Head-Fi.org › Forums › Equipment Forums › Portable Source Gear › Lossy Audio Codec's Comparison [HUGE amount of pics] [iTunes UPDATE on p.7]
New Posts  All Forums:Forum Nav:

Lossy Audio Codec's Comparison [HUGE amount of pics] [iTunes UPDATE on p.7] - Page 9

post #121 of 222
3/25/07 at 9:57am
Thread Starter 
I've uploaded them to my website, now making a big post with the pictures ... gonna take a while. (copy-paste 33 links in [IMG] thingy, onoz)
Reply
post #122 of 222
3/25/07 at 10:01am
ok give us the link when youre done. or at least a little taste here. maybe just the 192cbr and vbr and the 224 aac files
Reply
post #123 of 222
3/25/07 at 10:37am
Sir Nobax did you try to compress to WMA9.2 in WMP11?

I am still curious to see if there is any difference.
Reply
post #124 of 222
3/25/07 at 11:01am
Thread Starter 
BIG UPDATE
iTunes 7.1.1.5


Here is my latest update, all major encoder settings used in the newest iTunes. I've encoded the Main Wave several times with iTunes's AAC and MP3 encoder, with both CBR and VBR.

At first Glance the VBR versions of iTunes' AAC looks the same, this is not true, the files are marginally different. iTunes' encoder searches for 'low-bitrate spots' when in VBR-mode, this is makes the test sample very uneffective since the tone sweeps are all over the file, without silent gaps. iTunes give every second of the file the maximum bitrate, this is the reason why the filesizes, wich i posted a hour or so back, are very close.


Original Wave

You can get the file HERE.


iTunes AAC


AAC at 16kbps, this is the lowest bitrate iTunes can encode, i did it just to make the comparison more complete as this bandwith is way to low to sound decently.


AAC at 32kbps, this is quite usefull for speech only, but i would seriously consider HE-AAC or AAC+ instead


AAC at 64kbps


AAC at 96kbps, from this point music is actually listenable (IMHO) on speakers


AAC 96kbps VBR, as you can see, the difference between CBR and VBR is very minor


AAC at 128kbps, a lot of improvement since my first post


AAC at 128kbps VBR


AAC at 160kbps, not so much difference with the 128kbps sample


AAC at 160kbps VBR


AAC at 192kbps, if they would bump up the low-pass filter it would sound alot better, it is acousticly indisguisable from the 160kbps version when i did a quick listening test.


AAC at 192kbps VBR, same as above, but it has a 'bitrate storage' for very dynamic music (like classical music), probably none shall hear the difference between this and the 224 CBR file.


AAC at 224kbps, not woth the extra space, use the 192kbps VBR instead


AAC at 256kbps, here the AAC file's lowpass is at his maximum, close listerers with very refined equipment can hear minor improvement over the sub 256kbps files, others should not bother and go for VBR 192kbps or VBR 256kbps instead.


AAC at 256kbps, the best AAC file encode-able with iTunes


AAC at 320kbps, useless because of the 20k lowpass filter, nothing better then a VBR 256kbps file.

...picture limit.. MP3 coming up in a few minutes
Reply
post #125 of 222
3/25/07 at 11:22am
Thread Starter 
iTunes MP3



MP3 at 16kbps, looks quite pathatic


MP3 at 16kbps VBR


MP3 at 32kbps


MP3 at 32kbps VBR


MP3 at 64kbps


MP3 at 64kbps VBR


MP3 at 96kbps, this is the total opposite of iTunes AAC, where the 320kbps AAC files have a lowpass filter, on the 96kbps MP3 file its nowhere to be seen, strange guys there at Apple.


MP3 at 96kbps, showing that iTunes' MP3 VBR files use the same quality check as iTunes' AAC files


MP3 at 128kbps


MP3 at 128kbps VBR


MP3 at 160kbps


MP3 at 160kbps VBR


MP3 at 192kbps


MP3 at 192kbps VBR


MP3 at 256kbps


MP3 at 256kbps VBR


MP3 at 320kbps


MP3 at 320kbps VBR, ahhh finally i see some difference between a CBR and a VBR file, still inferiour to any other codec at that bandwith though.

..so the conclusion for iTunes' MP3 encoder,
...it's BAD.

..just kidding, for a MP3 encoder their low-medium encodings are okay, but BE WARNED above 160kbps its using totally different encoding algolrithmes, making especially the 192kbps version look awfull.

If you'd look at AAC's performance at lower bitrate's I dont see any reason for using iTunes' MP3 encoder. Anyways, the AAC encoder is quite nice at medium bitrates (160kbps-192kbps VBR), wich would be good enough for on-the-go for a lot of people.

