[Smoothbass12]

I recently imported my CD of John Mayer's Born and Raised album, and decided I would do a little experiment on myself. I uploaded the song "Speak for Me" (as well as a few others) to three different bit rates (128kbps, 256 kbps, and Apple Lossless) and compared the qualities with my rig (MacBook Pro to Fiio e17 DAC/amp to Beyerdynamic dt770). I heard absolutely NO difference (major let down:/). I was pretty surprised honestly, I do well picking out nuances most of the time. Would there be any reasons I don't hear a difference or are my ears just 'not good enough?' haha. I checked the info on the songs (and saw the dramatic file size difference) and noticed there was a common sample rate of 41K and common sample size of 16bit between the songs. I know that these are not as high as would be considered 'audiophile quality', but I have no idea how to correct that. I had the same info displayed on the fiio e17. How could I correct this? Would it relate to the fact that all were from the CD and that there would be a noticeable difference if downloaded from the iTunes store? And say I did correct it, would the difference even be worth it? (Yes, I just opened that can of worms.

)

 

[cel4145]

The difference can be subtle, more so than people make it out to be. There's a certain lossless "snobbiness" hype from people that have have never done any ABX testing.

Try these and see if you can tell the difference:

http://mp3ornot.com/index.php

A lot of people find it difficult or are unable to score more than around 50%. So one would think if 128k and 320k are hard to tell apart, so would be 320k and lossless, right? I can do it most of the time, but it's subtle.

That being said, I still buy CDs. But Amazon's new AutoRip would pretty handy for those that want mp3s ready-made for them out of their CDs.

 

[lorriman]

three different bit rates (128kbps, 256 kbps, and Apple Lossless) and compared the qualities with my rig (MacBook Pro to Fiio e17 DAC/amp to Beyerdynamic dt770). I heard absolutely NO difference (major let down:/). I was pretty surprised honestly, I do well picking out nuances most of the time. Would there be any reasons I don't hear a difference or are my ears just 'not good enough?' haha.

I wouldn't be surprised if you also couldn't tell the difference between the Mac's DAC and the FIIO. An amp makes a real difference when the heapdhones need one, but a lot of modern, standard, onboard dacs seem to be excellent; and pricey 3rd party offerings actually poorer than those they are meant to take-over from.

The only bad player I've heard of late was on the HTC Desire; very noisy, bad audio (to my subjective ears). I've heard that the audio on Samsung Galaxies is poor also, despite being touted as multimedia devices.

 

[lorriman]

three different bit rates (128kbps, 256 kbps, and Apple Lossless) and compared the qualities with my rig (MacBook Pro to Fiio e17 DAC/amp to Beyerdynamic dt770). I heard absolutely NO difference (major let down:/). I was pretty surprised honestly, I do well picking out nuances most of the time. Would there be any reasons I don't hear a difference or are my ears just 'not good enough?' haha.

I wouldn't be surprised if you also couldn't tell the difference between the Mac's DAC and the FIIO. An amp makes a real difference when the heapdhones need one, but a lot of modern, standard, onboard dacs seem to be excellent; and pricey 3rd party offerings actually poorer than those they are meant to take-over from.

The only bad player I've heard of late was on the HTC Desire; very noisy, bad audio (to my subjective ears). I've heard that the audio on Samsung Galaxies is poor also, despite being touted as multimedia devices.

 

[catspaw]

Personally i can tell the diffrence between Fiio E17 and laptops sound card (MSI 780DX). (Dac wise ofc)
 
The difference is small, rather okish for gaming, but a modest improvment in music.
The biggest difference i could notice was on diablo 2 soundtrack (on HE-400).
 
When testing quality of music, i am not 100% sure about this, but i think not all music generes benefit as much from kbps, but rather on the rip quality.
 
I can say that dance/electronic music at 128kbps for me sounds as good as a CD.
 
Also i "Assume" that you need a very high setup to notice a decent diffrence (like Flagship good with 2k+$ amps/dacs).
 
I think once you get to mid-fi (He-400/HD600/Akg.. 701 annis?), and bare in mind those were HI-fi some years ago, and our ears have not evolved biologicaly speaking in that time, the increments are rather small in sound quality, while the price keeps on increasing rapidly.
 
I belive it happens in a similar way in audio encoding, but only if you get better HPs.
 
After all, once i got the HE-400 i had to throw away about 80% of my mp3s as they were poor encodings.

 

[chewy4]

When the files are well encoded, most people have trouble with it. Newer LAME encoded mp3's are only different in minor ways(nothing you'd pick up with casual listening), and in some songs they're practically transparent.
 
Those samples on mp3ornot aren't level matched by the way. If you don't manually level match your files, the 128kbps file is usually going to be slightly quieter.

 

[lorriman]

Personally i can tell the diffrence between Fiio E17 and laptops sound card (MSI 780DX). (Dac wise ofc)

The difference is small, rather okish for gaming, but a modest improvment in music.
The biggest difference i could notice was on diablo 2 soundtrack (on HE-400).

Wouldn't that be due to the amping more than the dac?

 

[TheJazzyH]

Even on my HP Pavilion dv4's onboard audio (after disabling all of the "enhancements"), I was able to score 3/3 right on mp3ornot with my M50s (not much time spent, I know). The difference is barely noticeable, however. I probably wouldn't be able to continue this perfect streak for long.
 
I can't really distinguish 256kbps AAC from lossless, though, so as of now, I don't really mind buying from the iTunes store when no higher quality option exists at a reasonable price (usually J-Pop and J-Rock).
 
I hope this changes once I get a good DAC. I find my HP laptop to be fairly noisy.

