Newbie Question on Digital Audio up/oversampling on Analog vs Digital media. - AVS Forum
Forum Jump: 
 
Thread Tools
post #1 of 5 Old 11-21-2012, 06:19 PM - Thread Starter
Newbie
 
redison's Avatar
 
Join Date: Nov 2012
Posts: 2
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
Calling all experts, I will try to be brief.


I been getting interested in digitizing my old album collection from the 60s 70s and 80s. I've also been doing a little research about digital audio, particularly about sampling and bit depth. But I'm new to all this.

I'm using Audacity on a Mac Pro with a good turntable attached to the line in and doing some test recordings. So far everything's working well.

However I got to thinking that maybe I could do the same thing with my CD collection but rather than " rip them" into MP3s which have already done 10 years ago, maybe I could "upsample" my CD collection as well, as my vinyl records.


So here's my question

I can see the value of upsampling or oversampling above the CD rate of 44.1/16 on analog media such as my vinyl records [ See note 1 below ].

However even though you can convert digital media to higher sampling rates and bit depths and run them on certain equipment if you have it, I don't see what you're really gaining in doing so because the source is already "digital" so in the case of a compact disc which is obviously digital no matter what sampling rate or bit that the use your base source is still going to be 44.1 kHz/16. When I was discussing us with some friends I use the analogy of a digital photo, let's say at VGA resolution of 640×480 pixels and trying to upscale it to say HD 1920 x 1080 pixels, ( forgetting aspect ratio just for this example) to do so you'd need to fill in the information by adding pixels to the original 640 x 480 and it probably wouldn't be very good. I know there are programs like Photoshop and even Apple's iPhoto that will let you do these things but really these programs have to guess what color the pixels would be before adding them.

Actually this may be a really bad example I think just as I'm typing it.

To me the difference is that with digital audio you have a time component added compared to the static digital image.

Okay I think I probably confused everyone, but just to summarize I can see the value perhaps of the various sample rates and bit depths being applied to analog audio but I don't see how up sampling or oversampling digital audio can really make the sound any better, in theory anyway.


Note 1:

I have been reading about the Nyquist–Shannon sampling theorem and how it relates to digital audio suggesting that the sample rate needs to be twice "and then some" the number of the highest frequency which in audio usually means 20,000 kHz, and presumably there's nothing to be gained from sampling above that, which is why compact disks use a sample rate of 44.1 thousand kHz .... 2 X 20,000 "and a little extra"


Thanks for reading.
-Robert
redison is offline  
Sponsored Links
Advertisement
 
post #2 of 5 Old 11-21-2012, 07:30 PM
Senior Member
 
kraut's Avatar
 
Join Date: Sep 2012
Posts: 379
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 60
Quote:
, maybe I could "upsample" my CD collection as well, as my vinyl records.

You answered yourself as how much sense that would make...none.
Quote:
, and presumably there's nothing to be gained from sampling above that,
Correct, and just look at the spectral anlysis of a music peace and see how little info actually there is beyond 16 - 18 kHz.
Also - on CD payback a brickwall filter is applied to filter any frequencies not much above 20500Hz.

I have transferred L's to Wave and Flac and compared both. Higher rates than 44.1 and 16 bits is nonsense, waste of digital storage and makes no sense when taking into account the flaws of a medium like an LP. I have yet to hear an LP that does not have some surface noise, even aside from the lead in/out tracks and between tracks.

I also find it difficult to understand why anybody would rip in wav. instead flac or similar lossless compression formats. You do not loose sound quality, and it is far easier to tag in flac than in wav.
And you can produce CD's from any flac file without problems.

At 2000 Lps and the fact that transfer is 1/1, I have given up on transferring LPs, I rather find a download source.
redison likes this.
kraut is offline  
post #3 of 5 Old 11-21-2012, 07:41 PM
AVS Special Member
 
MarkHotchkiss's Avatar
 
Join Date: Mar 2007
Location: Long Beach, California
Posts: 1,476
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 82
Hi Robert,

Welcome to the forum.

I am not an expert on audio. In fact, I would consider myself a newby, compared to others on this forum. What I do understand is digital signal processing.

In general, your thinking is correct. There is no gain in oversampling your CDs. What is there is there, and re-sampling will not add anything to it. However, re-ripping into a lossless format, like FLAC, might be worthwhile, depending on the bitrate you originally ripped your MP3s to.

Your vinyl is a different issue, for two reasons. I've been digitizing a lot of vinyl lately, and have been using both 24-bit/96kHz and 24-bit/192kHz.

