own.

Out of the 10+ stereo cartridges I use, I haven't found one that has a balanced output.

I always use a program's built-in mono features when recording, track-splitting and noise removal on mono records. With a stereo cartridge, the stylus can bounce around inside the groove(especially when playing an older microgroove mono record cut 1mil with a .7mil stylus) which can either pick-up or create anomalies. Use a program's features to record a mono record in mono. Using "Y" cables should have ended after analog recording was over, like recording to magnetic recording tape.

 

Likewise! Which was a bit of a revelation when I started doing needle drops of my vinyls. (Sorry.....couldn't resist!). I haven't digitized any of the few mono releases I have, but for stereo I always calibrate the L/R channels using a test tone from a turntable setup album using the gain controls on my audio interface whenever I switch cartridges. Point being, despite the lengths and expense some audiophiles go through seeking the ultimate listening experience, it's highly likely they aren't even hearing the proper channel balance!

 

This is fantastic work! I can't believe I missed this post!

 

I'm wondering when AI assisted software will be available for needledroping? I mean AI that can analyse the record and make out what is a click and what isn't, and correct that without affecting the signal. And good noise reduction.

 

The RX software from Izotope supposedly uses AI for detecting problems in audio files. So appears it is already happening.

 

I'm not as sophisticated as some of you folks in using all of these tools for setting azimuth. First I use one of those low tech acrylic graduated or lined blocks and 10x headset to set azimuth by eye. Then my routine to get azimuth more correct is to record a track or two from a very clean near mint mono record. I usually use Miles Davis' Kind of Blue or Dave Brubeck's Time Out.

Then I pull those recorded tracks into IzotopeRX9 and perform Waveform Analysis. I look at the entire recording then go into each track and to select sections. What I examine is total RMS for each channel. Typically I will record a portion of the record from 1/3 the way in to 2/3 the way in per side.

My unpivot tonearm has a lever at the base that permits very small adjustments to azimuth. So in that way, I can tweak azimuth to improve the balance between left and right. Of course making those changes means I need to examine VTF again and also anti-skate. So I do that routine too.

In this way, I have been able to get a match as good as 0.15 dB.

When I record a mono record, such as The Barbara Carroll Trio Funny Face that I did the the other day, I re-examine that match for tracks or the entire recorded lp. I still have channels within 0.15dB.

This method is perhaps brute force and not as elegant.

 

AI is already here for audio processing, and it leaves a lot to be desired in my experience with it.

Again, it's not very good. There's a lot to be said for good ol' skill and intuition.

 

I actually find RX almost too sensitive for click removal. So yeah, I typically use it for scanning but fix manually if necessary.

 

Haven’t been here for a while and checking in. My cartridge was getting va lot of accumulated hours since buying new in 2014. I sold two rare LPs come months ago that my brother gave to me one Christmas many decades ago. I had read out them in an audiophile mag and since he was working at a major record distributor I asked if he might find me some sealed original releases. I didn’t like the at all on first play or any other tries. I ended up getting over $400 a piece on eBay so I bought myself a new discounted Sumiko Songbird HO MC cartridge to replace the old one. I absolutely hate setting up a new cartridge. Would rather get teeth pulled with no pain killer. Got through mounting, stylus force and overhang. I spent most of my day yesterday continuing to finish cartridge azimuth alignment I started a few weeks ago.. It was quite difficult to get convergence and it took me from right after lunch up to about 8:30 PM to complete. I initially used the 1KHz azimuth tracks on the Ortofon Test LP but had an awful time getting to a null point. For all its other merits, my arm does not have good design for making very small incremental azimuth adjustments. The linear tracking arm tube has a rotation sleeve that has to be gripped with two fingers and twisted with side rotation about the stylus axis. The range of rotation is very small and the bearing mating surfaces have friction so a tight grip is needed. There are no scale markings to go by. Even though I used a clear alignment block to set a 90 degree starting point, when I took my first measurement, the L and R crosstalk levels were off by 3.3dB. Making a tiny twist raised the offset to 6.6dB, the opposite of improvement. The rest of my trials would attempt to align crosstalk by reducing the level of one channel and increase the other until they got very close together. But every time I made a new adjustment it would either stay the same, get worse, or flip to the opposite direction. This happened over and over again. I also tried all the duplicate AZ bands on both sides of the Ortofon record to check consistency and still had same results. I was spending a lot of time on each trial performing noise reduction to get the best accuracy before calculating levels and that took time. I also made sure that my total RMS measurements matched closely with a separate set of ones gathered from a spectrum plot. Late in the day after more than 30 trials I decided to really speed things up by doing no calculations or full recordings at all. I just continuously recorded in my DAW while dropping the stylus in and out with each tweak while looking at the moving waveforms to see if one channel was increasing as the greater channel was decreasing. After each mechanical tweak, I sometimes had to restart twisting in the opposite direction if I flipped past a null. This was much more efficient, and eventually I saw visual evidence of quite equal amplitudes in the passing waveform displays. I stopped, recorded the full tracks, performed noise reduction, and took measurements. I also took another recording using the Analog Productions LP to see if I got the same results and I did, even slightly better. With the Ortofon bands I got a crosstalk match between L>R and R>L of 2.1dB with an RMS calculation and 2.3dB with FFT spectrum plot values. The results from the AP test LP recording was even better giving me a crosstalk match of 0.54dB with RMS measurements and a 0.4dB match off the spectrum plot. Full signal channel balance calculations were good too with 0.73dB (RMS) and 0.5dB (spectrum). Cartridge spec is 0.5dB. 1st harmonic distortion balance between R and L had a 0.38% match between the nicely low 0.7%L and 0.32%R calculations. Distortion from the Ortofon recordings were just a little higher (1.05%/1.38%). All good enough for me. I had a little bit of time before bed to take a listen of a couple favorite vinyl tracks and they sounded the best ever with deep 3D imaging and wide channel spread with lots of detail and naturalness. I only have VTA left and that is the least critical calibration. It will be easier because my arm has a great adjustment tool built in that can precisely put back VTA to my starting point if need be and it does not affect the overhang settings. This was really, really hard but I managed to get there. Now, up for some great needledrops!

