256kbps vs FLAC

[Aman]

Quote:

Originally Posted by vcoheda /img/forum/go_quote.gif after almost a straight year of strict cd use (wav), i recently played some mp3s on my cd player - some stuff i got from the internet. the rates were 128, 192, 256, 320. ALL sounded BAD - compressed and harsh, with almost all inner detail lost. forget A/B comparisons. listen to WAV for a year (or even 3 to 6 months) and then go to mp3 at any rate and on a good setup. the difference is what audio is all about, imo.

Fine, but your statements are going completely against the scientific facts behind audio compression. Use what makes you happy, but don't go around telling others that low-compression MP3s such as 256k and above sound any different to the human ear than a FLAC or any other lossless file.

 

[ataraxia]

Quote:

Originally Posted by Aman /img/forum/go_quote.gif Fine, but your statements are going completely against the scientific facts behind audio compression. Use what makes you happy, but don't go around telling others that low-compression MP3s such as 256k and above sound any different to the human ear than a FLAC or any other lossless file.

This only tells half the story. I would not doubt that many of the mp3s obtained from the internet have been made through mediocre compression methods. There is a world of difference between mp3s compressed through LAME and those that have not. This is not a statement "gong completely against the scientific facts behind audio compression", as not all compression executable files are created equal. There are those that compress efficiently, and those that do it less so. To say that it is impossible to tell the difference between compressed lossy files and compressed lossless files is incorrect.

 

[majkel]

This is enough for me to get a decent mp3 player capable of playing FLAC files, and a pair of Sennheiser PX100 to hear and point the differences between the FLAC and the mp3 320 kb/s best (optimized/unoptimized) compression. People who don't hear - be happy with your --preset-standard files and stop spamming this thread, please.

Compression differences are not only bandwidth related. THD level of the mp3 compression hardly ever goes lower than -90dB which is high enough to be audible. Even -120dB level of THD distortion is discernible in some situations, especially when comes from the signal path.

 

[Redo]

Quote:

Originally Posted by ataraxia /img/forum/go_quote.gif This only tells half the story. I would not doubt that many of the mp3s obtained from the internet have been made through mediocre compression methods. There is a world of difference between mp3s compressed through LAME and those that have not. This is not a statement "gong completely against the scientific facts behind audio compression", as not all compression executable files are created equal. There are those that compress efficiently, and those that do it less so. To say that it is impossible to tell the difference between compressed lossy files and compressed lossless files is incorrect.

I do believe proper ripping, encoder, encoding methods, and playback techniques make the most differences between formats. There are about a gazillion settings that alter the sound definitely, and most "user-friendly" encoders aren't set for a straight lossy encode. The biggest problem I have is I've always used Lame 3.97 and OGG for my ABX tests. These two encoders are very good, I guess I do it to test my ears to the max.


Quote:

Originally Posted by majkel /img/forum/go_quote.gif This is enough for me to get a decent mp3 player capable of playing FLAC files, and a pair of Sennheiser PX100 to hear and point the differences between the FLAC and the mp3 320 kb/s best (optimized/unoptimized) compression. People who don't hear - be happy with your --preset-standard files and stop spamming this thread, please.

I'm sorry, this is a discussion forum in which the topic is asking for everyone to chime in on their thoughts. You have stated ABX'ing is too stressful to make a valid comparison. This means you only think you can hear the difference, but offer no proof you can differentiate 320 LAME from FLAC from 320 Fraunhofer.

On top of it all, using an mp3 player with PX100's to state how easy it is for you to tell the differences between lossless and HIGH bitrate formats. ABX'ing with good equipment should be a walk in the park for your ears.

 

[Febs]

Quote:

Originally Posted by vcoheda /img/forum/go_quote.gif after almost a straight year of strict cd use (wav), i recently played some mp3s on my cd player - some stuff i got from the internet. the rates were 128, 192, 256, 320. ALL sounded BAD - compressed and harsh, with almost all inner detail lost. forget A/B comparisons. listen to WAV for a year (or even 3 to 6 months) and then go to mp3 at any rate and on a good setup. the difference is what audio is all about, imo.

