Latest Thread Images

Why 24 bit audio and anything over 48k is not only worthless, but bad for music.
Jul 20, 2015 at 3:52 PM Post #706 of 3,525
interpolate said:
  It makes me want to throw all my high-res gear in the bucket 
popcorn.gif
Click to expand...

I'd recommend the FS forum, plenty of audiophiles ready for you to unload the kit on.
 
Jul 20, 2015 at 4:00 PM Post #707 of 3,525
Yeh maybe but I'm keeping the Unicorn and Fairy Dust.
 
Jul 20, 2015 at 4:10 PM Post #709 of 3,525
I converted some WAV 24/96 to DSD format (DSF) and you know the difference is indecipherable other than the processing power to output a signal. DoP tends to settle at 174K for some reason with jRiver Media Player.
 
Jul 22, 2015 at 2:42 AM Post #710 of 3,525
Decided to try this out. I just got a bunch of Michael Jackson's albums on CD and I wanted to see how they sounded differently from the HD masters.

"Butterflies" from Invincible
24/96 HD


CD


24/96 converted to 24/44.1 (-r 44100 -o "filename".flac via SoX)


Pretty close. I think I could pass it if I did it again. The differences are not nearly as large as other HD vs CD masters I've heard.

foo_abx 2.0.1 report
foobar2000 v1.3.8
2015-07-21 22:59:55

File A: 07 Butterflies-CD.flac
SHA1: bdae6fd77dc7540ec35d1e48b68565b10848e54f
Gain adjustment: -9.85 dB
File B: 07 Butterflies-HD-441.flac
SHA1: 790dbd114cffe43520335ce39cff258b0fa27ba5
Gain adjustment: -9.09 dB

Output:
DS : Geek Pulse X Infinity 1V5 Output, 16-bit
Crossfading: NO

22:59:55 : Test started.
23:02:31 : 00/01
23:03:22 : 01/02
23:03:50 : 02/03
23:05:03 : 03/04
23:05:21 : 04/05
23:05:35 : 04/06
23:05:48 : 05/07
23:06:24 : 06/08
23:06:35 : 07/09
23:06:48 : 08/10
23:06:48 : Test finished.

----------
Total: 8/10
Probability that you were guessing: 5.5%

-- signature --
8ce3f60e31b3ca2e52382cd507aa34b932f28ba5

24/96 converted to 16/44.1 (-r 44100 -b 16 -o "filename".flac via SoX)


Myth debunked; 24-bit audio makes no audible difference; 928 kbps average sounded just as good as 1632 kbps average. I was just guessing for these tests since I couldn't even tell A apart from B.

foo_abx 2.0.1 report
foobar2000 v1.3.8
2015-07-21 23:30:27

File A: 07 Butterflies-HD-CD.flac
SHA1: 00557c6ebe4256509eaab281d5af5b8d7454f238
File B: 07 Butterflies-HD-441.flac
SHA1: 790dbd114cffe43520335ce39cff258b0fa27ba5

Output:
DS : Geek Pulse X Infinity 1V5 Output, 16-bit
Crossfading: NO

23:30:27 : Test started.
23:33:26 : 00/01
23:37:31 : 00/02
23:38:06 : 00/03
23:38:45 : 01/04
23:39:13 : 02/05
23:39:34 : 02/06
23:40:07 : 03/07
23:40:28 : 03/08
23:40:38 : 03/09
23:40:45 : 03/10
23:40:45 : Test finished.

----------
Total: 3/10
Probability that you were guessing: 94.5%

-- signature --
e6a89011212e5129d89489bc68e6189151d7926b






"Smooth Criminal" from Bad (HD) and Bad [2001 Special Edition] (CD)
24/48 HD



CD


24/48 converted to 24/44.1 (-r 44100 -o "filename".flac via SoX)


Flawed. The timing between the two files is different, as evident by the spectrograms too, and it's easy to pick them out.

foo_abx 2.0.1 report
foobar2000 v1.3.8
2015-07-21 23:19:34

File A: 10 Smooth Criminal-CD.flac
SHA1: 6f6dc6adb56a17fdb1672643bf0be14e2778e004
Gain adjustment: -8.76 dB
File B: 10 Smooth Criminal-HD-441.flac
SHA1: 9d8e137cbe0d27369d0442125aea611d471e5ace
Gain adjustment: -9.02 dB

Output:
DS : Geek Pulse X Infinity 1V5 Output, 16-bit
Crossfading: NO

23:19:34 : Test started.
23:22:31 : 01/01
23:22:42 : 02/02
23:22:53 : 03/03
23:23:06 : 04/04
23:23:12 : 05/05
23:23:18 : 06/06
23:23:26 : 07/07
23:23:31 : 08/08
23:23:39 : 09/09
23:24:35 : 10/10
23:24:35 : Test finished.

----------
Total: 10/10
Probability that you were guessing: 0.1%

-- signature --
b4208571ae58ca867651b80e7cdbf9f99c1ddc58
 
Jul 22, 2015 at 5:54 AM Post #711 of 3,525
miceblue said:
Decided to try this out. I just got a bunch of Michael Jackson's albums on CD and I wanted to see how they sounded differently from the HD masters.
 
Click to expand...

 
Something like DiffMaker should be able to align the files or even compensate for minor time expansion/contraction, if you wanted to have another go at the last comparison. For the 8/10 case, it would be interesting to see the null/difference file to see what the mastering differences are.
 
Jul 22, 2015 at 6:06 AM Post #712 of 3,525
I should be recording some vinyl soon,so I may upload a blind test for you people. No graphs, just ears.
 
Providing I can get the thing rigged up.
 
