IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Coming from FLAC files to WavPack
spicymeatball77
post Apr 11 2013, 20:58
Post #1





Group: Members
Posts: 26
Joined: 16-January 09
Member No.: 65658



I have just recently gotten a Sansa Clip+ and put RockBox on it. I have been running my FLAC files on the device and have switched to WavPack. While I love the lossless quality, I would like to get more content on my 32GB sd card. I'm considering moving to high bitrate lossy WavPack. I have a fairly discerning ear, so I'm not certain this is even a good idea. I'd appreciate advice from anyone who has performed a similar process. Is the lossy format transparent? Does the soundstage suffer much?

I'm thinking about using the -b400fx3 switch after playing with them a bit.
Go to the top of the page
+Quote Post
probedb
post Apr 11 2013, 21:30
Post #2





Group: Members
Posts: 1193
Joined: 6-September 04
Member No.: 16817



Only you can test that. Do some ABXing and find out.
Go to the top of the page
+Quote Post
Gainless
post Apr 11 2013, 21:36
Post #3





Group: Members
Posts: 169
Joined: 28-October 11
Member No.: 94764



If you're going for lossy instead of lossless anyway, it would make more sense to use the codecs which are made for this purpose from the beginning. Vorbis and AAC (not Nero or FAAC!) should be the best there, so I guess you should give these at least a try.
Go to the top of the page
+Quote Post
ChronoSphere
post Apr 11 2013, 21:58
Post #4





Group: Members
Posts: 448
Joined: 11-March 07
Member No.: 41384



I went with wavpack hybrid to not have to keep two collections of the files. On my pc I have the .wv + .wvc, on my clip+ only the .wv
Went with 385kpbs for the lossy part because it somehow made the combined size smaller. Probably because less has to be corrected at higher bitrate.

You could try running the battery bench with wavpack vs. other codecs. For my clip I had a 10 minute longer runtime than flac, but I didn't test the other lossy format because having to convert the music every time I want to add music to my player is too much of a hassle for me.
Go to the top of the page
+Quote Post
spicymeatball77
post Apr 11 2013, 22:19
Post #5





Group: Members
Posts: 26
Joined: 16-January 09
Member No.: 65658



QUOTE (Gainless @ Apr 11 2013, 15:36) *
If you're going for lossy instead of lossless anyway, it would make more sense to use the codecs which are made for this purpose from the beginning.

Are you saying that because they were designed for lossy they have psychoacoustic models already? Or do you recommend for some other reason?
Go to the top of the page
+Quote Post
Gainless
post Apr 11 2013, 23:04
Post #6





Group: Members
Posts: 169
Joined: 28-October 11
Member No.: 94764



QUOTE (spicymeatball77 @ Apr 11 2013, 23:19) *
QUOTE (Gainless @ Apr 11 2013, 15:36) *
If you're going for lossy instead of lossless anyway, it would make more sense to use the codecs which are made for this purpose from the beginning.

Are you saying that because they were designed for lossy they have psychoacoustic models already? Or do you recommend for some other reason?

Yes, modern lossy codecs are not intended to give any bit-perfect output, but to "fool" your ears with more or less complex psymodels. Lossy WavPack for sure has some kind of psymodel too, but not as advanced. There's actually no reason in your situation to use anything different but Mp3, AAC, etc.: They will give you better quality, while also being far more supported. You could start with Lame or the AAC encoder from Winamp/Itunes at 192 kb/s and ABX (a program for blind testing) it against WavPack at similar bitrates.

This post has been edited by Gainless: Apr 11 2013, 23:07
Go to the top of the page
+Quote Post
greynol
post Apr 11 2013, 23:05
Post #7





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



If you haven't already, please read this prerequisite:
http://www.hydrogenaudio.org/forums/index....showtopic=16295

There is no point in proceeding with the discussion until this has been satisfied.


--------------------
YOUR EYES CANNOT HEAR!!!!!!!!!!!
Go to the top of the page
+Quote Post
shadowking
post Apr 12 2013, 02:26
Post #8





Group: Members
Posts: 1523
Joined: 31-January 04
Member No.: 11664



QUOTE (spicymeatball77 @ Apr 12 2013, 06:58) *
I have just recently gotten a Sansa Clip+ and put RockBox on it. I have been running my FLAC files on the device and have switched to WavPack. While I love the lossless quality, I would like to get more content on my 32GB sd card. I'm considering moving to high bitrate lossy WavPack. I have a fairly discerning ear, so I'm not certain this is even a good idea. I'd appreciate advice from anyone who has performed a similar process. Is the lossy format transparent? Does the soundstage suffer much?

I'm thinking about using the -b400fx3 switch after playing with them a bit.


