IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Using FLAC to avoid problems with lossy -> lossy transcoding?
Jim P.
post Aug 31 2013, 19:50
Post #1





Group: Members
Posts: 9
Joined: 21-July 08
Member No.: 56074



My music collection is about 95% FLAC (nearly all ripped from my original CD's). But I do have a smattering of MP3's that I have acquired for one reason or another. On the side, I like to have everything also available in MP4 (for portable devices). Converting my FLAC to MP4 is of course no problem. But when I attempt to transcode the MP3 to MP4 I get a warning message saying that this will result in a further degradation of quality. (I could of course just leave the MP3's as-is, but I want to have everything uniform.)

I have two questions:

1) just how bad IS the loss of quality going from MP3 to MP4? Is this a significant concern? My MP4 settings are "Very High", around 300 kbps.

2) am I over-thinking the problem if I transcode MP3 > FLAC (which doesn't increase quality at all, but doesn't decrease it either), and then go from FLAC to MP4? Does that effectively work around whatever issue there is with lossy to lossy transcoding?

Thoughts?
Go to the top of the page
+Quote Post
saratoga
post Aug 31 2013, 20:21
Post #2





Group: Members
Posts: 4844
Joined: 2-September 02
Member No.: 3264



Just copy the mp3 files directly to the portable device. Do not transcoded.
Go to the top of the page
+Quote Post
Jim P.
post Aug 31 2013, 20:30
Post #3





Group: Members
Posts: 9
Joined: 21-July 08
Member No.: 56074



QUOTE (saratoga @ Aug 31 2013, 12:21) *
Just copy the mp3 files directly to the portable device. Do not transcoded.



Yeah, that's what I actually have done thus far. But it irks me to have two different formats when I'm so close to uniformity. I'm probably just being OCD.

Still curious about lossy > lossy transcoding, but I probably should be in a different forum for that anyway.....
Go to the top of the page
+Quote Post
Rollin
post Aug 31 2013, 20:35
Post #4





Group: Members
Posts: 189
Joined: 5-March 08
Member No.: 51815



mp3->flac->mp4 has the same quality as mp3->mp4
Go to the top of the page
+Quote Post
jensend
post Aug 31 2013, 20:44
Post #5





Group: Members
Posts: 143
Joined: 21-May 05
Member No.: 22191



That isn't over-thinking, it's just misunderstanding. If the only source you have is a lossy file, decoding the lossy format and then losslessly compressing the result only gives you a larger file with the same distortions as your lossy source. Encoding the resulting FLAC as MP4 won't give you a warning, because the encoder will no longer be able to tell there's distortions just from the file format, but the end result will be exactly the same loss of quality as the MP3-MP4 "direct" transcode.

(All you've done is stick an invertible operation and its inverse in the chain, which together are the same as doing nothing, just like adding a number and then subtracting it: decode MP3 to PCM -> losslessly compress PCM to FLAC -> decode FLAC to PCM, bit for bit identical to the PCM from decoding the MP3 -> encode PCM to MP4.)

Having everything "uniform" doesn't really provide any advantage. At least for the forseeable future, any device that supports MP4 will support MP3.

Please read the HA Wiki entry on transcoding carefully before proceeding any further.
Go to the top of the page
+Quote Post
Jim P.
post Aug 31 2013, 21:26
Post #6





Group: Members
Posts: 9
Joined: 21-July 08
Member No.: 56074



QUOTE (jensend @ Aug 31 2013, 12:44) *
Encoding the resulting FLAC as MP4 won't give you a warning, because the encoder will no longer be able to tell there's distortions just from the file format, but the end result will be exactly the same loss of quality as the MP3-MP4 "direct" transcode.


Sigh, I suspected as much. Thanks for the responses!
Go to the top of the page
+Quote Post
testyou
post Sep 1 2013, 00:20
Post #7





Group: Members
Posts: 99
Joined: 24-September 10
Member No.: 84113



You can try transcoding them, and then performing ABX tests to determine if the generation loss is a problem for you.