 

I need guidance on using IzotopeRX9 (or any version) on use of the Resampling function and on subsequent "export regions to file."

I've used the program recommended base line settings for Resampling. But I've been experiencing some sharpness in high frequencies that I wonder are not influenced by these settings. Note all of my dubs on either a Tascam DA-3000 or a Tascam DV-RA1000HD are vinyl recorded at 96/24. Then I process in IzoptopeRX9 to Red Book for making CDRs for home and automobile use.

So my question is what have those of you using this software been doing on settings for 1) filter steepness, 2) cutoff shift and 3) pre-ringing? What are the pros and cons to changing the default settings?

Once I process in the Resampling step, I export regions to file, having prior marked all individual tracks. I select 16bit and Izotopes MBIT+ dithering routine.

Thank you.

 

I wish I could help you with the resampling, but I rarely do anything above 44.1 these days. But, with dithering, I always experiment for the most transparent sound for each project.

 

Can you record at 24/88.2?
That's what I do, then I resample to 44.1, then dither to 16bit, all with the default settings.

 

I never use RX for downsampling to CD rates from higher ones. I use Ozone 9 for that because I like to engage the Maximizer and sometimes EQ to get as many bits of resolution I can in the final 16-bit domain without clipping. Ozone has fixed settings when downsampling on an export except for dither on/off. RX can normalize to any final level without clipping so from a recorded track accuracy stand point it is the better way to go for exporting at lower sample rates and word lengths.

 

I have Ozone9 but I don't know how to use it in the application you note. Are you willing to walk me through that? Thank you in advance.

If I read y0ur process correctly you resample in Ozone then come back to RX to export? Sorry, I'm a bit confused by what you noted in process.

 

I can. But is that not the same as recording at 96/24 then using the defaults to resample to Red Book and then dither to 16bits on export?

 

Thanks. I dither down with MBIT+ because of what Barry Diamant told me. I have not experimented. Sounds like I should.

 

I do use MBIT+, but I experiment with noise shaping.

 

My apologies for not being as clear as I should have been. After I have recorded my vinyl tracks at 96K/24-bit I always bring those files into RX first. I edit out extra unneeded dead space at the front and end of the track. Then I perform all the noise reduction I think it needs (mostly declick, but sometimes Spectral Denoise, and Spectral Repair). Lastly I insert Cosine fades at beginning and end for smoothness on startup and ending. I don’t do any normalization in RX, but sometimes even out large differences between loud and very soft sections with Clip Gain adjustments (so they will have intelligibility in car listening or other environments). Once in a great while I might use Music Rebalance to raise the levels of vocals or instruments I think are too low in the original mix. Then, I export files with a postfix at the end of their names like “_RX”saving into a folder I create or one that already exists. Then, I bring the cleaned-up file into Ozone 9 for final remastering steps such as EQ, Dynamic EQ, Imager, etc, but always Maximizer for limiting and rescaling loudness up to near 0dBFs without clipping. I like Ozone for this because I can load up a lot of tracks and export them all at once with re-sampling and word length reduction to .WAV files (can use to burn CDs). My .WAVs are actually kept only as backup after I convert them to Apple Lossless .m4a files for my Apple Music server on a Mac Mini. That sits in the den to play music through our Apple TV in the living room audio system but is also used to keep my playlists up to date on iphone, iPods, and iPads. I welcome further questions if you have any.

