ripping vinyl at 24bit 96Khz with Audacity

Hi,
New forum member here, thank you for having me
Having invested over the years in a very high end vinyl replay system I am interested in ripping some of my more collectable LPs to digital.

I have Audacity 3.0.2 running in Windows 10 on a Lenova Yoga laptop, ultimately file storage will be on my QNAP NAS.

I am capturing the audio via a line level output from my hi-fi to a Presonus Audiobox 2.0 interface connect to the laptop over USB. So far so good (it is working,this post is about optimising it)
In the audio interface I have the resolution set to 24bit 96Khz in the control panel for the Presonus
In the Audacity project I created 1 stereo track which is also 24bit 96000Hz
and off I go… (it works, I have successfully ripped my first LPs)

What I want as an outcome is 24bit 96KHz Wav files stored on my network drive,
It seems Audacity for Windows does not record natively in WAV so I then use the “Export as WAV” function selecting “Signed 24bit” encoding option
Again all good, I have tested it and have made a WAV file I can now play in any media player and stream back to my HiFi, the media players confirm the WAV file is indeed 24bit 96KHzresolution and it sounds good to my ears

So all seems good, but I want to ask the forum members if there are any further optimisations I can do, by optimisation I mean removing any processing. I do not want to transcode, i do not want any volume levelling, AGC, sample rate conversion, bit depth quantisation etc, I would hope that the WAV file is a bit transparent image of was captured by the audio interface and recorded in first instance by Audacity

Does Audacity force a transcode when you export to WAV? are there any options to record natively in WAV? have I followed the correct path so far?

Tips welcome
Thanks
Guy

Audacity works internally with “32-bit float PCM”, which is capable of exceptional sound quality.
24-bit WAV files contain 24-bit PCM data.

When recording 24-bit audio data, the conversion to 32-bit float is perfect (totally lossless). You can think of it as “padding” the data with some extra zeros (it’s actually a bit more complicated than that in the technical detail, but that’s a rough idea of what happens).

On export, Audacity converts back to 24-bit (or whatever format you use).
The extreme high quality that Audacity uses internally means that Audacity can perform processes such as amplifying and fading without rounding errors.

It is worth noting that while marketing hype will tell you that “hi res” audio is superior, in practice that cannot be guaranteed. Higher sample rates are more prone to IM distortion from high frequency interference, and higher sample formats (> 16-bit) are only really useful when the analog noise floor is below about -70 dB. I have observed cases where the sound quality from some hardware is actually better at 44.1/16bit than 96/24bit (not audibly better, but measurably better).

Usually, the difference between 16-bit 44100 Hz and 24-bit 96000 Hz is indistinguishable in blind (ABX) testing. In some cases there “may” be very subtle improvements at 24/96, but it is dependent on the hardware being used, the original audio being sufficiently high quality, and the person listening having exceptionally good hearing. Is such a small (possibly insignificant) difference worth three times as much disk space? For some it is, for others it isn’t. Personally I can’t hear the difference between 16/44.1 and 24/96, so I use 16/44.1.

Your setup sounds good.
Using “Amplify” or “Normalize” will not damage the audio at all and can be safely used. All that these two effects do are arithmetic operations on the numeric sample values, which at 32-bit float are extremely accurate (much more accurate than the A/D, D/A converters in the Presonus Audiobox).

Thanks, I’m glad you think my set up sounds good, re hi res vs cd resolution, i’ve worked in and around audio professionally for 35 years and I have been involved in several initiatives over those years in delivering “better than CD quality” / “hi res” to market, the actual validation in terms of sound quality over that time has been mixed, I can tell stories of minds being blow by quality only to find it was 16bit 44.1 and similar stories of hi res audio impressing on other occasions, i’ve seen it all and heard it all, my personal beleif is that dynamic range control and mastering are where it can go a bit wrong and well mastererd 16bit sounds amazing. Reason for digitising my LPs at 24/96 is kind of “why not”, ive got the hardware to do it… ive got the storage space, so might as well give my files the best possible start. I’m not going to die in a ditch defending hi res however for all the reasons you write
:slight_smile:

I absolutely agree. A badly mastered recording will still sound bad no matter how high the “resolution”.

That sounds perfectly reasonable to me.

24-bit also has the advantage that if at a later date you decide that you do want to apply some effects, or transcode into a different format, then you are starting “closer to the original” than if you were using a 16-bit format. Although the differences are still likely to be very subtle, additional processing may make previously inaudible differences noticeable.

24-bit FLAC format is a popular format for “archive” recordings. It has exactly the same quality as 24-bit WAV (and can be converted to 24-bit WAV without any losses), but has significantly smaller file size and better support for metadata. The down-side is that some audio players still don’t support FLAC, despite it being around for over 20 years!

