The latest MP3 discussion thread (moved from the DVD-A discussion)

[ssully]

I might not bet on being able to hear the difference between 96kHz and 192kHz, but I would bet on hearing the difference between 160kbps mp3 and CD, depending on the quality of the original source of course. But most 160kbps mp3s I've heard I've recognized as crap easily before checking the bitrate.

That's a better bet, but still a long shot. As I said, the codec matters, as well as the source audio, as well as the listener's training. It's easy to make the first state-of-the-art, it's unusual to find source audio that doesn't encode well, and listeners with trainign or native acuity good enough to tell good 160 kbps from source are rare.

Now, I suppose it's possible technology has developed to better encode mp3s at 160kbps with better sound quality. I'd have to hear it to believe it though.

MP3 codecs underwent substantial evolution in the 2000's. At this point it's all fine-tuning. And of course, no one has ever been obligated to stick to a constant 160 kbps bitrate; variable bitrate (VBR) centered around 192 kbps is what I use. Paranoiacs use 320kbps constant bitrate. All three of these offer massive filesize reduction (and thus greatly enhanced portability) compared to lossless, with little or no subjective audio degradation for most listeners and sources.

 

[ssully]

Almost all material? I am only interested in what I do listen to and with that I can certainly tell the difference between 160 kbps MP3 or AAC compared to lossless uncompressed PCM. It is hardly a scientific test if it is limited to almost all material, that is sort a vague requirement. Does almost all material mean 51%? Does it mean anything other than music with a wide dynamic range? I am not an MP3 fan but what little I understand from people that do listen to MP3 a lot, they acknowledge a compromise in sound quality justified by the convenience and/or cost factor and they can tell the difference.

What people are these? People rarely report in detail on what basis they find any audio 'compromised', though sighted,non level-matched, bias-prone listening seems to be the most common mode, among those reporting audio difference generally.

I could never convince myself that I could tell the difference between a properly mastered stereo CD and properly mastered stereo SACD, I tried and failed. Since so many claim that they can, I just assumed it was either my hearing or my modest equipment. The same would probably be true if I compared identical masterings of high resolution stereo PCM to stereo CD. I have compared MP3 to CD and didn't find that challenging to distinguish. Just have a friend play a 160 kbps MP3 of a favorite recording, then listen to the CD without knowing which is which and see if you can get it right a statistically significant percentage of the time. I am willing to believe you can.

What music do you tend to listen to? What codecs and bitrates have you tested? Have you compared them a fair number of times, with software ABX, which is more 'double blind' than having a friend put on a CD and an mp3?

I have always assumed so many listen to MP3 because it sounds good enough, but almost all can tell the difference.

That's not supported by the extant listening test evidence (Hydrogenaudio.org's public listening tests, and a few conducted by magazines). Indeed, results indicate a lot of people are already struggling to tell lossy from source at 128 kbps, if a decent codec is used.

 

[ssully]

True, but I have read many times over that mp3 sucks and I finally found out what it sounds like.

Now consider where and how easily one could shoot holes in that logic.

You're right!
I haven't heard a lot of mp3 music tracks because I prefer not to.

Suppose your preference is based on hasty, poorly-supported conclusions about mp3 sound from a very limited sample?

 

[ssully]

What is this...pick on Perpendicular day??
Alright, I was asking for it!

Anyway, I know what I hear when I hear it.

No, you only *think* you do.

But don't worry, it's not just you, all of us are susceptible to that particular conceit.

 

[ssully]

In regards to sub-topic #3

Okay as the original poster of the "mp3 backlash", thank you for the comments that allow me to dis the article as baseless. People don't care and the world keeps spinning. It also gives me license to say that the vinyl comeback is so small as to be off the radar. So vinyl fanboys, it's over before it's begun. But I will defend the other statement; most CDs are just fine; it is the time spent sitting still (paying attention) and LISTENING that is making the difference in the appreciation of the music and NOT so much the medium. That's healthy. Whew.

As to MP3 quality, here's a specific story. When I started using iTunes in 2005, it all went in at 128 bit rate. Most of the time it was okay but others not. One of the biggest obvious failures was the cymbal work on In The Court Of The Crimson King. Whereas this light work on the drum kit was a marvel to hear in pristine glory, this bit rate mangled it so badly as to be unlistenable IMHO. It was not much better at 160 or 192; finally at 256 it straightened out. Hell, with storage so cheap, I just went with 320 on everything now and don't look back anymore.

Again, it's not *necessarily* the bitrate alone; the codec matters too. Particularly as 128 is on the 'edge' of artifact audibility, the codec really matters at that bitrate -- historically some codecs have done 128kps better than others (see Hydrogenaudio.org's public 128kbps listening tests for data on this). In all cases, it's too easy for our perceptions to be swayed by nonaudio factors, so 'proof positive' that you heard a negative difference needs some sort of bias-controlled comparison.

If there's any take-home point it's that you shouldn't just say 'the 128 kbps version of this tune sounded like crap', you should at least specify the codec too, and whether the comparison was in any way controlled for bias and level.

 

[Jeffrey Michael]

No, you only *think* you do.

But don't worry, it's not just you, all of us are susceptible to that particular conceit.

No, it's what ever YOU think.
I know what I know.

 

[ssully]

No, it's what ever YOU think.
I know what I know.

But what you 'know' can be wrong.