For all you know, those MP3s may have been transcoded 6 times before you downloaded them. Forget listening to WAV for a year, and listen to some well-encoded MP3s rather than some questionable crap that you downloaded and then you can at least make an informed comparison.

Quote:

Originally Posted by majkel /img/forum/go_quote.gif This is enough for me to get a decent mp3 player capable of playing FLAC files, and a pair of Sennheiser PX100 to hear and point the differences between the FLAC and the mp3 320 kb/s best (optimized/unoptimized) compression. People who don't hear - be happy with your --preset-standard files and stop spamming this thread, please.

Maybe you should go back and re-read the question that the original poster asked before you accuse people who are addressing that very question of "spamming."

 

[majkel]

I performed a couple of times ABX tests. 320kb/s LAME is damn easy to distinguish - it's just a tad brighter than the original. It's more difficult with the FhG or Blade codec. If you ask me to distinguish ogg Q8 or higher quality file from the CD content - I give in. They sound the same for me, at least from a mp3 player so I encourage people to use them on the mp3 players if they can, in order to save space. Q8 is just around 256kb/s - see topic.

And again, I don't need blind tests for myself anymore because I know what I hear, and again - I use my music for listening for pleasure, not for performing blind tests, so I care about the differences I hear while listening for pleasure.

I did not mean answering the question as spamming but posts suggesting there are no differences. If somebody doesn't hear - convert a mp3 file (any) and original lossless content both to wav format. Then compare it bitwise or their time domain graphs. Are they the same? I just hate a point of view "I am deaf which means sounds don't exist".

 

[EnOYiN]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif I performed a couple of times ABX tests. 320kb/s LAME is damn easy to distinguish - it's just a tad brighter than the original. It's more difficult with the FhG or Blade codec. If you ask me to distinguish ogg Q8 or higher quality file from the CD content - I give in. They sound the same for me, at least from a mp3 player so I encourage people to use them on the mp3 players if they can, in order to save space. Q8 is just around 256kb/s - see topic. And again, I don't need blind tests for myself anymore because I know what I hear, and again - I use my music for listening for pleasure, not for performing blind tests, so I care about the differences I hear while listening for pleasure.

I have heard and seen this a lot of times before and I regret that I would have to say that most people just lie about this. (not saying that you do too, mind you) Evidence for this shouldn't be too hard to find here on Head-Fi.

Quote:

Originally Posted by Aman /img/forum/go_quote.gif Almost no human ear can distinguish between a low-compression MP3 file (256k and up) and a lossless FLAC file, all else equal - no matter what equipment used.

This is simply not true and there have been people proving that they can hear differences. This has to do with the limits of an mp3 and the encoding process. So called artifacts are audible for someone who trained himself to be able to hear them. Some things are harder to encode than others.

 

[serpico]

From my experiences, I notice differences between genres of music. My wife and I collect many different genres of CD's and whenever I rip them in any lossy format, some are ok and others are very horrible. I'm sure there is a technical explanation for this because of the different styles of music. I've wasted enough time in my life re-encoding my cd's many times, that I stopped for now and just play it through a cd player.

Perhaps the young generation likes the 128 bit files because it is enough for the pop radio songs. Maybe when they get older, some will learn about audio and visit us here!

 

[SR-71Panorama]

[size=x-large]MP3 VS LOSSLESS: THE TEST:

Metallica Sad But True[/size]

If you cant hear a difference, then you dont need lossless.
I suspect everyone who tries this test will hear a difference.
I hear a difference immediately in the cymbals. It is so obvious it stares you in the face.

And Im talking about 320 bit, not some anorexic 128 bit.

