Head-Fi.org › Forums › Equipment Forums › Sound Science › MP3 vs FLAC, revisited
New Posts  All Forums:Forum Nav:

MP3 vs FLAC, revisited

post #1 of 16
Thread Starter 

I have googled this topic and there seems to be a wealth of information concerning this topic.  I would like to add my 2 cents and most of the threads that I found were old.  As I am big into headphone listening, I decided to post here.  I have the AKGK701 paired with the HeadRoom Maxed Out Home Headphone Amplifier at home and a Stax Pro Lambda system with its solid state amp at work. I've compared files using my headphone system, custom car audio system, and home (not headphone) audio system.  Here are my observations.

 

I have found that Exact Audio Copy (with AccurateRip) is essential in getting quality mp3s from my CD library.  Together with Lame V0 (VBR, approximately a 288 kbps bit rate overall), MP3 sounds great.  EAC (test and copy) + LAME does a much better job than the much quicker iTunes ripping. 

 

There are four important variables in being able to distinguish between quality mp3s and FLAC: (i) the quality of the equipment you own, (ii) the quality of your hearing, (iii) your trained ability as a critical listener, and (iv) environmental noise or distractions.  A final variable is even more important: some people can distinguish between the two when they focus their attention and try, but the important question is how much your enjoyment of the music is improved by FLAC.  If you can hear a difference but consider it too small a difference to matter, then you don't need FLAC.  There is one final variable, consideration for other people.  Even if your ears aren't the best, you're not a critical listener, or are easily pleased by music, you *may* want to use FLAC if other people will be appreciative it. 

 

Most people (probably 90% or more) will be satisfied with good mp3s, because one or more of the criteria mentioned above are not met.  For instance, it would be impossible for almost everyone to tell a difference between mp3 and FLAC in the average car going down the highway (poor equipment in a noisy environment).  People on the move (iPod) will also likely not notice a big difference due to midfi/lowfi equipment and environment.

 

Some mp3s are much more easily distinguished as different than live (or CD).  Namely, many people can distinguish 128 kbps mp3s from FLAC.  I really don't think there is any reason for 128 kbps mp3s considering the storage available today.  Free Pandora delivers a 128 Kbps stream of music, and it is tolerable to listen to even though it's lesser quality is obvious.  Pandora One upgrades the performance to 192 Kbps.  MOG is the winner (at least to me), providing its music at 320 kbps.

 

However, each person should decide for themselves what format is right for them.  They should do this not by listening to what other people claim is right nor by simple guessing based on the variables listed above.  There are some people who claim they can hear the difference when a person places a dime underneath one of their speakers.  Of course, I don't believe that.  Conversely, I know that there are some people who have excellent hearing and are very critical listeners that can hear things correctly when I can't.  I suggest making a CD, USB, or other digital recording (according to the source you will use) of FLAC and different mp3s that you know well.  Volume match the songs if possible. Allow a friend to play different tracks and see if you can tell a difference using the audio system and while in the environment that you will usually be listening.

 

I tested myself.  In the quiet with my audiophile-quality home stereo, I can tell the difference between 288 mp3 and FLAC with A vs B comparison.  Although my equipment is of audiophile grade, my ears are those of a 44 year old male who has heard too many live concerts and high-powered sonicators.  It would be very challenging for me to correctly distinguish the two, without A to B comparison and within 20 seconds for each song, 7 times correctly out of 10 songs.  (My personal criteria for establishing a clear difference that matters). I really can't tell the difference between uncompressed wav files and FLAC files.  Knowing this, I decided to record my entire library in FLAC.  However, I will convert my library to high quality mp3's for my iPod.

 

For good digital music, a good DAC and jitter control are very important, but currently those things should not cost a mint.  You can download studio quality music direct to your computer onto media that does not scratch or corrode (www.hdtracks.com), but that can be backed up onto CDs or DVDs.

post #2 of 16

You should try using foobar2000 with the plugin 'ABX comparator'. It makes it really easy to to ABX comparisons, and it also magically creates a log file in which you can post here/look over.