Reason one is that the digital-to-analog process requires absolutely no energy above the Nyquist frequency (half the sample-rate). So you need an analog anti-alias filter that will not start rolling off until it's above 20kHZ, but is effectively at zero by Nyquist. For 48kHz, that would be fron 20kHz to 24kHz - a really steep filter. Sampling at 96kHz greatly relaxes the requirements by allowing the filter to roll to zero over an octave, and 192kHz allows it to operate over two octaves. So the higher sampling-rate allows for a better analog front-end.

Reason two is that you often want to post-process the digital copy, to remove snaps, crackles and pops that may be on the original. Any DSP algorithms that you use to do that, or even if you edit them by hand, will operate better with the higher sampling rate. So, in effect, you are not improving the quality of your recording by using the higher sample-rate, but reducing the degradation that might occur from post-processing later.

24-bit is useful during the original recording to allow for the volume level to be normalized without losing any dynamic range. And if you do any post processing, 24-bits might be a minimum safe depth to start with. But a bit depth more than 16-bits in the final product might not be audible. I have no opinion one way or the other on that, as what we can hear is way outside my expertise.
redison likes this.
MarkHotchkiss is offline  
post #4 of 5 Old 11-21-2012, 08:41 PM
AVS Special Member
 
mcnarus's Avatar
 
Join Date: Nov 2007
Posts: 6,183
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 123 Post(s)
Liked: 324
For ripping CDs, resolutions above 16/44.1 are pointless.

For analog, Mark H. is right in theory. (Although just to be clear, to account for post-processing losses you want to increase the bit depth, not the sampling rate.) In practice, however, 16/44.1 will give you plenty of headroom for whatever processing you're likely to do, and no reasonable soundcard (including the one in the Mac Pro) will cut off any frequencies you are likely to hear, unless you are an exceptional teenage girl. Nothing wrong with using higher resolutions, of course, if you have the space.
redison likes this.

If you can't explain how it works, you can't say it doesn't.—The High-End Creed

mcnarus is offline  
post #5 of 5 Old 11-21-2012, 10:04 PM - Thread Starter
Newbie
 
redison's Avatar
 
Join Date: Nov 2012
Posts: 2
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 10
OK Thanks guys,

As far as compact discs are concerned, I'm glad to see your comments agree with what I thought. I'll probably still rip them but into 44.1/16 FLAC or ALAC files. The only problem I have with FLAC files is that they're not supported by Apple's iTunes or their iPod products, however it's a simple solution just to convert him to a very similar format called ALAC which, you guessed it is an acronym for Apple lossless audio codec. I imagine this is a simple licensing problem with Apple. :-) I have downloaded some live concerts from a site called amazingly " live downloads com" where some bands which will give you the option between FLAC or ALAC at the same cost, both at 44.1/16.

Okay so now on the LP vinyl records, using the program audacity I have it set up to initially record under our quality preferences of 96 kHz and a sample depth of 32-bit float. The 96 kHz I just chose for testing but you can choose rates from 8 - 192Hz (or presumably greater just by entering a number in the project rate field). the 32-bit float bit depth setting is quite interesting because from what I've read it allows essentially an infinite dynamic range, however the problem is not only the file size but the fact the very few DACs, even the really expensive ones will play this. That is okay because all I'm using the settings for is the initial raw digital capture which I'll save to file as an audacity project. I will archive these files and back them up obviously, file space actually is not a problem for me right now I'm running a 36 TB NAS server ( by the way I am storing other stuff and backing other stuff to it as well ).

Using these original files I can make multiple exports in audacity, right now I'm testing out exporting them at 96 kHz and 32 bit PCM, as well as 96 kHz and 24 bit PCM in uncompressed AIFF format.

Just for your interest in the file sizes involved here, they are all for one side of one record which is about 24 minutes, the initial RAW recording file at 96 kHz and 32 bit float ranges from 3.5 to 4 GB, and exported AIFF file at 96 kHz and 32 bit is about 1.1 to 1.4 GB and the exported a AIF file 96 kHz at 24 bit ranges from 700MB to 1GB. -Remember that these are all uncompressed AIFF, and I'm just playing around with some of the settings. I have saved a couple in 96/24 FLAC format and they come in around the 500 to 600 MB range.

OK now the thing is I'm not an expert at using audacity or any other digital recording software. I should be getting a book or manual in hard print about running audacity tomorrow, there are a lot of settings to understand such as " remove clicks " " normalize ".

