IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
turning M4A to something useful for my mp3player, transcoding iTunes AAC
HTA
post Nov 6 2012, 21:59
Post #1





Group: Members
Posts: 14
Joined: 5-November 12
From: Saudi Arabia
Member No.: 104339



Hello,

I have a large library of music purchased from iTunes mostly in LC AAC form in M4A non drm container

i want to transcode this lossy format to something usefull to play in my new MP3 player which only accepts these formats:

http://kb.sandisk.com/app/answers/detail/a...d-music-formats


i tried converting from lossy AAC to losseless Flac(0 compression) using default foobar2000 encoder

but i was able to hear diffrence in ABX tests up to 70% of the time( i can post logs)

what do you recommend for me to use to transcode and is foobar2000 AAC decoder good enough? can't shake the feeling that i am losing some information in the transcoding process using foobar2000's default encoder/decoder to transcode FLAC(0)



thanks
Go to the top of the page
+Quote Post
skamp
post Nov 6 2012, 22:51
Post #2





Group: Developer
Posts: 1343
Joined: 4-May 04
From: France
Member No.: 13875



QUOTE (HTA @ Nov 6 2012, 21:59) *
i want to transcode this lossy format to something usefull to play in my new MP3 player which only accepts these formats:


Instead of transcoding (which will only increase filesize if you go lossless, or degrade quality is you go lossy), you could install Rockbox, an alternative firmware that does support AAC (.m4a files from iTunes). No transcoding required with that option.

QUOTE (HTA @ Nov 6 2012, 21:59) *
but i was able to hear diffrence in ABX tests up to 70% of the time( i can post logs)

A 5% probality that you're guessing (or less) is required to determine with a high degree of confidence that you do hear a difference.

QUOTE (HTA @ Nov 6 2012, 21:59) *
can't shake the feeling that i am losing some information in the transcoding process using foobar2000's default encoder/decoder to transcode FLAC(0)


It is highly unlikely that foobar2000's decoder is doing anything wrong here. Also, if you still want to go the FLAC route, at least use the default compression setting (-5), you'll get smaller files.


--------------------
caudec.net
Go to the top of the page
+Quote Post
yourlord
post Nov 6 2012, 23:25
Post #3





Group: Members
Posts: 172
Joined: 1-March 11
Member No.: 88621



I'll echo the idea that you should just rockbox the thing and then transcoding becomes a moot issue.

I'm assuming that since you are transcoding to FLAC -0 that you seem to think there is a quality difference in the compression levels. There is no quality difference in FLAC compression levels. They are all completely lossless. You will get the exact same waveform output from flac regardless of compression level. The compression level is there to balance encoding workload with your requirements or available power. A really slow machine may not be able to encode flac -8 in real time or better but it may be able to encode flac -0 fast enough. No matter if you use -0 or -8, the resulting output once decoded will be 100% bit identical.

I agree with skamp that you're not successfully ABXing the 2 files. Flac, being lossless will produce exactly the same output that the aac decoder produced. There simply is no difference between listening to the aac file and listening to a transcode of the aac file from the same aac decoder to any lossless format.
Go to the top of the page
+Quote Post
DVDdoug
post Nov 6 2012, 23:42
Post #4





Group: Members
Posts: 2441
Joined: 24-August 07
From: Silicon Valley
Member No.: 46454



QUOTE
tried converting from lossy AAC to losseless Flac(0 compression) using default foobar2000 encoder
That doesn't make sense... In either case, the same data should be hitting the DAC. It's just a question of when the AAC file is decoded.

If something is different it's NOT caused by the lossless FLAC encoding/decoding.
Go to the top of the page
+Quote Post
HTA
post Nov 6 2012, 23:59
Post #5





Group: Members
Posts: 14
Joined: 5-November 12
From: Saudi Arabia
Member No.: 104339



Thanks guys it i learned alot

so the compression level are just regarding cpu resources, ok i will try more compressed FLAC in the future

i just done another(third) ABX test and i will attach the log file


i choose a song with dynamics to more emphasize the differences between formats if that makes any sense

same volume level and just closing my eyes and listening(so hard effort)



what i meant DVDdoug:

