Your Vinyl Transfer Workflow (sharing best needledrop practices)*

This is my practice. I use it simply for digital playback of recordings on my desktop system.

I am thinking of replacing a basic Behringer UCA222 with a Motu M2 to act as ADC from unbalanced line out on a phono preamp (or record out on integrated), recording using Audacity running on Linux. I have read around and dipped into this thread and I think I can use RCA->TS plug adapters into the two line inputs on the M2. Am I correct?

As I see it the M2 will have more headroom that the Behringer (which suffers from too little on some hotly recorded records which causes clipping which I have no control over), I hope that the ADC is better. There may be other advantages.

Can anyone confirm I’m on the right lines with this setup? Has anyone any tips for getting decent results? I don’t do much cleanup after recording. Apologies for the basic nature of this query as I know some of you are way beyond this level of recording!

You can look at my profile if you need any info regarding what I’m working with.

 

Yes, that works fine.

Make sure to do the cleanup before recording - the cleaner the record, the better the result usually is.

 

Thanks - that is what I wanted to be confirmed! It is surprisingly difficult to find information relating to how to connect the stereo amp RCA line outs to the M2 or similar devices given they are generally aimed at recording instruments. I also understand the source has to be as good as possible.

If I get the M2 I imagine I would adjust the gain on the Motu to avoid clipping (using the handy display). Any further advice on gain gratefully received. I will generally adjust Linux PulseAudio recording device to apply 0dB gain (or slightly below) to keep things from peaking beyond -6dB. As I say, I don’t generally do any post-processing other than the occasional use of the Audacity Effect menu->Repair feature to deal with any obvious and isolated clicks. No further gain/normalisation applied in general. To date I have generally recorded at 48kHz and 16 bit as that’s all my existing ADC is capable of.

Please let me know if there are any obvious blunders to avoid or anything different I should be doing. Again, I appreciate this is needledrop 101. I have read posts in this thread which are beyond my current knowledge but am prepared to be schooled

 

My Motu M2 arrived and I set it up today. It was surprisingly easy to set up and use. I'm using Audacity on Linux and nothing else. The chain is Rega P8/Apheta 3 > Rega Aria > Rega Elicit-R > Motu M2. The Motu is connected to one of the Record Out on the amp (- I didn't bother connecting directly to the Aria).

Anyway, I made a couple of recordings. If anyone cares to listen and make suggestions as to how I could improve the basics, please do - I appreciate it. The objective of any recordings I make is simply for me to listen to on my desktop system. I'm not looking for perfection but I'd like to avoid simple mistakes or add some better practice.

The first track is recorded at 48kHz. The Audacity Click removal tool has been used with default settings. No other processing in Audacity. The track Pitch the Baby was recorded from an original but not exactly quiet copy of Cocteau Twins Heaven Or Las Vegas.

Dropbox - Cocteau Twins-Pitch_the_Baby.flac - Simplify your life

The second track is recorded at 96kHz. No post processing in audacity - no click removal. The track Gestures is from Carmen Villain's Only Love From Now On (2022).

Dropbox - 2023-01-07 - Carmen Vilain Gestures.flac - Simplify your life

In general, is it worth recording to 96kHz or higher from LP if the intention is not archival or for a lot of further processing? Does the bit depth of 32 bit floating point make sense or should I select 24 bit PCM?

For users of the M2, how do you manage to set the two independent gains on the two inputs to avoid imbalance? I played it by ear and eye to avoid peaks too far past -6dB.

These are simple or basic questions but I appreciate there may be no simple answers.

I will remove the tracks from Dropbox after a short while. If you listen - thanks! No obligation to comment!

 

Same as you: I set the levels by ear. The volume knobs are not exactly level, which complicates things a bit.