 

I don't know if I'd call a 7/10 a positive result, as you'd be able to get that 11% of the time just by randomly guessing. Besides, if you say you can hear a clear difference, then you should be able to do heaps more trials while getting a good success rate. I think 13/16 is an unofficial guideline on a positive result, while I personally aim for a 0.1% probability value.

 

In my experience, the most important factor that determines whether or not you can distinguish the files is the actual song itself. Not even the genre itself, because I've had tests from the same genre where I'd fail one miserable, but pass the other. It has a lot to do with how the psychoacoustic compression is masked by what's actually going on in the track.

 

I've said it before and I'll say it again, the differences between high bitrate mp3 and lossless are tiny. I don't even consider mp3 to be a 'downgrade', but rather, 'different sounding'. I still cannot comprehend how one can get more musical enjoyment from lossless than mp3 from the actual sound differences, but perhaps it's night and day for some people. For me, it's just the peace of mind and the ability to transcode into whatever I might need at the time, as I'd never be able to listen to a song and instantly tell if it's mp3 or lossless without an ABX comparison.

 

Uncompressed wav files and FLAC files decode bit-for-bit exactly the same, there are no differences between the files unless you believe in FLAC eating viruses. tongue.gif

post #3 of 16
Quote:
Originally Posted by deadlylover View Post

 

I've said it before and I'll say it again, the differences between high bitrate mp3 and lossless are tiny. I don't even consider mp3 to be a 'downgrade', but rather, 'different sounding'. I still cannot comprehend how one can get more musical enjoyment from lossless than mp3 from the actual sound differences, but perhaps it's night and day for some people. For me, it's just the peace of mind and the ability to transcode into whatever I might need at the time, as I'd never be able to listen to a song and instantly tell if it's mp3 or lossless without an ABX comparison.


Spot on. I feel it's the "cool thing" to not only say you're able to immediately tell the difference between lossy and lossless, but to spit upon the former. It's really a shame, since lossy audio compression is really one of the truly revolutionary software AND audio developments of the last 20 years.

post #4 of 16
Thread Starter 

I did do the foobar ABX comparison when using my computer.  As my home system is not connected to a computer, that was not possible there.  The same goes for the car stereo. The FLAC did sound better to me in my home system. Storage is not a problem. So, that is a very cheap way of increasing the sound quality of my home system overall - using FLAC instead of mp3.  Although my car audio system is expensive, the environment really doesn't make FLAC much (if any) of an improvement. Besides, I can carry my entire library on my iPod in mp3.

 

Although the theory is that FLAC = CD, I doubt that is possible in absolute terms when I am ripping the Cd with a $70 computer drive.  Simplistically, the copy is never as good as the original.  The theory of FLAC = WAV is more valid.  However, as a scientist, I know that theories are sometimes wrong.  In my case, the theory is not wrong. I could not tell the difference between FLAC and WAV.

 

post #5 of 16
Quote:
Originally Posted by rbeverjr View Post

Although the theory is that FLAC = CD, I doubt that is possible in absolute terms when I am ripping the Cd with a $70 computer drive.  Simplistically, the copy is never as good as the original.  The theory of FLAC = WAV is more valid.  However, as a scientist, I know that theories are sometimes wrong.  In my case, the theory is not wrong. I could not tell the difference between FLAC and WAV.


Really? I get bit-perfect rips (verified with AccurateRip) using a $20 DVD drive.

 

FLAC = WAV is not a theory. The data, once extracted, is exactly the same bit for bit, unless there is an error somewhere which is extremely rare. The only difference between the two during playback is the number of hard drive seeks and CPU cycles. Do you lose data when you place files in a ZIP?

post #6 of 16

People buy computer optical drives (non-BluRay) for more than like $15-25?  I guess the external and slimline ones cost more, but $70 still is too much.

 

Anyway, the audio data on the CD is digitally encoded, interleaved, and encoded with a Reed-Solomon code for forward error correction (data redundancy).  There's one bit of redundancy for every three bits of audio data.  With some scratches, with an imperfect optical drive reader, etc., there can be some errors in reading the data from the disc, sure.  But for an RS code, this means that if 1/8th or fewer of all the bits read from the disc (in each code block) are incorrect, the errors can be completely recovered, so you get an exact replica of the data on the CD.  Under some reasonable conditions, you can get a perfect rip with 100% data intact.