 

The default "CD" settings for resampling and dither are better than custom settings, IME. I tried steeper resampling curves and more aggressive dither settings, and all of it wound up being inferior to iZotope's defaults after version 5, I believe. You're doing more damage with the spectral effects and Music Rebalance. Any effect that applies to the entire waveform is going to change the sound.

Also, are you recording with iZotope, or using a different app for that? When recording with iZotope on Mac, my recordings go straight into 32-bit float in the app, even though the ADC only provides 24-bit int. 32-bit is better for editing, though (192 would be better for audio capture as well, if your ADC supports it, though the higher sample rate is probably not as useful for edits as being in 32-bit).

Exporting markers should have no audible impact. It's just an exact copy of a portion of the waveform in a separate file.

 

I appreciate very much your work trying different dither and resampling options. It s good to know that the defaults are perfectly fine. The app I use for recording is Pro Tools Ultimate on a Mac. I have a fully balanced signal path from my MC cartridge pins through a fully balanced phono pre and then into my Avid 192 I/O professional A2D/D2A interface hardware.

I very, very, rarely use spectral repair within the body of the music unless there is some kind of high frequency white noise bursts or those strange vinyl “scraping” noises that declick can’t repair at any setting. About 99% of the time they are masked by moderate to high music levels. They usually reveal themselves during the very quietest passages but most often as the track gently ramps up or slowly fades out. That’s where I use spectral repair with the freeform selection tool and try replace or attenuate modes. Other strange noises sometimes occur at very low frequencies like “thumps” and these can often be heard during music levels. I have a few saved declick settings that will sometimes repair them but more often it creates artifacts and the noise floor starts singing where the thump once was. Then the only way to transparently fix them is with spectral repair in “replace” or “attenuate” mode with multiple trials. Sometimes just doing an edit “cut” of a thump with the rectangle selection too (that is easily seen in the bottom of the spectral view) works well. If there is a lot of rough hiss or tonal grunge during fade-out I find too audible I will use spectral denoise with a learn on the noise once the music has stopped and next do trials with different strengths and region boundaries. I just had to do that the other day on a track. With an additional carefully placed cosine fade-out for the very end I was able to get a pretty clean and non-distracting finish to the song. But, I never use denoise throughout a song. Any residual analog tape hiss is best left alone so all of the original recording is preserved. I have some extremely gentle declick presets that I have developed by testing them on music transients listing to what is being removed to know that they will only remove ticks. not transients. I listen to the entire song to see if there are any very fast HF transients that might be altered by my presets and then only apply them during parts of the music that are free of that kind of content, then go after any ticks/clicks manually removing each individual one.

Per the comment on the “damage” caused by rebalancing some stems, I do respect the majority ambition of the participants here to preserve the purity of the original recording/mix while removing only audible noise in a manner that does not disturb the music. I’m in sync with that most of the time. There are outliers though where I love the music but not the mix like maybe some great vocal work is buried behind the instruments, or other stems out of balance. For those I’m willing and able to use intelligent software tools like Rebalance to lift up the vocals a few dB, or when I feel a track needs more extended remixing I’ll use RIPX software to breakout more individual stems like vocals, bass, drums/percussion. guitar, etc and do a more laborious full remix that will result in much higher enjoyment when I listen to the finished track. I’m not claiming that I can do or have done a better job that the original production crew. Its just that I find greater enjoyment of the music from my efforts, and it is an awful lot of fun to do.

 

Yeah, I do come from the position of attempting the most faithful capture possible, so the "damage" comment does come from that standpoint, not from one of judgement.

In my experiments and attempts to avoid altering the overall sound, I've become extremely wary of most of the tools, as even a 0.0 sensitivity declick on the whole waveform of a 32/192 recording seemed to flatten the apparent soundstage depth. So at most I do very targeted manual declicking (extremely tedious), then resample for my own uses. Given the playback apps that I use having good dither implementations, I usually use 32/44 for personal use.

I would recommend trying recording at 192kHz, though, even if you ultimately only do playback at 44.1kHz. It provides a more realistic capture, and allows the resampler to make better guesses. Try it yourself. Record the same tune at both 96 and 192, resample both to 44, then compare and see if you hear the difference.

 

You can use Ozone just as you use RX for resampling and dither ing. But, as he noted, Ozone gives you more options.

I make use of the Ozone Maximizer in every project I do. It takes experimenting, but you can extract as much quality out of your files as possible. And, I don't care what the purists on this forum think: if you are careful, you don't have to settle for a low-volume (quiet) master. And, screw the LUFS -14 settings. Bump that baby up to -12, -11, or -10 LUFS if you can, and keep your sound clean and dynamic!

If you want to try Ozone, just download it, install it, and access it from RX through the "Plug-in" tool.