Just to give you an example, the audacity online tutorial or manual highlights these editing actions in an LP workflow:



====================== Paste Begin ==============================


"Remove DC offset
DC offset can occur at the recording stage so that the recorded waveform is not centered on the horizontal line at 0.0 amplitude. Use the Normalize effect to remove any DC offset. Put a check mark in "Remove any DC offset..." but leave "Normalize maximum amplitude..." unchecked.

Remove subsonic rumble and low frequency noise
Use Effect > High Pass Filter... with a setting of 24 dB per octave rolloff, and a cutoff frequency of 20 - 30 Hz to remove unwanted subsonic frequencies which can cause clicks when editing. If your record is warped, this will definitely generate unwanted subsonics, in which case consider a lower cutoff frequency.

This step can probably be omitted given a flat record and high quality turntable, arm and cartridge.

Remove clicks and pops
There are a number of ways you can use Audacity to remove clicks and pops from your recording.

Use the Click Removal effect on either selected regions of audio or on the whole project. Preview the effect with different settings to get the best results.

Clicks which did not get removed with Click Removal can be treated individually with other methods. These are only really useful if you have a relatively small number of clicks and pops to deal with, otherwise the approach may be too labour-intensive and time-consuming:

Try Audacity's Repair effect. This repairs a very short length of up to 128 samples by interpolating from the neighbouring samples. You will need to zoom in to see the individual samples to use this effect.

For somewhat longer regions of audio, try:

Draw Tool. You also need to be zoomed in to the individual samples to use this. Some patience may be needed with this tool, but the principle is to put samples back into line with their neighbours so that a smooth contour is presented.

Effect > Hard Limiter.... This is an extreme compressor effect, but can be effective used on an individual click. There is no need to zoom right in to sample level to use this.

Remove hiss and high frequency noise
Get a noise sample from either the lead-in grooves immediately before the music starts, or from a lead-in between tracks. Apply the Noise Removal effect with Noise Reduction set to no more than 12 dB (9 dB is a good guideline), Frequency smoothing 300 Hz and Attack/decay time 0.25 seconds.

Noise reduction is always a compromise, because on the one hand you can have all the music and a lot of noise, and on the other hand, no noise and only some of the music. Try different settings on the "Noise Reduction (dB)" slider until you get the best compromise.

Whether you need to use Noise Removal will depend on the quality of your LPs and your stylus and cartridge.

Clean the inter-track gaps
These are rarely truly silent so you may want to replace them with silence by selecting the gap and using CTRL+L. Reduce the inter-track gap as desired to around a maximum of 2 seconds, though you may wish to use a shorter gap or even no gap at all for some recordings.

Note that CD burning software almost always adds a 2-second gap between tracks by default. Check for any options to turn this off, or for "gapless burning" or "Disc-at-once (DAO)" options that you can enable.

Adjust label positions
If you are using a 2-second gap, adjust the label position as desired to be 0.5 seconds before the start of the next track. To move the label, drag by its center circle.

Amplitude adjustment
Normalize the amplitude of the recording. Use Effect > Normalize as the last editing step to bring the amplitude to around -3.0 dB. The Normalize effect can be set to either:

adjust the amplitude of both stereo channels by the same amount (thus preserving the original stereo balance) or
adjust each stereo channel independently (this can be useful if your equipment is not balanced).

Compression
The Compressor effect reduces the dynamic range of audio. One of the main purposes of reducing dynamic range is to permit the audio to be amplified further (without clipping) than would be otherwise possible.

Compressor makes the loud parts quieter and (optionally) the quiet parts louder. It can be very useful for listening to classical music in a car. Such music normally has a wide dynamic range and can thus be difficult to listen to in a car without constant volume re-adjustment."

=========================== Paste Finished =======================



The reason I mention all of these is partly in response to part of kraut's reply :

"Higher rates than 44.1 and 16 bits is nonsense, waste of digital storage and makes no sense when taking into account the flaws of a medium like an LP. I have yet to hear an LP that does not have some surface noise, even aside from the lead in/out tracks and between tracks."

Which of course is true and I knew that from the beginning.

However even if these flaws can't be that much reduced, then for me and this is really just a subjective call, I'd still like the music portion to be as high resolution as possible.

And some vinyl recording simply never made it to CD or digital downloads, they are the equivalent of " out-of-print ".



Thanks again for reading and your replies.

-Robert
redison is offline  
Reply Audio Theory, Setup, and Chat

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off