 

Recording at 96kHz allows one to see distortion. This can contain information that allows further manual processing or analysis, such as viewing the transient spikes of dust clicks or the spurious splatter of mistracking siblance. There is no need to actually hear these contents, so audio can be downsampled to 48kHz for consumption, or just recorded at 48kHz if you won't be doing anything advanced with the audio. 48kHz ensures you aren't simply recording and reproducing ultrasonic noise.

We can see in the mellow second file, no content above 20kHz - a digital recording on the vinyl, plus the click filter likely removing any transients (the rise above 25kHz is the ADC's noise shaping):




32 bit is nonsensical. The AK5552 ADC chip spec gives noise -115dB and dynamic range 115dB for particular signals, but a combined THD+N of 106dB. That's 18 bit. The rumble and noise of any phono preamp you can plug in will be above 16 bit level. We only use 24 bit because it is academically a format where multiple stages of processing doesn't degrade the audio, but you can save the final product in 16 bit.

You can use a 1-to-2 Y cable to provide the same electrical signal to both Motu inputs. Then monitor for identical levels (easier to evaluate if you play a test tone). You can then put the Y cable on the phono input from your turntable, and ensure even your phono preamp is included in the balancing act.

Quality, the music sounds fine. You cut out the silence, so we have to use a downsampled FFT, sized to isolate a sliver of the file's end - to see 50Hz and 100Hz line noise and harmonics distinguishable above rumble, and the tonearm resonance right at 10Hz.

 

Thanks @Stan94 for the user feedback! I look forward to using this device and will happily compare notes at any stage.

Thank-you @harby . You packed a remarkable amount into that answer but exactly the sort of info I was looking for.

For my purposes, at this stage I have no intention nor the skills, knowledge or tools to perform the advanced processing so 48kHz is sufficient. I must check but I think the 32 bit floating point must have been a default (I did not set it). I will select 24bit for my purposes but if I understand you correctly, even recording 16bit would still be sufficient.

There was no de-click on the second sample (CV, 2022 release) but your observation regarding the 20kHz cut-off for content is interesting and makes sense. Yes, there is no doubt that is a digital recording.

Your suggestion to use a Y splitter to help with the setting gain levels is a great one.

I should have included a bit more silence at the start and end but what you say about the graph makes sense.

Thanks for your comments re the quality. Maybe I could have selected better source material but I think they’re not unrepresentative. I didn’t try to select any “best sounding record”.

I haven’t compared to my previous efforts using a Behringer UCA222 but I am fairly sure this device will give better results. It has gain control and a lot more headroom. The UCA222 clipped with hotly cut records.

 

I use the M2's level meters to ensure I have plenty of signal coming in without the M2s meter turning yellow orange red. Then I switch to my FM tuner and use interstation noise to set the balance. Then I set the recording level using the metering in my recording program. If you don't mess with the the M2 input levels you can adjust your program recording levels only when you record other material.

Nice needledrops by the way.

 

I record at 44.1Hz per channel (88.2 for both channels). I use the spectrogram to check if there's any information above 20kHz: if there's something I keep it, if there's nothing I resample down to 44.1.

 

Ok - I was thinking I might get away from adjusting gain using the Linux audio subsystem but it does make some sense given the fiddly nature of having to adjust left and right gain independently with the device. I’m not sure if one way of setting gain is better than another or whether both have their place.

I do have the option to change to higher gain on my phono pre-amp. In fact the manufacturer recommends the higher gain setting (69.3dB) for that cartridge. I lowered it (to 63.5dB) partly to give a bit more headroom but also for reasons mentioned in the previous post, clipping when recording with the previous device. The setting is via dip switches which are fairly inaccessible (with my setup) so I won’t be changing them often or without good reason.

Thanks for your encouraging words re the recording.

Edit: fixed the phono gain numbers!

Edit2: That’s another great way to set the L-R gain evenly. I don’t have an FM tuner hooked up but I get the idea.

