2. misc-audio
  3. 1194498225

Analog vs. digital segment on PBS

The show "Wired Science" on PBS this week has a good segment on analog vs. digital with a relatively quick blind panel test on analog vs. digital. I think they replay the show during the week if you can catch it. Nice to see some of the hobby getting some primetime attention, if PBS can be considered primetime of course! They have a couple recording engineers speaking about the merits of each and a blind listening test between a recording group (whose music they use for the test) and some unbiased recording engineers.
Also some info on frozen brains... either way it's a great show for general technology every week.
jimmy2615
tgrisham
1,291 posts
 
Albert, do you think the typical CD "quality" we experience is due to the disc itself. I believe I hear a difference when I listen to CD-Rs from standard Redbook CDs using the EAC software, but it is not double blind. Any thoughts?
tomcy6's avatar
tomcy6
2,621 posts
 
They couldn't do Lp vs cd, everyone would have known which was which by the pops and crackles:)

I have read a few reviews of cd players lately where the reviewer states that the sound of the cd was close to if not equal to the sound of the vinyl version. I expect to hear that more frequently as time passes. I think that cd players are just starting to reach their potential and that there will be sonic improvements in cd players for a long time to come.
albertporter's avatar
albertporter
4,949 posts
 
Albert, do you think the typical CD "quality" we experience is due to the disc itself. I believe I hear a difference when I listen to CD-Rs from standard Redbook CDs using the EAC software, but it is not double blind. Any thoughts?
Tgrisham

In the case of the listening tests I did in New York, the digital master had much higher sampling rates than the consumer CD.

That's probably the biggest difference, or perhaps there are errors that the shiny disc creates while spinning, compared to tape against the playback head?

Another friend of mine was involved in the JVC XRCD project, he too is a recording engineer and audiophile. A few years ago at Stereophile, he and I were having a long conversation and he expressed his frustration that his digital masters sitting on the hard drive were incredibly good, but the VERY FIRST transfer to anywhere else, it moved toward digital nasty.

Note here, perhaps digital on hard drive is converted to analog at the studio for LP's production? If so, this might explain why DSD masters pressed to LP (often) sound better than Redbook.

Maybe this is a clock thing? Maybe the hard drive is more perfect and other formats are required to evolve the sound from that point on?

As for your tests with re-recording CD's, you are possibly getting error correction and that's what you hear. I've heard others say this is possible, many burn a "better" CD after software compares 100 times (or whatever).

You say you did not do double blind, no need, I'll take your word for the results.

I'm an advocate of relaxed listening over double blind. Long term allows you to relax, absorb and learn all at the same time. Double blind tests are like cramming for an exam the night before, instead of studying all semester :^).

I hope Tousana will contribute to this thread again.
tousana
16 posts
 
Albertporter, first let me say to your friend that is working with the JVC XRCD
project, that those CD's too me are some of the finest CD's made that will play in a standard CD player. I've said before that if all CD's were produced with such care the format would not get bashed so much. The list is long of things that can happen while recording and or mixing to affect the sound of the recording. I don't remember if they touched on the fact in the PBS segment that analog is very forgiving as far as the recorded level is concerned and that some of the warmth that is achieved from analog recording can be caused by the natural compression of the signal as it gets recorded. This happens when the level of the recorded material gets to hot for the tape to handle, but in digital when levels get to hot it is all or nothing. Digital is not a forgiving medium when mistakes are made. If you make it out of the studio with a good mix there are other things that can go wrong between finishing the mix and the actual release of a CD. One of the main things is the mastering of a project where for the last 10 yrs. or so the loudness wars have pretty much ruined the sound of many recordings, in every type of music. This is the phenomenon of everybody wanting their recording to be as loud or louder than the next persons. Loudness = compression = loss of dynamic range (no real difference between the loud and soft passages in the music), which could all lead to possible distortion on the CD itself, or even the distortion of the electronics being used in the playback of the CD.
These are just a few of the the things that must be taken into account to produce a great sounding recording, regardless of it being put on CD or vinyl. There is very long list of things that can go wrong, but I'd need to write a book to explain them.
rouvin
26 posts
 
The results stated at the end of the testing, 20 brief samples/group, if I remember correctly, were: Engineer group correct 55% of the time; Musician group correct 53% of the time -- just slightly above chance, 50%. These percentages seem difficult to interpret or parse. They appeared to be summed for each of the two groups. The difficulty is determining what constituted a correct response in computing the percentages. If a sample were digital and one member of a pair responded digital and the other analog, how would this be counted? For instance: Is this a group miss because there was one correct and one incorrect, or is it two responses, one correct, the other incorrect? Was there a differential between members of each pair in percentages identified correctly (one engineer right 60% and the other 50% = 55% for the engineer group)? If one group member were wrong more than 50% of the time would this be an inverse correlation exceeding chance? (Wrong 65% of the time could be interpreted as being able to differentiate digital and analog even though misidentified; and would negatively affect the reported outcome "correct" statistics.)Was there a difference in identifying digital or analog material correctly?

My experience in reading about these issues is that technical matters about how the testing is done, such as some of the possible problems about the setup and administration of the testing on PBS identified in some of the posts, and how statistics are generated, organized and analyzed are often ignored in favor of a debate among the believers and doubters in each camp. If there are problems in any step of the procedures, the results cannot be meaningfully understood.

Given all of the issues identified in the procedure and the number of Rumsfeldian unknown unknowns can anything be concluded from this segment?
More to discover