The lossy mode is definitely transparent. I find 300k to be perfect quality on normal music while 350k (-b350x5) yields excellent quality for even the most demanding stuff.




--------------------
Wavpack -b450
Go to the top of the page
+Quote Post
Kohlrabi
post Apr 12 2013, 05:12
Post #9





Group: Super Moderator
Posts: 1002
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



QUOTE (shadowking @ Apr 12 2013, 02:26) *
The lossy mode is definitely transparent. I find 300k to be perfect quality on normal music while 350k (-b350x5) yields excellent quality for even the most demanding stuff.
Is there any incentive to waste about 160-200 kbps by not using a modern MP3 or AAC encoder?

This post has been edited by Kohlrabi: Apr 12 2013, 05:12


--------------------
Audiophiles live in constant fear of jitter.
Go to the top of the page
+Quote Post
shadowking
post Apr 12 2013, 10:06
Post #10





Group: Members
Posts: 1523
Joined: 31-January 04
Member No.: 11664



QUOTE (Kohlrabi @ Apr 12 2013, 15:12) *
QUOTE (shadowking @ Apr 12 2013, 02:26) *
The lossy mode is definitely transparent. I find 300k to be perfect quality on normal music while 350k (-b350x5) yields excellent quality for even the most demanding stuff.
Is there any incentive to waste about 160-200 kbps by not using a modern MP3 or AAC encoder?



I guess if one does not want multiple collections or transcoding there is yet another option in the encoding world. I think a traditional encoder @ 150..200 yields excellent if not transparent quality in most cases . If one can be happy with with a single mp3 library there is an advantage with universal playback and small files, However many here are managing multiple libraries and that somewhat the advantage of non-mp3/aac formats.


--------------------
Wavpack -b450
Go to the top of the page
+Quote Post
spicymeatball77
post Apr 12 2013, 14:28
Post #11





Group: Members
Posts: 26
Joined: 16-January 09
Member No.: 65658



I did some ABX testing and took Gainless' advice to try a codec specifically created for high quality lossy encoding. I compared WavPack lossless to Ogg q 10. The test results tell me that I really can't tell the difference. I realize now that I should have run both tests in a row, but it doesn't detract from the idea that I just don't know the difference, especially when you consider that on test 1 I may have gotten it correct but I thought the Ogg sounded better.

CODE
foo_abx 1.3.4 report
foobar2000 v1.2.4
2013/04/12 08:08:26

File A: C:\Users\...\Desktop\03-Holy.ogg
File B: C:\Users\...\Desktop\03-Holy.wv

08:08:26 : Test started.
08:11:19 : 01/01 50.0%
08:12:10 : Test finished.

----------
Total: 1/1 (50.0%)


CODE
foo_abx 1.3.4 report
foobar2000 v1.2.4
2013/04/12 08:12:59

File A: C:\Users\...\Desktop\03-Quake.ogg
File B: C:\Users\...\Desktop\03-Quake.wv

08:12:59 : Test started.
08:17:34 : 00/01 100.0%
08:17:54 : Test finished.

----------
Total: 0/1 (100.0%)
Go to the top of the page
+Quote Post
pdq
post Apr 12 2013, 14:37
Post #12





Group: Members
Posts: 3371
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



QUOTE (spicymeatball77 @ Apr 12 2013, 09:28) *
I realize now that I should have run both tests in a row, but it doesn't detract from the idea that I just don't know the difference, especially when you consider that on test 1 I may have gotten it correct but I thought the Ogg sounded better.

You misunderstand the purpose of the test. It is not to decide which sounds better, but whether or not something is indistinguishable from the original.

In this case the WavPack lossles IS the original, and the test is whether or not the ogg sounds different. Whether you thought the ogg sounded better or worse is immaterial, as long as you could tell that it sounded different.

FWIW, it is highly unlikely that you could tell an ogg q 10 from the original.
Go to the top of the page
+Quote Post
spicymeatball77
post Apr 12 2013, 15:03
Post #13





Group: Members
Posts: 26
Joined: 16-January 09
Member No.: 65658



QUOTE (pdq @ Apr 12 2013, 08:37) *
Whether you thought the ogg sounded better or worse is immaterial, as long as you could tell that it sounded different.

FWIW, it is highly unlikely that you could tell an ogg q 10 from the original.

Agreed, I do not feel I can tell the difference. I only mentioned that I thought the ogg file "sounded better" is just further evidence to support that any differences were merely perceived and not actual.
Go to the top of the page
+Quote Post
greynol
post Apr 12 2013, 16:00
Post #14





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



Placebo effect in action. Did you read the link I gave? It essentially tells you to do anything but run tests with a single trial.

If your interest is the preservation of soundstage, you should really begin testing at 80kbits and the work your way up until you find something that satisfies your needs. Even with the most discerning ears, other problems appear long before soundstage is compromised.