"tried converting from lossy AAC to losseless Flac(0 compression) using default foobar2000 encoder"

is i used all the encoding/decoding components from the stock foobar v1.1.14a. no manual command lines was used


regarding the Rockbox i will search the idea but i was avoiding over concerns of bricking the device. but seems to be the better way.


thanks
Go to the top of the page
+Quote Post
HTA
post Nov 7 2012, 00:00
Post #6





Group: Members
Posts: 14
Joined: 5-November 12
From: Saudi Arabia
Member No.: 104339



foo_abx 1.3.4 report
foobar2000 v1.1.14a
2012/11/07 01:19:26

File A: C:\Users\Hamza\Music\Flac\Madonna\Like a Prayer\Madonna - 01 - Like a Prayer.flac
File B: C:\Users\Hamza\Music\iTunes\iTunes Media\Music\Madonna\Like a Prayer\01 Like a Prayer.m4a

01:19:26 : Test started.
01:20:38 : 01/01 50.0%
01:21:58 : 02/02 25.0%
01:22:51 : 02/03 50.0%
01:23:02 : 03/04 31.3%
01:23:25 : 03/05 50.0%
01:24:02 : 04/06 34.4%
01:25:59 : 05/07 22.7%
01:27:33 : 06/08 14.5%
01:29:30 : 07/09 9.0%
01:29:55 : 07/10 17.2%
01:31:23 : 08/11 11.3%
01:32:15 : 08/12 19.4%
01:33:02 : 08/13 29.1%
01:34:39 : 09/14 21.2%
01:37:29 : 10/15 15.1%
01:39:06 : 11/16 10.5%
01:40:56 : 12/17 7.2%
01:41:53 : 13/18 4.8%
01:44:27 : 14/19 3.2%
01:45:18 : 15/20 2.1%
01:45:21 : Test finished.

----------
Total: 15/20 (2.1%)
Go to the top of the page
+Quote Post
JunkieXL
post Nov 7 2012, 01:14
Post #7





Group: Members
Posts: 359
Joined: 3-April 05
Member No.: 21165



There is no reason to up convert a lossy AAC file to a lossless FLAC file. The sound quality does not improve and you are increasing the file size by double at the least. That was what DVD Doug was referencing.

If you are transcoding a lossy file, transcode it to another lossy file; OGG, MP3, etc.
CK
Go to the top of the page
+Quote Post
skamp
post Nov 7 2012, 01:25
Post #8





Group: Developer
Posts: 1343
Joined: 4-May 04
From: France
Member No.: 13875



QUOTE (JunkieXL @ Nov 7 2012, 01:14) *
There is no reason to up convert a lossy AAC file to a lossless FLAC file.


Yes there is: converting to a different format for compatibility reasons with no loss of quality.

QUOTE (JunkieXL @ Nov 7 2012, 01:14) *
If you are transcoding a lossy file, transcode it to another lossy file; OGG, MP3, etc.


That would reduce quality, although it might or might not be audible. But the first solution is not a bad one.


--------------------
caudec.net
Go to the top of the page
+Quote Post
kornchild2002
post Nov 7 2012, 01:33
Post #9





Group: Members
Posts: 2043
Joined: 8-April 05
From: Cincinnati, OH
Member No.: 21277



Except that the resulting files are rather large. A 4 minute file with the same sound quality as the source AAC file (likely 256kbps) would be almost 10 times higher when using FLAC without gaining any quality. That's all fine and dandy if you want to sacrifice storage space on a device by that much (while not gaining quality) but I don't see a reason for it. Rockbox would be a much better solution since it would play the AAC files directly and the OP wouldn't need to fill their player up with lossless files that retain lossy quality.
Go to the top of the page
+Quote Post
nu774
post Nov 7 2012, 01:44
Post #10





Group: Developer
Posts: 476
Joined: 22-November 10
From: Japan
Member No.: 85902



QUOTE (kornchild2002 @ Nov 7 2012, 09:33) *
A 4 minute file with the same sound quality as the source AAC file (likely 256kbps) would be almost 10 times higher when using FLAC without gaining any quality.

