sound quality, mp3 vs cd's

[fmw]

Actually, if YOU put forward a factual claim, it's YOUR homework. Your tests are one thing (and thanks for the details). The claim that nobody has ever detected a difference is not supported by your individual tests, however.

.

Your opinion, not mine. Because someone says it on the internet doesn't mean much to me unless I know who is saying it. Help yourself to all the citations you like. I'm not going to look them up. Just don't believe me. Call me confused. That's less work for me and I don't really care. As far as I know, nobody on the planet can distinguish a 320 mp3 from a wav file. Nobody has provided me any credible information to the contrary. I don't consider it meaningful to listen to non music to judge something that is applied to music. It's is nitpicking, immaterial and ridiculous. I'm out of this debate.

 

[OttoMatic]

Here is a link to some Hydrogenaudio listening tests. I haven't made any sense of it at this point, but there does appear to be quite a bit of information there.

 

[lukejr]

I prefer CD to MP3, even the ones ripped at 320 kbps. This conclusion is from informal listening sessions. There are subtle differences that when played on a good stereo system, it's apparent the CD triumphs over MP3s. These differences are detail, instrument spacing, imaging, etc.. Again I stress that the differences are subtle, but still noticeable.

I'll try to do a test one of these days when I have some time. Basically the methodology I'll follow will be a blind test (the listener), and a helper (who switches the CDs). Listener blindfolded, sits in one spot and listens to a portion of a song (30 secs), then listens to the same song on either the master CD (X) or ripped MP3 version (Y). There will be no trickery. If X was played first, then Y should be played second. And vice versa. The reason why no trickery should be involved is it creates "stress" in the experiment which might influence the listener to possibly make a guess. The test's objective is to disprove the null hypothesis (there is no difference in sound). The scoring of above 60% on 10 playbacks should be at least noteworthy. Above 70% is significant. Anything above 80% is very significant and not likely guessing.

Let me know if this methodology is valid for the skeptics. I'd love to know whether a self professed discernible listener can tell the difference.

 

[adwilk]

Wont the test need to be level matched? Based on louder being perceived as better?

 

[lukejr]

The listener has the option and should exercise it to listen to the music at different volumes to see if he or she can discern differences. There will be no volume bias if this is carried out.

 

[WmAx]

The best method is to use a computer as the switching system and source. You can connect a DAC to the computer if you have insufficient quality sound card, of course, assuming your computer has digital outputs to feed a normal DAC.

You need to use the original source wave. Then convert the wave file to mp3 using a high quality encoder like the latest Lame encoder in a valid preset - remember that Lame is designed to have presets that have the programmed encoding methods that are best for human perception.

After you make the mp3, convert the mp3 to wave. This is important, just in case the playback software handles the mp3 different from the wave, such as a delay in start, or slight level chnage or any other variable that can act as an identification cue. Now use an audio editor and synch the original wave and ripped mp3 wave to the same time positions and then adjust level to be equal if there is any differce - there should be none.

Now use foobar with the abx plug in or use PCABX a standalone ABX program. The ABX software will let your computer handle the randomization and you can set the tracks to loop or play any section of the wave you so desire in real time.

-Chris

 

[lukejr]

I use winLAME for all my MP3 encoding so I am familiar with it. Since I have a laptop, it might not be possible to use the computer as the source since obviously the sound card and headphone output sucks. I am unwilling to spend money to buy a DAC. I think if I do the test I would use my DVD player as source connected to a Little Dot MK III amp. My listening will probably be through headphones (Grado SR60s or Sennheiser HD595), however, I might try using speakers as well.

 

[fmw]

Wont the test need to be level matched? Based on louder being perceived as better?

Not in this case because you are using the same playback equipment. You are only changing the software. As long as the two files are made the same way they will be level matched. Same thing with wire tests. Wire doesn't affect volume. Level matching is required for most every other comparison, however.