 

Copies don't necessarily lose fidelity in any sense...or do you think that when you send an e-mail, the text they receive will be different than what you sent?  Or maybe for the image attachment?  Or whenever you copy a file to a flash drive?  edit: I see I was late and Head Injury already went through a similar line of reasoning heh


Edited by mikeaj - 12/10/11 at 7:11pm
post #7 of 16
Thread Starter 

I appreciate the information. Obviously, I am not a computer guru, but I am glad that my listening tests agree with your assessment, FLAC = WAV (and apparently =CD too). Looking over this forum, this seems to be a good place to pick up the science behind the sound!

post #8 of 16
I can't tell the difference between 320kbps and FLAC, but that doesn't mean there is no difference, and given that I store my music on a 1TB HDD I have no problem with using the FLAC format.
post #9 of 16
Thread Starter 

I'd like to remind people not to make the mistake I did. I'm not sure that Apple Lossless codec is the same quality as FLAC.  I have seen posts disputing that, but have never compared for myself.  However, I made a mistake when comparing the sound quality of the Apple Lossless codec (through my trusty iPod) with a CD.  The unfortunate thing about the iPod is that it's USB connection can feed digital or analog signal depending on how the manufacturer of what you are plugging it into has implemented the dock.  I had thought my iPod was connecting digitally to my stereo system, but it wasn't.  So, instead of using my fine audiophile 24 bit, 96 kHz DACs, the Apple Lossless was using the mediocre DACs of the iPod.  Needless to say, the CDs clearly sounded better in this comparison.

post #10 of 16
Quote:
Originally Posted by rbeverjr View Post

I'm not sure that Apple Lossless codec is the same quality as FLAC.



It is, as well as WAV, APE, SHN and all other lossless compression formats :)

post #11 of 16

I agree with you, but I have noticed the difference in flac and mp3 in certain genres. I personally think that all electronic music should be exported via flac or wav because it often times needs that extra detail. It's a noticable difference in detail and it seems like it often times has "cleaner" bass. The cleaner bass is often times how I notice the difference between flac and mp3 in a more difficult genre like pop or classical.

post #12 of 16

Hi guys

 

I moved from MP3 to ALAC a while back and i am very happy with the choice (my MP3s were a mix of bitrates so i have found a decent increase in the sound quality overall)

 

With FLAC one thing i noticed (after buying tracks from HDTracks) is that my FLAC files are smaller in size than ALAC (Apple lossless) which i use for my iPod/iPhone.

 

Is this just my imagination/am i decoding my cds incorrectly/have i had too much to drink at Christmas (I am Irish)

 

Thanks

 

Mark

post #13 of 16
Quote:
Originally Posted by markgreville View Post

Hi guys

 

I moved from MP3 to ALAC a while back and i am very happy with the choice (my MP3s were a mix of bitrates so i have found a decent increase in the sound quality overall)

 

With FLAC one thing i noticed (after buying tracks from HDTracks) is that my FLAC files are smaller in size than ALAC (Apple lossless) which i use for my iPod/iPhone.

 

Is this just my imagination/am i decoding my cds incorrectly/have i had too much to drink at Christmas (I am Irish)

 

Thanks

 

Mark


Lossless compression codecs use different algorithms to compress. Some take longer, some save more space. The FLAC website has some comparisons, though I don't know if the other codecs have been updated recently. FLAC has one of the faster encoding times, the fastest decoding time, and an average compression size. None of the differences are really that large.

 

So no, you're not imagining things. Each codec will end up at a different size, but they'll all contain the same data once extracted.

post #14 of 16

FLAC and ALAC compression is supposed to be almost equal, though (source), so if you have a significant difference in size, you might want to check your encoding settings.

post #15 of 16
Quote:
Originally Posted by andrewberge View Post

FLAC and ALAC compression is supposed to be almost equal, though (source), so if you have a significant difference in size, you might want to check your encoding settings.



Well the average (I assume arithmetic mean) of the resultant files are similar; but I would assume this refers to highly aggregate samples and there may or may not be large discrepancies between individual files. It may well be that FLAC and ALAC compress different types of songs to different degrees of effectiveness.

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › MP3 vs FLAC, revisited