 

[obobskivich]

The difference can be subtle, more so than people make it out to be. There's a certain lossless "snobbiness" hype from people that have have never done any ABX testing.

Try these and see if you can tell the difference:

http://mp3ornot.com/index.php

A lot of people find it difficult or are unable to score more than around 50%. So one would think if 128k and 320k are hard to tell apart, so would be 320k and lossless, right? I can do it most of the time, but it's subtle.

That being said, I still buy CDs. But Amazon's new AutoRip would pretty handy for those that want mp3s ready-made for them out of their CDs.

+1.

I'd also add that the material can dictate how aggressive your encoder can get as well. That having been said, anymore I tend to rip VBR WMA-L, mostly because disk space is cheap and I don't really care about conserving it anymore (years ago it was a concern, but honestly storage is cheap today, so why fuss?). I too buy CDs - I like the convenience more than anything else.

 

[cel4145]

When the files are well encoded, most people have trouble with it. Newer LAME encoded mp3's are only different in minor ways(nothing you'd pick up with casual listening), and in some songs they're practically transparent.

Those samples on mp3ornot aren't level matched by the way. If you don't manually level match your files, the 128kbps file is usually going to be slightly quieter.

Sure. It should be slightly quieter because it is missing some frequencies, even though subtly so. But if you level match it, then you've just made the sum of the frequencies that are still there louder than they are in the higher bitrate file.

So I would go for not level matching since we are trying to hear if anything is missing.

 

[Jaywalk3r]

When the files are well encoded, most people have trouble with it. Newer LAME encoded mp3's are only different in minor ways(nothing you'd pick up with casual listening), and in some songs they're practically transparent.

Those samples on mp3ornot aren't level matched by the way. If you don't manually level match your files, the 128kbps file is usually going to be slightly quieter.

Sure. It should be slightly quieter because it is missing some frequencies, even though subtly so. But if you level match it, then you've just made the sum of the frequencies that are still there louder than they are in the higher bitrate file.

So I would go for not level matching since we are trying to hear if anything is missing.

If level matching is not performed, then any difference detected by the listener will likely be due to volume difference, not actual difference in quality. Humans can detect very small differences in volume, and louder sources tend to sound "better," regardless of bit rate. Only when two different bit rates can be distinguished from each other at identical volume levels is that distinction meaningful.

For the comparison to produce any meaningful result, level matching is mandatory.

 

[cel4145]

If level matching is not performed, then any difference detected by the listener will likely be due to volume difference, not actual difference in quality. Humans can detect very small differences in volume, and louder sources tend to sound "better," regardless of bit rate. Only when two different bit rates can be distinguished from each other at identical volume levels is that distinction meaningful.

For the comparison to produce any meaningful result, level matching is mandatory.

You can't win either way. Humans can also detect differences in certain frequencies being emphasized.

If you take a lossless file and convert it an mp3, you are leaving out some data, some aspect of the music, right? So the audio data that is lost is no longer generating SPL. But if you played back what is left in the mp3 so that the audio data that is still there is playing at the same volume as that same data is in the lossless file (not the extra data), the mp3 should, overall, output less SPL--there is less of the music left in the new version. Now if you level match the mp3 to the lossless in overall SPL, then you have just made the audio data that is still there louder than it is in the original version. That could make the mp3 sound better because you have now given some parts of the music more emphasis.

Now if it turns out the mp3 encoder decreases the output level for the parts of the music that are still there, then that's another problem. Perhaps they add distortion. I've sort of run to the limits of my understanding of MP3 encoders

 

[Jaywalk3r]

If you take a lossless file and convert it an mp3, you are leaving out some data, some aspect of the music, right? So the audio data that is lost is no longer generating SPL.

No, that's not correct.

That the quantity of data is reduced does not imply that sound is missing, i.e., no longer producing SPL. At least for audible frequencies, much of the "missing" data is estimated and replaced during playback. Compared to the original data, the estimations inevitably have errors, we should be careful not to assume that the error always results in the lossy passage being quieter than the original. It's just different. If we wish to determine whether or not that difference is audible, we must level match for our ABX tests.

 

[cel4145]

No, that's not correct.

That the quantity of data is reduced does not imply that sound is missing, i.e., no longer producing SPL. At least for audible frequencies, much of the "missing" data is estimated and replaced during playback. Compared to the original data, the estimations inevitably have errors, we should be careful not to assume that the error always results in the lossy passage being quieter than the original. It's just different. If we wish to determine whether or not that difference is audible, we must level match for our ABX tests.

Thanks. I didn't realize it can introduce error in the encoding/decoding process.

 

[Jaywalk3r]

No, that's not correct.

That the quantity of data is reduced does not imply that sound is missing, i.e., no longer producing SPL. At least for audible frequencies, much of the "missing" data is estimated and replaced during playback. Compared to the original data, the estimations inevitably have errors, we should be careful not to assume that the error always results in the lossy passage being quieter than the original. It's just different. If we wish to determine whether or not that difference is audible, we must level match for our ABX tests.

Thanks. I didn't realize it can introduce error in the encoding/decoding process.

I think you might be interpreting error in a manner different than intended. An estimate is virtually always wrong, but can often be very close to correct. Suppose you measure the length of a carrot with a ruler that is only graduated down to 1/8 inches. If one end of the carrot is lined up with the zero mark, the other end is extremely unlikely to lie exactly on a n/8 mark. The exact length remains unknown, but can be estimated with reasonable accuracy. The difference between the actual length and the estimated length is an error, but it might not be (probably isn't, at least in this example) consequential.