I'm happy to keep a variety of file formats if they are from lossy sources, you do not need arbitrary rules of uniformity.
Go to the top of the page
+Quote Post
nu774
post Sep 1 2013, 05:25
Post #8





Group: Developer
Posts: 514
Joined: 22-November 10
From: Japan
Member No.: 85902



If you want lossy -> lossy transcoding with fb2k, just do it directly.
lossy -> FLAC -> lossy only to shut up fb2k warning is a complete non-sense, and actually is worse than simple direct transcoding.
Since FLAC only supports integer PCM, the latter process inevitably involves with quantization (with or without dither / noise shape).
On the other hand, in case of direct lossy->lossy transcoding fb2k will use float32 as intermediate PCM format as far as encoder supports it. Unnecessary quantization doesn't take place.
Although difference may not be audible, you are just making it worse by adding chain of task. Do it simple.

Go to the top of the page
+Quote Post
db1989
post Sep 1 2013, 10:27
Post #9





Group: Super Moderator
Posts: 5275
Joined: 23-June 06
Member No.: 32180



To be clear, encoding lossy to lossy involves an intermediate uncompressed stage, just as encoding lossy to lossless; the losslessly compressed result in the former case will be the same as the intermediate decompression in the former case.

An interesting question that I like to pose in these recurring threads and ones with similar ideas, and which I donít recall having gotten an answer for, is a thought experiment that reframes the original question: How could an intermediate conversion to lossless protect against transcoding issues? In other words, what attributes of that process would grant it quality-restoring powers? There are none. With a basic understanding of the technology, such questions become impossible.
Go to the top of the page
+Quote Post
andrew_berge
post Sep 1 2013, 17:20
Post #10





Group: Members
Posts: 108
Joined: 17-November 09
From: Canada
Member No.: 75012



QUOTE (nu774 @ Sep 1 2013, 00:25) *
lossy -> FLAC -> lossy only to shut up fb2k warning is a complete non-sense, and actually is worse than simple direct transcoding.
Since FLAC only supports integer PCM, the latter process inevitably involves with quantization (with or without dither / noise shape).
On the other hand, in case of direct lossy->lossy transcoding fb2k will use float32 as intermediate PCM format as far as encoder supports it. Unnecessary quantization doesn't take place.
Although difference may not be audible, you are just making it worse by adding chain of task. Do it simple.


Huh. I never realized that.
The one thing i thought of is that FLAC can't handle samples above 0dB, while this can happen (and quite often, i've found) in mp3, resulting in a loss of quality (possibly audible?) when converting mp3 to FLAC.
Go to the top of the page
+Quote Post
greynol
post Sep 1 2013, 17:35
Post #11





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



QUOTE (nu774 @ Aug 31 2013, 21:25) *
Do it simple.

Which in this case is not to transcode at all.


--------------------
YOUR EYES CANNOT HEAR!!!!!!!!!!!
Go to the top of the page
+Quote Post
phofman
post Sep 1 2013, 18:54
Post #12





Group: Members
Posts: 283
Joined: 14-February 12
Member No.: 97162



The question is what DAC technology can handle samples above 0dbFS.
Go to the top of the page
+Quote Post
greynol
post Sep 1 2013, 19:21
Post #13





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



The question was whether lossy -> flac -> lossy was better than lossy -> lossy.

As to the off-tooic question about DACs and samples above 0dBFS, the worthwhile answer will only be found in an ABX test.


--------------------
YOUR EYES CANNOT HEAR!!!!!!!!!!!
Go to the top of the page
+Quote Post
nu774
post Sep 2 2013, 03:34
Post #14





Group: Developer
Posts: 514
Joined: 22-November 10
From: Japan
Member No.: 85902



QUOTE (phofman @ Sep 2 2013, 02:54) *
The question is what DAC technology can handle samples above 0dbFS.

You can just turn software volume down if you worry about lossy files with peaks above 0dBFS (or you can simply apply replaygain).
That's completely different from playing files that are already hard clipped.
Try the attached file. AAC is original with peak of +3dBFS or so, and FLAC one is converted from AAC using fb2k (without applying gain or something).
Attached File(s)
Attached File  3dB.zip ( 156.21K ) Number of downloads: 27
 
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: 23rd July 2014 - 15:38