It will be The KLF - White Room recorded at 24/48, so I can save space and processing time. Some other vinyl like older Pink Floyd, Jazz maybe..who knows. There's a lot to choose from.
 
Jul 22, 2015 at 8:34 AM Post #713 of 3,525
interpolate said:
  I should be recording some vinyl soon,so I may upload a blind test for you people. No graphs, just ears.
 
Providing I can get the thing rigged up.
 
It will be The KLF - White Room recorded at 24/48, so I can save space and processing time. Some other vinyl like older Pink Floyd, Jazz maybe..who knows. There's a lot to choose from.
Click to expand...

 
This proof will be amusing to some of us.  When we did recordings in the 1980's we learned a thing or two about analog recordings.  We did digital recording as early as 1982 - and universally liked the results - but didn't release anything mainly due to obscene cost of multitrack digital rigs. So multitrack-tape-to-vinyl remained the default until about 1990. 

Here are some of the realities that seem to have been lost in the intervening years:
 
  • The dynamic range of vinyl was anemic.  A high noise floor combined with the fact that the stylus would not stay in the grove at high volumes meant that we compressed EVERYTHING more than anyone wanted. You absolutely needed something like dbX expansion to restore proper sound levels at playback. I don't see much expansion these days. But you audiophiles don't even use EQ which is funny to me since EVERY recording had extensive amounts of EQ to deal with limitations of vinyl and acoustic recording issues.  
     
  • Tape was the best medium for playback. Whenever possible we would scoop an original recording on reel-to-reel. We'd take it home and know we had a recording superior to vinyl.  None of us wanted vinyl except for maybe drink coasters.
     
  • We knew tape degraded over time. Not only did the mylar substrate get brittle and break, you were always getting magnetic remanence decay from recorded tapes sitting in a roll. This means even 'remastered' recordings are now suspect.  That's too bad.  If we used more digital equipment in the 1980's, we would have been better off. 
 
 
Enjoy your vinyl - we never did.
 
Jul 22, 2015 at 9:02 AM Post #714 of 3,525
I'm not an engineer, HiFi buff, nor do I have a multi-thousand dollar system but, IMO, a LOT of what sounds good on Vinyl, Tape, CD or otherwise is because of what goes on in the initial recording process.
 
Like the music or not Steely Dan's "Asia" was recorded in 1977 and it still "sounds" good on vinyl or CD. It's a great recording. 
 
I haven't read the article yet but I have read others that stated 24/96 was the optimum because the 24 bit gave one more head room for recording transients and the 96Khz sampling rate cut down on the duration of minor imperfections where as above that, at say 192Khz, the listen couldn't perceive any difference in SQ.
 
I'll give the article a read.
 
Jul 22, 2015 at 9:09 AM Post #715 of 3,525
I'm not trying to prove anything. Sod it, i'll just record the vinyl for my own personal pleasure then. It's because it's there really no other reason.
 
Jul 22, 2015 at 10:30 AM Post #716 of 3,525
I read the article. Like I said, I'm not an engineer but I was a audio tech and business owner. The article makes a lot of sense. So, as stated or implied in other articles as well. Both Dither and component quality in recording may have more of an impact on SQ than higher bit or sampling rates. 
 
Jul 23, 2015 at 1:12 PM Post #717 of 3,525
gorquin said:
  I read the article. Like I said, I'm not an engineer but I was a audio tech and business owner. The article makes a lot of sense. So, as stated or implied in other articles as well. Both Dither and component quality in recording may have more of an impact on SQ than higher bit or sampling rates. 
Click to expand...


FWIW: Dithering with respect to different time bases is not an issue if the source sample rate is an integral multiple or divisor of the playback rate. For example: Source @ 176.4 will downsample fine to 88.2 or  44.1 without dithering. The interpolation of missing bits is not destructive. Look for these multiples if you are interested in hi-res content. Higher sample rates don't always mean better sound if your DAC does not directly support a sample rate or an integral divisor.  For example: iPhone DACs are decent but iOS do not support much beyond 44.1/16 - yet.  96KHz and 192 KHz content will not generally downsample well as 88.2 KHz and 176.4 KHz.  

Also, variation in bit depth never causes dithering. It can result in minor dynamic compression depending the loudness range being represented in source and playback. 

Here is a good summary of the issues (Hint: Don't read if you believe in magic):

http://bitperfectsound.blogspot.ca/2015/05/how-does-sample-rate-conversion-work.html
 
Jul 23, 2015 at 3:06 PM Post #718 of 3,525
gr8desire said:
 
FWIW: Dithering with respect to different time bases is not an issue if the source sample rate is an integral multiple or divisor of the playback rate. For example: Source @ 176.4 will downsample fine to 88.2 or  44.1 without dithering. The interpolation of missing bits is not destructive. Look for these multiples if you are interested in hi-res content. Higher sample rates don't always mean better sound if your DAC does not directly support a sample rate or an integral divisor.  For example: iPhone DACs are decent but iOS do not support much beyond 44.1/16 - yet.  96KHz and 192 KHz content will not generally downsample well as 88.2 KHz and 176.4 KHz.  

Also, variation in bit depth never causes dithering. It can result in minor dynamic compression depending the loudness range being represented in source and playback. 

Here is a good summary of the issues (Hint: Don't read if you believe in magic):

http://bitperfectsound.blogspot.ca/2015/05/how-does-sample-rate-conversion-work.html
Click to expand...

 
If the data word length doesn't decrease, the existing dither suffices even though the sample rate changes.
 
The purpose of dither is to randomize the quantization error, and while changing the sample rate changes the bandwidth of the quantization error, it doesn't change its randomness.
 
You must log in or register to reply here.

Users who are viewing this thread

Total: 4 (members: 0, guests: 4)
Back
Top