 

[majkel]

SR-71P., I guess most people listen to undemanding music and maybe there are hardly any differences. For me it is enough to take some tunes from the Papa Roach - Infest album. Every lower than 320kb/s mp3 bitrate changes the guitar timbre. Another reason for hearing no differences is because this subtle differences is often lost in crappy mp3 players or soundcards. I sold at least two times players which didn't allow me to hear the full, rich midrange, which I can achieve from good DAPs, soundcards or CD players.
Another thing is that in all mp3's the soundstage is unstable, especially when Joint Stereo is used. And again - if your equipment does not provide correct or just observable soundstage (meaning discernible width and depth/distant), you cannot hear how bad are mp3's.

 

[inasafeplace]

Why not rip using V0 instead? CBR is a total waste of precious space if u wanna know...

 

[FrederikS|TPU]

In an ideal world FLAC should sound the same no matter what compression rate (0-8), but depending on how good the deconding works you can get different results.

If you examine a decoded FLAC file compression level 8 to say a level 1 there will be no difference bit wise. In theory the only things that differs is the artifacts that the decoding process produces at different compression levels.

I'm with Quaddy on this one it's all just a matter of finding out which compression level that gives the best result on your hardware.

EDIT:Half a sleep while writing.

 

[KyPeN]

Quote:

Originally Posted by FrederikS|TPU /img/forum/go_quote.gif From a compression algorithmic point of view the different FLAC compression levels have a significant impact on the bit per bit output. In an ideal world FLAC should sound the same no matter what compression rate (0-8), but since most of the hardware used for decoding uses a different logic depending on compression the decoded result can sound different. If you examine a decoded FLAC file compression level 8 to say a level 1 there will be no difference bit wise. In theory the only things that differs is the artifacts that the hardware decoder produces at different compression levels. I'm with Quaddy on this one it's all just a matter of finding out which compression level that gives the best result on your hardware.

From a "compression algorithmic point of view," you don't know what you are talking about. The FLAC is decoded in software to a PCM before output to a hardware DAC. Like I said before, diff the outputs of a FLAC file decoded to a WAV like all players do with the original WAV file and you'll see they are the same.

This is straight science. FLAC can not sound different from the original WAV file. It simply can not. Lossless means without loss. I don't understand where the ambiguity comes in.

 

[Febs]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif I did not mean answering the question as spamming but posts suggesting there are no differences. If somebody doesn't hear - convert a mp3 file (any) and original lossless content both to wav format. Then compare it bitwise or their time domain graphs.

This is a straw man argument. You have set up an argument that no-one here is making, and then refuted that argument, rather than addressing the issue that is actually the subject of the thread, which is whether the differences between MP3 and lossless are audible. No-one has argued that there are no differences between MP3 and the original lossless content. The fact that MP3s range in size from anywhere from 1/4 to 1/10 the size of the original file makes that fact obvious. The point of this discussion is not whether there are any differences, but whether those difference are audible.

Every controlled listening test that I have ever seen or participated in has demonstrated that well-encoded, high bitrate MP3s are indistinguishable to most people. I do not discount the possibility that you may be one of the people who can hear the difference, though I believe that it is at least equally likely that there is some other explanation for whatever difference you may be hearing.

As EnOYiN pointed out, considering how many people here assert that they can hear a difference between MP3 and the uncompressed original, and considering how easy it is to prove that assertion, there is a surprising dearth of substantiated claims. Frankly, I don't understand this. After doing some ABX testing myself, I found that on my portable players, LAME -V5 is completely transparent to me, which allowed me to put substantially more music on my portable players without sacrificing any (subjective) quality at all. I archive all of my music in FLAC anyway, so if I ever find a track that has audible artifacts, I can simply reencode it at a higher setting. But for portable purposes, I don't see any reason to waste limited storage space with encodings that don't offer any audible improvement over MP3.

 

[vcoheda]

^^ sorry febs. ipod + bithead isn't exactly the most ideal setup for determining if there are audible differences between wav and mp3.

get back to me when you get something better.