Mistaken with 3 times?
256 * 10 exceeds the size of uncompressed PCM in redbook format.
Go to the top of the page
+Quote Post
HTA
post Nov 7 2012, 01:50
Post #11





Group: Members
Posts: 14
Joined: 5-November 12
From: Saudi Arabia
Member No.: 104339



i was hoping for a transcoding solution since i enjoy very much learning laugh.gif

QUOTE (skamp @ Nov 7 2012, 03:25) *
Yes there is: converting to a different format for compatibility reasons with no loss of quality.


makes sense that each time going lossy is losing information

also making sense is rockbox because it allows it (AAC) hassle free


but i would love to get back at why it sounded different for me(m4a vs flac0 transcoded via stock foobar2000) maybe i will try doing manual command lines for transcoding just for the sake of comparsion. maybe use daaf2 or somethting. i got to read more
Go to the top of the page
+Quote Post
skamp
post Nov 7 2012, 01:51
Post #12





Group: Developer
Posts: 1343
Joined: 4-May 04
From: France
Member No.: 13875



QUOTE (kornchild2002 @ Nov 7 2012, 01:33) *
A 4 minute file with the same sound quality as the source AAC file (likely 256kbps) would be almost 10 times higher when using FLAC without gaining any quality.


What? That would be 2560 kbps. Uncompressed 16/44.1 PCM (which iTunes AAC files decode to) is 1411.2 kbps (that's 5.5 times higher than 256 kbps), and my entire FLAC collection averages at 880 kbps (that's 3.4 times higher). A far cry from "10 times higher".

QUOTE (kornchild2002 @ Nov 7 2012, 01:33) *
Rockbox would be a much better solution since it would play the AAC files directly and the OP wouldn't need to fill their player up with lossless files that retain lossy quality.


Yes, definitely. I'm the one who suggested it in the first place smile.gif


--------------------
caudec.net
Go to the top of the page
+Quote Post
saratoga
post Nov 7 2012, 06:24
Post #13





Group: Members
Posts: 4718
Joined: 2-September 02
Member No.: 3264



QUOTE (HTA @ Nov 6 2012, 20:50) *
but i would love to get back at why it sounded different for me(m4a vs flac0 transcoded via stock foobar2000) maybe i will try doing manual command lines for transcoding just for the sake of comparsion. maybe use daaf2 or somethting. i got to read more


Did you actually decide to do 20 tests in advance, or just do tests until you got below 0.05?
Go to the top of the page
+Quote Post
HTA
post Nov 7 2012, 06:58
Post #14





Group: Members
Posts: 14
Joined: 5-November 12
From: Saudi Arabia
Member No.: 104339



good question. at first i was confused as to how to perform the test but after i learned how to notice the difference at 10/20 attempt(17%) i thought i should try to extend to 20 and stop and by the end i was going 8 consecutive good guesses. i was able to spot differences the M4A had much smoother sound to my ears and i'm sure i could get better result i could restart the test with better results now

the transcoded M4A sounded choppy at times in lifting music segments also stereo transition between the two channels was different(one was smoother in transition from right to left than the other)

i'm using the faad2 command line tool to write wav as i speak. maybe i'll compress it to flac(0) and restart the test just for comparsion. i now know that 0 compression isn't prefered but just to elimanate the variables in the test

This post has been edited by HTA: Nov 7 2012, 07:03
Go to the top of the page
+Quote Post
Kohlrabi
post Nov 7 2012, 07:43
Post #15





Group: Super Moderator
Posts: 953
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



I don't support the sentiment that you shouldn't just transcode the files to MP3 or OGG. Do it, perform some listening tests, and if everything's fine to you stick with it.


--------------------
Audiophiles live in constant fear of jitter.
Go to the top of the page
+Quote Post
Porcus
post Nov 7 2012, 08:56
Post #16





Group: Members
Posts: 1779
Joined: 30-November 06
Member No.: 38207



If you want to transcode, then my suggestion is that you keep a naming like e.g. ".m4a.mp3". Then for a file that sounds bad, you know where it came from and can encode a new one.

(Of course you keep the original!)