Regarding multiple collections, what is the difference between having lossy + lossless and lossy + correction? You have two sets of files either way.

Again, I think time should be spent ensuring the OP understands how to judge performance of lossy encoding before being complicit in reinforcing the notion that he "needs" to use a grossly inefficient format. This is especially the case when there was an expressed desire to fit more music on a portable.

This post has been edited by greynol: Apr 12 2013, 17:43


--------------------
YOUR EYES CANNOT HEAR!!!!!!!!!!!
Go to the top of the page
+Quote Post
spicymeatball77
post Apr 12 2013, 16:23
Post #15





Group: Members
Posts: 26
Joined: 16-January 09
Member No.: 65658



QUOTE (greynol @ Apr 12 2013, 10:00) *
Placebo effect in action. Did you read the link I gave? It essentially tells you to do anything but run tests with a single trial.

If your interest is the preservation of soundstage, you should really begin testing at 80kbits and the work your way up until you find something that satisfies your need. Even with the most discerning ears, other other problems appear long before soundstage is compromised.

Thank you, I plan on doing a better analysis next time especially considering your soundstage advice. Will post results if interested.
Go to the top of the page
+Quote Post
ChronoSphere
post Apr 12 2013, 16:23
Post #16





Group: Members
Posts: 448
Joined: 11-March 07
Member No.: 41384



QUOTE (greynol @ Apr 12 2013, 17:00) *
Regarding multiple collections, what is the difference between having lossy + lossless and lossy + correction? You have two sets of files either way.
Yes, but you only have one set of metadata complete with covers, replaygain, custom tags etc. When transcoding from foobar, it doesn't preserve replay gain tags or cover (or I don't know how), so I have to go back and re-tag the files after conversion. With wv+wvc I just copy the wv and I'm done.

Not to mention pure lossy (say mp3) + lossless (say wv) uses more space than wv+wvc

This post has been edited by ChronoSphere: Apr 12 2013, 16:24
Go to the top of the page
+Quote Post
greynol
post Apr 12 2013, 16:27
Post #17





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



That sounds like a problem with your choice in file management software.

QUOTE (ChronoSphere @ Apr 12 2013, 08:23) *
Not to mention pure lossy (say mp3) + lossless (say wv) uses more space than wv+wvc

For use on a portable?

Taking this comment generally, surely you aren't taking Kohlrabi's point into account. Are you prepared to back this up with data that includes double-blind listening tests?

This post has been edited by greynol: Apr 12 2013, 16:58


--------------------
YOUR EYES CANNOT HEAR!!!!!!!!!!!
Go to the top of the page
+Quote Post
ChronoSphere
post Apr 12 2013, 18:09
Post #18





Group: Members
Posts: 448
Joined: 11-March 07
Member No.: 41384



You're more than welcome to suggest me a (free) conversion software that does keep all the tags *shrug*

As for the size part, I'm not sure what you mean. Having two separate collections would mean you have, for example

- one wv of a song, 16,6 MB
- one mp3 of a song, 4,52 MB

This totals in ~21MB size

Having the same in wv+wvc would result in

- one wv of a song, ~4,6Mb
- one wvc of a song, ~12MB

totalling in ~17MB MB size. That's 4Mb more space per song, and I have a big music library.
I don't particularly care about "hdd space being cheap", I don't have the money to afford more anyway.

ABXing would not change the fact that keeping both the lossless and the lossy results in higher space usage, while wavpack hybrid has both lossy and lossless in one package at nearly the same size as the lossless alone. Thus lossy, no matter at which bitrate is a +alpha on space usage.

Of course going pure lossy would give you much better space efficiency, because the transparency can be reached with much lower bitrate than going pure lossless. The reason I go hybrid is not because I am claiming to hear the difference between lossless and lossy (which I don't), but because of space and metadata reasons. When on my PC I am using my ripped lossless archive, when I'm on my portable devices, I use the lossy part.
Go to the top of the page
+Quote Post
greynol
post Apr 12 2013, 18:18
Post #19





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



I didn't recall the correction file being smaller than the lossless file by about the size of the lossy file.

Thanks for pointing out what I really should have investigated prior to posting.


