I tried about a year ago to rip some records I bought, without much success. I think I’m going to give it another try with your help. Here goes…
I’d like to end up with a 16/44 rip and a 24/96 rip. My sound card is capable of recording at 24/96. Would it be better to record in audacity with 32-bit float and then change (dither/resample) to my desired rate, or just stick with recording in sample rate I want to end up with? The last time I attempted ripping, I dithered with Izotope RX and for some reason when I tried to encode the final WAV with FLAC it gave me an error. Is there a recommended way of downsampling?
Thanks for your time, and please excuse my ignorance of vinyl and audio.
By default, Audacity works in floating-point internally and you should leave it that way. Soundcards (ADCs & DACs) are integer devices but the software will make the conversion.
I’d like to end up with a 16/44 rip and a 24/96 rip. My sound card is capable of recording at 24/96.
There’s no harm in using 24/96 other than large files. But, it’s not necessary with analog vinyl. Think about it… You can easily hear vinyl defects, especially noise but sometimes distortion and frequency response variations, but you can’t hear any limitations of 16/44… CDs are better than human hearing and you can make a CD or digital file that sounds like vinyl. Records are worse than human hearing and you can’t make a vinyl record sound like a CD.
Is there a recommended way of downsampling?
Just change the sample rate (lower left corner of the Audacity window) and then set the bit depth when you export. (Some downsampling algorithms “measure” better than others, SoX has a reputation for being one of the best, but I’ve never heard any difference, no matter what software I was using.)
Dithering is normally recommended whenever you downsample, but you shouldn’t hear any difference unless you dowmsample to 8 bits. Plus, dither is low-level noise so you’re just adding to the existing vinyl noise.
I’m guessing you had trouble with FLAC because you changed some setting in iZototope and ended up with a really unfortunate setting for downsampling and dithering.
If you don’t work with the audio, 16 bit 48 kHz is all you need for playback.
If you count on doing some noise removal, eq or filtering, record at 24 bit and export 16 bit when you’re done.
Take into account that even if your interface is capable of 96 kHz, the filters might not be optimal for that sampling rate. Most professionals only use half the capacity of the interface, to stay away from filter problems, unless they use really high-end gear. So if your interface supports 96 kHz, 48 probably gives the best results.
Hoping you can answer a similar question…let’s say I want an end result of 16/44. Would I bet better off (quality wise) recording at say 24/32 bit depth - 96khz sample rate, and then downsampling to my desired 16/44. Would this result in a higher audio quality as opposed to recording natively at 16/44?
You aren’t recording at 24-bit unless you have a 24-bit interface and you’re choosing Windows WASAPI host. Otherwise you are recording at 16-bit which is being converted to 24-bit resolution if Default Sample Format is 24-bit or to 32-bit float if Default Sample Format is 32-bit float.
If “bit perfection” is your requirement, and you are not running effects that change the sample amplitudes, you could set Default Sample Format to 16-bit, set project rate to 44100 Hz, and turn dither off in Quality Preferences (dither should not be applied when exporting to a 16-bit audio file from a 16-bit project, but it’s applied due to a bug).
If you are running effects that change the sample amplitudes, you should set Default Sample Format to 32-bit float because Audacity processes internally in 32-bit float. Otherwise you will get lossy downconversion to the track rate with every effect.
Some may argue that at 96000 Hz you will benefit from analogue anti-aliasing, but https://xiph.org/~xiphmont/demo/neil-young.html argues that down and generally argues that extreme sample rates are counter-productive. Have a read of that and make your own mind up. What is unarguable is that 96000 Hz takes more than twice the disk space of 44100 Hz, and so there is increased risk of recording dropouts if the computer is slow.
Ok, so I have a very limited understanding of audio and I’m sorry that I’m having trouble following. My takeaway from your replies and reading the xiph.org article is that it doesn’t make sense to listen to music at bit depths higher than 16 or sampling rates higher than 44/48. However, if I plan on editing the audio i.e. removing clicks/noise/hum (Izotope or ClickRepair), it might make sense to Record at 24/96 and then downsample to 16/44 for my final product? Oh and with regards to dropouts, I’ll be recording on a SSD, So I should be fine, is that correct?
My first major use of Audacity was to convert my vinyl collectionn (and subsequently my wife’s vinyl collection).
The objectives were to produce good quality compressed audio on my iPod but also to provide high quality uncompressed files to play on my hi-fi rig (old kit but still comparatively high end QUAD 33/303 feeding QUAD ESL-57 electrostaticspeakers)
My turntable was fed to an ARTcessories phono pre-amp which in turn fed an Edirol UA-1EX external USB soundcard and this into a USB port on my PC (I did start out with a cheapy USB turntable but the sound quality was rubbish).
I recorded with Audacity set with its default quality settings of 32-bit float, 44.1 kHz - the Edirol soundcard worked at 16-bit.
I did very little post-processing apart from claening the track ends and beginnings and the inter-track gaps. I started out processing the clicks and pops by hand and that was very time consuming - but after a steer from Koz I started using ClickRepair, that saved me loads of time and produced excellent results - see this sticky thread: https://forum.audacityteam.org/t/click-pop-removal-clickrepair-software/1933/1
I exported a 32-bit WAV file from Audacity immediately after capture and processed that file through ClickRepair to produce a cleaned 32-bit WAV file which I imported back into Audacity for further processing and export.
When I exported I used the default settings of 16-bit PCM stereo WAV - I used triangular dither setting at the time but now I would recommend shaped dither (which I think is the default).
So the upshot: what I managed to produce were some really excellent results which sounded as good as listening to the vinyl through my hi-fi rig - and in many cases better (beacuse the clicks and pops were removed).
Again, you will not record 24-bit unless you have a 24-bit soundcard and choose WASAPI host - otherwise you will record 16-bit which will be expanded to 32-bit float if you set Audacity’s Default Sample Format to 32-bit float.
I would definitely use 32-bit float Default Sample Format if you are using Audacity to do click and noise removal or any other editing other than deleting silence or noise. If you are doing all filtering in other applications you may prefer to use 16-bit Default Sample Format assuming you only have a 16-bit soundcard.
Personally I would not bother with 96000 Hz unless you do a blind test and can tell the difference between that and a recording at 44100 Hz. It is “possible” that whatever is the default rate for the soundcard will sound better than other rates because that is how the card has been designed. Look at the documentation for the card.
Yes you should be OK for avoiding dropouts if you have a Solid State Drive (SSD).