How well do you hear music?

[DesolationAngel]

Well, this is interesting... NPR give you six pieces of music reproduced as uncompressed WAVs, 320kbps and 128kbps MP3's.

Listened through a pair of Mackie monitors, running from an Apogee Mini-DAC... and I got 1 out of 6. BUT, the interesting thing to me, is that 4 out of my 5 incorrect answers were with the 320kbps selected. Only once did I select the 128kbps one and I knew, the instant I hit it that I was wrong and that I was choosing based on something in my head other than what I was hearing.

How do you do? NPR How Well Can You Hear Audio Quality?

[sdelsolray]

Originally Posted by DesolationAngel

Well, this is interesting... NPR give you six pieces of music reproduced as uncompressed WAVs, 320kbps and 128kbps MP3's.

Listened through a pair of Mackie monitors, running from an Apogee Mini-DAC... and I got 1 out of 6. BUT, the interesting thing to me, is that 4 out of my 5 incorrect answers were with the 320kbps selected. Only once did I select the 128kbps one and I knew, the instant I hit it that I was wrong and that I was choosing based on something in my head other than what I was hearing.

How do you do? NPR How Well Can You Hear Audio Quality?

Did you familiarize yourself with the six audio files before you took the test? If not, your test results aren't valid.

[DesolationAngel]

Originally Posted by sdelsolray

Did you familiarize yourself with the six audio files before you took the test? If not, your test results aren't valid.

Would you like to, perhaps, elaborate on my failure?

[sdelsolray]

Originally Posted by DesolationAngel

Would you like to, perhaps, elaborate on my failure?

It's not your failure. The test's results are not accurate because you were not familiar with the six audio files before you took the test. Become familiar with the six files by name and identity. Take the test afterwards. Post your results.

Unfortunately, NPR doesn't seem to allow for this on their webpage. NPR does not understand the basic requirements for comparison listening tests.

[Doug Young]

I think NPR is just trying to do a "real world" test. It's not scientific, it's not whether, having heard the original, can you pick it out. They're just letting you see if you hear anything that would make you pay for the higher quality. If it's so close that you have to be super-scientific about it to even be sure there's a difference, who cares? The comments are interesting. One person pointed out that the lower resolution on one track de-emphasized the crackles in one original recording, thereby making it sound "better". I've had that same experience with sometimes preferring a lower quality recording because it hides some flaws.

People also mentioned that these weren't necessarily great candidates. Interesting how dated the Neil Young track sounded, quality-wise, and the classical piece had some warble from being (I assume) an old tape recording, so you're left figuring out which version was the highest quality reproduction of a low-quality recording.

FWIW, I got 4 out of 6. Couldn't begin to tell which JayZ track was "better", they all sounded bad. My other miss was Katie Perry, possibly for the same reason :-)

[DesolationAngel]

I didn't think for a second that NPR were trying to conduct serious scientific research, just a quick 'hey, everyone is saying how crappy MP3's are but can you tell the difference between compressed and uncompressed' type of thing.

I think, in a couple of days, when I've forgotten the answers, I'm going to give it another go in my studio on the Dynaudio's rather than the Tapco's...

[Silly Moustache]

My pc is linked via the sound card (a good one) into an aged Teac AX55 amplifier thence to very old Sony 2 x SS7200 with 10", 5" and 1" speakers.

Good sound. Of the music samples I could bear to listen to for any length of time I got the classical piece and the Suzanne Vega piece right, most wrong with the 320 most often chosen.

The Neil Young track sounded blurry and any resolution, the shouty pop sings - well it really doesn't matter does it?

[tdq]

With headphones and a computer, I got 6 for 6. But I also have a lousy internet connection so I picked the ones that wouldn't load

[jbhiller]

Like some of the others, I got 1 correct but on all my wrongs I chose the 320kbps and I really think they sounded better than the 128kbps. It was usually a toss up between the .wav and the 320kbps or a slight edge to the 320kbps. Hmmmmm.


I used a pair of decent Sony noise cancellation headphones but not my somewhat audiophile grade stereo (Creek integrated amp and Totem Arrow speakers).

For my tastes, and I fashion myself some level of an audiophile, some compression is fine, but it should be a bit higher bit rate than 128kbps. Apple Lossless (<128), 3200kbps, and other codecs work just fine for me.

[jetcode]

One thing I have learned over the years is that the ear needs to be educated to what is attainable. It is impossible to do without time in front of a reference system or minimally a high end system. I went with a stereo 30w Prima Luna tube amp, Wharfedale 7 Diamond Anniversary bookshelf speakers, and Emu 0404 (ESS9018) DAC. A budget system.

Something strange happened on the way to nirvana ... The sound is more dimensional, bigger sound stage, more separation which exposes greater detail. I can hear things now I've never heard before. Like cymbal splashes and bad mixes. When a good mix plays it's magic. There is a wonderful presence in the high end, a sparkle that is often missing from down sampled tracks.

In practice I can live with less because I have for soooooo long. MP3 still sounds decent. If it didn't it would have never been adopted as a standard.

[Andy Howell]

Message deleted - posted in wrong topic!

[Betelgeuse]

This is really interesting and something that I have stumbled across recently when reading about soundwaves and how sound works in general. Hearing sound is not purely mechanical and involves the inner ear which processes signals which arrive as sound waves, the brain can then use this processed signal to deliver to us what we perceive as sound.

Psychoacoustics is the scientific study of sound perception and is used in data compression of audio. Through psychoacoustics we know that certain differences between waveforms are from a human perspective imperceptible and data compression takes advantage of this fact and in theory a human should be unable to tell the difference between an MP3 at 320kbs and a WAV. MP3's at anything less than 320kbs are compressing elements that are perceptible by humans so it becomes a compromise of sorts.

I am not a scientist and because of that my explanation may not be very good but here is a link to wikipedias page regarding psychoacoustics
http://en.wikipedia.org/wiki/Psychoacoustics the whole subject covers many fascinating areas including one which we should all be familiar with and that is binaural beats.
When tuning a guitar, as the two notes get closer together you can hear a wah-wah noise, these are binaural beats and they only occur because of the way the brain perceives and processes sound, they don't really exist except in our heads... it's all fascinating stuff.

[Pete on guitar]

That's a really interesting test. Makes you realise that anybody who says MP3s are rubbish is talking rubbish - or at least, never had to use cassettes.

Only thing I would say though, is that A-B testing is never quite as conclusive as it sounds like it should be. I have two pairs of headphones, both reasonably high quality, and if I try to A-B test them they both sound similarly good. But one pair (Grados) are unquestionably the ones I slip on to really enjoy getting lost right in the music.

Pete