 

Is this a setting within your recording software? I notice that with that device if I change the sample rate within Audacity (just one global setting AFAIK) there is a gap when you start recording (I assume) due to the M2 being sent and instruction to change sample rate which causes a hiccup in its sending data for a moment. (i think I read that during my pre-purchase research and observed it today so made an assumption that’s what was happening).

 

That's not how it works.

 

Don't tell my 192kHz x8 = 1.5MHz...

BTW (and to use that bandwidth), has anyone experimented with parallel recording with a large number of channels, then combined them back in software? I've got one 8 channel DB25 balanced that is 15k input (114DR/-105THD+N), and one 8x that is 50k single-ended 1/4" (114DR/-100THD+N). Cirrus multibit sigma-delta without the noise shaping ultrasonics of new chips.

"For N amplifiers in parallel, the amplifier noise power is reduced by N and the input referred voltage noise density is reduced by √N. Put another way, each time the number of amplifiers is doubled, the amplifier noise power decreases by 2 and the amplifier's input referred voltage noise density decreases by √2." = beyond Motu M4 on 15 year old hardware.

Custom 1-to-4 adapter cables is a lot of connectors for an experiment in SNR way beyond vinyl...

 

Reminds me of a classic saying: “Just because you can doesn’t mean you SHOULD.”

 

Is Ozone a stand alone application or do you have to host it in another software? I have installed it on a PC and can find no executable files to launch it.

 

From the Google search, "Is Ozone audio program standalone?"

"Surprisingly, Ozone 10 can no longer be used as a standalone application, and is now only available as AAX, AU and VST3 plugin formats."

"iZotope Ozone 9 can be used either as stand-alone application or as a plugin inside your DAW."

Depends which you have. O10 NO. <O10 Yes

 

You are not going to believe this but I spent over an hour on Google trying to figure out how to run it

 

I can believe it. People say Google is your friend. I say Google isn't a very good friend.

 

I make my recordings on a Mac mini computer with Audacity software. The audio input of the computer is set at 24/88.2, so is the sample rate within Audacity. The Motu driver is not installed since it made Mac OS hang.
In Windows, there is a delay in playback when changing the sample rate on the fly (or if you switch between two files with different sample rates) WITH the Asio driver; if you're listening thru the regular audio codec then there's no delay when you're switching between 16/44 and 24/96 for example.

 

I too think Isotope should have left Ozone 10 with the same standalone option as V9

 

Would that be the Cirrus 5381 with a multi-bit delta sigma modulator? Have seen quite a few academic papers published proposing different multi-bit architectures for A2D converters. Some consumer hardware out there now employing multi-bit D2A chips too. I upgraded my external D2A box fed by my MacPro’s optical output to a Schiit Audio Bifrost2 D/A that uses the TI DAC8812 converters. The time and frequency domain optimized digital filter is their own custom one realized inside a Analog Devices Sharc DSP IC.

 

4x CS5361 for the first (CS5381 is a pin-compatible spec upgrade).

"The CS5361 uses a 5th-order, multi-bit, delta-sigma
modulator followed by digital filtering and decimation.
This removes the need for an external anti-alias filter.
The ADC uses a differential architecture which provides
excellent noise rejection."​

The "everything upsampled to DSD" DACs are the ones to be wary of, as that's just marketing pushing old technology.

DACs now have their own digital filtering, often selectable by different registers (possibly not fully presented to end users). "Upsampling" to one is just filtering twice. The DACs have firmware allowing different filter IP from the manufacturer to be uploaded, again, rebranded deceptively.

 

My latest thing is to play the LP using my Parks Audio Puffin with "Magic" ON. From there, I take it into my PS Audio NPC analog inputs. Kinda weird to go analog to digital to analog to digital to analog, but it sounds great to me.

 

You might look into the spdif output mod and record that.

Kinda weird to process recorded audio in a box when free software (SoX.exe input.wav output.wav RIAA) can do the same.

 

The Bifrost 2-64 allows the user to bypass their custom digital filter and see which mode is preferred by listening comparison tests.