Many years ago early b90s when CDs were just out I played some music for a couple of friends - it was an LP that I had “ripped” to tape (TDK Super Avilyn on my Nakamichi BX-2) played back on the Nak though QUAD 33/303 and QUAD ELS-57 electrostatic speakers. One of the friends commented

These CDs sound really amazing don’t they.

I didn’t like to disillusion her …

Peter.

I would hope that the WAV file is a bit transparent image of was captured by the audio interface and recorded in first instance by Audacity

It should sound identical (assuming nothing “goes wrong”). [u]Here is a short story about an informal demonstration/experiment[/u].

With some click & pop reduction the digital copy can sound even better than the original!* And some older records sound a little “dull” so I sometimes boost the highs a bit. I ASSUME modern records are more consistent but back in the vinyl days sound quality varied a lot and most were mediocre, at least the rock records… The rumor was that classical records generally had better quality, but that’s not what I was listening to.

Your choice of recording software doesn’t affect quality (again assuming nothing goes wrong). The software just has to “capture” the digital audio stream and send it to the hard drive. And, Audacity doesn’t have the ability to add effects in real time effects during recording. (Windows can mess with the audio, but when that happens it’s usually drastic AGC, etc., and it’s pretty obvious. And Windows will automatically
re-sample if your hardware is not capable of matching the software settings.)

I would hope that the WAV file is a bit transparent image of was captured by the audio interface and recorded in first instance by Audacity

“Full disclosure”: By default Audacity will [u]dither[/u] when you export, but you can turn it off. Dither is very low-level added noise which is supposed to sound better than quantization noise, but under normal conditions you can’t hear quantization noise, dither, or the effects of dither at 16-bits or better anyway. Plus, vinyl noise is much greater than dither noise so you can consider it “self dithered”. The “rule” is to dither whenever you downsample and since Audacity works in 32-bit floating-point it doesn’t “know” if you are really downsampling or not.

What I want as an outcome is 24bit 96KHz Wav files stored on my network drive,

The only issue with WAV is that metadata/tagging is not well-standardized or widely supported. Virtually all software players support FLAC. FLAC also makes a good archive format because it’s lossless so it can be converted to any lossless or lossy format anytime in the future, it does support metadata, and the files are almost half the size of WAV.

Speaking of metadata… Audacity can’t add the album artwork. I use MP3Tag (which works with almost all formats, not just MP3). I also just find it easier because I can select several files at once for the common information and then I usually just have to enter the song title and track number for each song. Usually you can find the artwork somewhere online. If you have to scan it, an LP cover is too big for a standard scanner but you can use "photo stitching’ software to patch-together 4 separate scans.


\

  • You can “clean up” the digital copy to reduce/remove “snap”. “crackle”, and “pop” so the digital can be better than the vinyl (although you rarely achieve “CD quaility”). Audacity has 3 tools that can help:
    The Click Removal Effect is automatic.
    The Repair Effect is manual.
    As a “last resort” you can zoom-in and use the Draw Tool to manually re-draw the waveform.

For background hiss & hum you can use the regular Noise Reduction Effect. Noise Reduction can be tricky because you can get artifacts (side effects) depending on how bad then noise and how aggressive the the noise reduction is. (The same is true for click & pop removal.) Sometimes I’ll use Noise Reduction just on the fade-in and fade-out and live with any artifacts. (Of course you can mute completely between tracks.)

Or there are specialized 3rd party applications:
[u]Wave Corrector[/u] is fully automatic and it’s free.
[u]Wave Repair[/u] ($30 USD) works manually.

I’ve used Wave Repair for several years. Since it’s manual means it only “touches” the audio where you identify a defect. But of course, it’s VERY tedious and time consuming. I’ve spent a day or a weekend fixing-up an LP. It offers a few different repair methods and in most cases (but not all you can get audibly perfect results. Sometimes the worst clicks and pops are the easiest to fix, maybe because they are easier to “find”. Another thing you might not like is that it only supports 16/44.1 or 16/48. But if you have a bad click that you can’t fix otherwise, it would be worth reducing the resolution.

The developer of Wave Repair has a [u]website[/u] with several other software recommendations (maybe somewhat outdated) and a TON of other information about digitizing vinyl. One application he doesn’t mention is Izotope RX ($130 USD and up). It gets great reviews but I’ve never tried it.

Not really relevant for 24-bit because the dither level is so low (about -130 dB, or to put it another way, about 100 times lower than the quietest sound that can be produce in CD format).


Sort of, but there’s a difference.
Vinyl surface noise will mask quantization noise, whereas dither noise prevents quantization noise.

I did (and still do) listen to rock & classical

Deutsche Grampohon produced some of the best classical IMVHO, really great engineering - and the ones I had converted nicely to digibits with Audacity …

Peter.

thank you all, so far I have not spent time de-clicking, this is an unfettered rip after exported to WAV, feel free to have a listen and give me your comment
https://files.mycloud.com/home.php?brand=webfiles&seuuid=a08f68c565d478e373fcdb72228287a4&name=Yello___Toy___Side_C