Still, anyone with any requests, please ask, ill have a look at it. (now searching for MP3-Xing-encoder) Also, anybody who has a non-Sonicstage ATRAC encoder plz PM me or post a link in this thread.
Reply
post #126 of 222
3/25/07 at 11:32am
Thread Starter 
Whoops .. .. i made a mistake .. 192, 160 and 128 mp3 are switched :S.
Gonna fix it right away.

[EDIT] oh .. i didnt, iTunes uses makes a weird turn around 160kbps, switching form "an okay encoder" to "omg this sucks"...
Probably got something to do with a low-pass filter or EVIL APPLE PEOPLE ( ) who dont want MP3 to be better then AAC at higher bitrates.
Reply
post #127 of 222
3/25/07 at 12:38pm
hmm, 160 vbr for itunes mp3 is actually really great looking. looks better than almost everything else except for the nero 224 vbr or 320aac
Reply
post #128 of 222
3/25/07 at 3:30pm
Great. Now how do they SOUND compared to each other?

Sorry to beat a dead horse, BUT...
Reply
post #129 of 222
3/25/07 at 6:55pm
Nice job!
Reply
post #130 of 222
3/26/07 at 1:14am
NOOOOOOOOOOOOOOOOOOOOOOO!!!

I've encoded around hundreds of songs with iTunes at 320 mp3 VBR!!! Does this mean I should re-encode all my music with another encoder? If so, what would you recommend (for mac)?

Thanks!
Reply
post #131 of 222
3/26/07 at 7:00am

Pics

.
Dude:

Nice work, I dig it!

Do I understand correctly then, that VBR is, generally speaking, a better way to go for [sound quality] - if file size is not an issue I care about. It sounds counter-intuitive to my mind however, as I would think that VBR would be subject to more data loss and 'compansion' artifacts (is that a word?), compared to plain old Constant Bit Rate.

Andrew D.
cdnav.com

.
Reply
post #132 of 222
3/26/07 at 7:53am

Great Job!

Thanks and great work!

Don't listen to those naysayers who don't understand what these graphs depict! It's funny that some people think that even though a codec's spectrograph depicts gaps and distortion, it can somehow sound better than a codec whose spectrograph more closely resembles the original WAV's spectrograph.

It's almost like saying that headphones that have flatter frequency responses with less distortion may reproduce music worse than a headphone with a FR graph that shows wild peaks just because the headphone with the wild FR may have its wild peaks in "inaudible areas."

Flat FR + No Distortion = WIN no matter how you put it, just like a spectrograph that more closely resembles the original after compression = WIN no matter what you think. An intact spectrograph reflects high efficiency encoding. You don't need a subjective listening test confusing people even more.

The only significant problems about these test that I can think of is how each codec handles special cases (like cymbals) and that you should cutoff the graphs at maybe 16-18K as to not give head-fier's the impression that data above 16K matters. Most headphones suffer from severe attenuation at high frequencies and most music has very little audible noises above 16K.
Reply
post #133 of 222
3/26/07 at 12:49pm
how about WMP11 encoded mp3 file compare to iTunes encoded mp3 file?
Reply
post #134 of 222
3/26/07 at 12:56pm
Quote:
Originally Posted by ObiHuang View Post
Don't listen to those naysayers who don't understand what these graphs depict! It's funny that some people think that even though a codec's spectrograph depicts gaps and distortion, it can somehow sound better than a codec whose spectrograph more closely resembles the original WAV's spectrograph.

It's almost like saying that headphones that have flatter frequency responses with less distortion may reproduce music worse than a headphone with a FR graph that shows wild peaks just because the headphone with the wild FR may have its wild peaks in "inaudible areas."
Headphones do not intentionally discard inaudible data. Lossy codecs do. These graphs are nothing at all like a headphone's frequency response graph.
Reply
post #135 of 222
3/26/07 at 1:22pm
Quote:
Originally Posted by jewman View Post
I've encoded around hundreds of songs with iTunes at 320 mp3 VBR!!! Does this mean I should re-encode all my music with another encoder? If so, what would you recommend (for mac)?
Don't worry, your choice of encoder is propably fine. However, there have been some problems with the iTunes mp3-encoder in the past, but I don't know about the present situation. For mp3 lame is propably a better choice. iTunes AAc encoder is supposed to be very good.

I know this has been said numerous times before, but you really can't stress it enough. There is no point comparing lossy encoders using pictures. Encoders are meant for something you listen to, not something you look at. The very basic principle of encoding algorithms is to remove inaudible parts of the signal. But inaudible does not mean invisible.

If you decide to re-encode your songs, I would recommend program called Max, which is a very good ripping/encoding software for mac OS X.


(btw. Does someone know any good image->audio converters. I'm thinking about comparing some image compression algorithms using my headphones )
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Portable Source Gear
Head-Fi.org › Forums › Equipment Forums › Portable Source Gear › Lossy Audio Codec's Comparison [HUGE amount of pics] [iTunes UPDATE on p.7]