--------------------
One day in the Year of the Fox came a time remembered well
Go to the top of the page
+Quote Post
Nessuno
post Nov 7 2012, 09:09
Post #17





Group: Members
Posts: 422
Joined: 16-December 10
From: Palermo
Member No.: 86562



QUOTE (Kohlrabi @ Nov 7 2012, 08:43) *
I don't support the sentiment that you shouldn't just transcode the files to MP3 or OGG. Do it, perform some listening tests, and if everything's fine to you stick with it.

And it would be very peculiar if he perceived differences between AAC and its lossless transcoding but not between AAC and a lossy one! laugh.gif

Seriously, if he is really able to perceive differences (a 15/20 could still means guessing, after all), there must be something flawed in the process. I don't know how Foobar works, but for example, are the decoders and encoders used to generate the second file the same used to listen? If not (and given the premises) maybe one of them is buggy.

BTW: yes, 256kbps AAC files are roughly three times smaller than their FLAC counterpart.


--------------------
... I live by long distance.
Go to the top of the page
+Quote Post
nu774
post Nov 7 2012, 10:17
Post #18





Group: Developer
Posts: 476
Joined: 22-November 10
From: Japan
Member No.: 85902



Lossy codecs like AAC can have track peak higher than 1.0, and is decoded into float32 inside of fb2k.
Since FLAC cannot handle float format, quantization to int16 or something is required anyway, where some dither might be applied, and it can be clipped (if he didn't apply gain modification to avoid it).
Other AAC decoder might simply output to int16 by default.

In short, it's impossible obtain exact copy of AAC using FLAC.
Of course audibility is a different matter, though.
Go to the top of the page
+Quote Post
Nessuno
post Nov 7 2012, 10:56
Post #19





Group: Members
Posts: 422
Joined: 16-December 10
From: Palermo
Member No.: 86562



QUOTE (nu774 @ Nov 7 2012, 11:17) *
Lossy codecs like AAC can have track peak higher than 1.0, and is decoded into float32 inside of fb2k.
Since FLAC cannot handle float format, quantization to int16 or something is required anyway, where some dither might be applied, and it can be clipped (if he didn't apply gain modification to avoid it).

And if he does, audible differences in this scenario could derive simply from little differences in track gain. This will be compatible with the subjective differencies the OP is reporting in a precedent post ("smoother sound" and the like).

QUOTE
In short, it's impossible obtain exact copy of AAC using FLAC.

Do you mean "exact copy of decoded output"?


--------------------
... I live by long distance.
Go to the top of the page
+Quote Post
nu774
post Nov 7 2012, 11:23
Post #20





Group: Developer
Posts: 476
Joined: 22-November 10
From: Japan
Member No.: 85902



QUOTE (Nessuno @ Nov 7 2012, 18:56) *
Do you mean "exact copy of decoded output"?

Ah, yes.
Go to the top of the page
+Quote Post
Garf
post Nov 7 2012, 12:45
Post #21


Server Admin


Group: Admin
Posts: 4853
Joined: 24-September 01
Member No.: 13



QUOTE (nu774 @ Nov 7 2012, 10:17) *
Lossy codecs like AAC can have track peak higher than 1.0, and is decoded into float32 inside of fb2k.
Since FLAC cannot handle float format, quantization to int16 or something is required anyway, where some dither might be applied, and it can be clipped (if he didn't apply gain modification to avoid it).
Other AAC decoder might simply output to int16 by default.


The float32 decode is still going to have to be output through the soundcard, so it'll clip just as well on playback.

Given what happened here, a gain difference or other difference when decoding sounds most likely, though the differences reported don't particularly make a lot of sense. (It's also not entirely clear from the comments in this thread, but there appear to have been multiple ABX trials and only one was reported, so it's possible he's not actually hearing a difference)
Go to the top of the page
+Quote Post
HTA
post Nov 7 2012, 15:32
Post #22





Group: Members
Posts: 14
Joined: 5-November 12
From: Saudi Arabia
Member No.: 104339



i was NOT able to replicate last night result consistantly.

i feel i owe to tell you guys this

there are lots of variables in the tests. and this can drag alot of discussion of ABX theory. so i will keep it simple and update if there was reliable results after Transcoding to lossy.



thanks
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 20th April 2014 - 00:45