--------------------
YOUR EYES CANNOT HEAR!!!!!!!!!!!
Go to the top of the page
+Quote Post
ChronoSphere
post Apr 12 2013, 18:28
Post #20





Group: Members
Posts: 448
Joined: 11-March 07
Member No.: 41384



The size of the correction file depends on how low you go on the lossy bitrate (since it has to correct more for a low bitrate lossy). I experimented with the bitrates and at 385kbps the wv+wvc results in files that are around the size or smaller than flaCL -11, while going down on the bitrate (~250) would result in wv+wvc being larger than flaCL -11, so I put that as my "sweet spot" since I didn't want my library to grow.
Go to the top of the page
+Quote Post
BearcatSandor
post Apr 13 2013, 02:11
Post #21





Group: Members
Posts: 149
Joined: 18-May 10
From: Montana, USA
Member No.: 80732



I found the same sweet spot. I go with -b 384. Does it make any difference that 384 is evenly divisible by 8 and 385 isn't, given that wavpack doesn't use floats in it math?

(if this is question that makes me look hebetudinous, i apologize in advance :") )


--------------------
Music lover and recovering high end audiophile
Go to the top of the page
+Quote Post
spicymeatball77
post Apr 13 2013, 18:08
Post #22





Group: Members
Posts: 26
Joined: 16-January 09
Member No.: 65658



This time I ran an ABX on FLAC vs. OGG (q6). It appears that I could tell a difference, but got fooled a couple times. Apparently there was a 22.7% chance I guessed on them all.
CODE
foo_abx 1.3.4 report
foobar2000 v1.1.6
2013/04/13 11:52:02

File A: C:\Users\...\Desktop\12-The Menzingers-I Can't Seem To Tell.flac
File B: C:\Users\...\Desktop\12-The Menzingers-I Can't Seem To Tell.ogg

11:52:02 : Test started.
11:54:41 : 01/01 50.0%
11:56:19 : 01/02 75.0%
11:57:44 : 02/03 50.0%
11:59:55 : 03/04 31.3%
12:00:12 : 04/05 18.8%
12:01:36 : 04/06 34.4%
12:04:12 : 05/07 22.7%
12:04:12 : Test finished.

----------
Total: 5/7 (22.7%)


Whereas FLAC vs. OGG Q8 had me completely fooled. I think I found my bitrate. Thanks guys!
CODE
foo_abx 1.3.4 report
foobar2000 v1.1.6
2013/04/13 12:13:39

File A: D:\Clip+\OGG-8\Alkaline Trio-This Addiction-2010\12-Kick Rocks.ogg
File B: D:\FLAC\#-D\Alkaline Trio\2010-This Addiction\12-Alkaline Trio-Kick Rocks.flac

12:13:39 : Test started.
12:14:19 : 01/01 50.0%
12:15:57 : 02/02 25.0%
12:17:02 : 02/03 50.0%
12:18:24 : 02/04 68.8%
12:18:41 : 02/05 81.3%
12:19:40 : 02/06 89.1%
12:20:03 : 02/07 93.8%
12:20:59 : 02/08 96.5%
12:21:00 : Test finished.

----------
Total: 2/8 (96.5%)


This post has been edited by spicymeatball77: Apr 13 2013, 18:23
Go to the top of the page
+Quote Post
Kohlrabi
post Apr 13 2013, 18:34
Post #23





Group: Super Moderator
Posts: 1002
Joined: 12-March 05
From: Kiel, Germany
Member No.: 20561



QUOTE (spicymeatball77 @ Apr 13 2013, 18:08) *
This time I ran an ABX on FLAC vs. OGG (q6). It appears that I could tell a difference, but got fooled a couple times. Apparently there was a 22.7% chance I guessed on them all.
CODE
foo_abx 1.3.4 report
foobar2000 v1.1.6
2013/04/13 11:52:02

File A: C:\Users\...\Desktop\12-The Menzingers-I Can't Seem To Tell.flac
File B: C:\Users\...\Desktop\12-The Menzingers-I Can't Seem To Tell.ogg

11:52:02 : Test started.
11:54:41 : 01/01 50.0%
11:56:19 : 01/02 75.0%
11:57:44 : 02/03 50.0%
11:59:55 : 03/04 31.3%
12:00:12 : 04/05 18.8%
12:01:36 : 04/06 34.4%
12:04:12 : 05/07 22.7%
12:04:12 : Test finished.

----------
Total: 5/7 (22.7%)
You'd want to (at least!) achieve 5% confidence to have a conclusive result. With your result the 7 trials are not enough to be conclusive.

This post has been edited by Kohlrabi: Apr 13 2013, 18:48


--------------------
Audiophiles live in constant fear of jitter.
Go to the top of the page
+Quote Post
greynol
post Apr 13 2013, 20:01
Post #24





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



Fooled, or happen to guess incorrectly? That first log does not show a distinction.

Do at least 12 trials and do not have the results revealed until the test is finished.

This post has been edited by greynol: Apr 13 2013, 20:04


--------------------
YOUR EYES CANNOT HEAR!!!!!!!!!!!
Go to the top of the page
+Quote Post
Porcus
post Apr 29 2013, 07:45
Post #25





Group: Members
Posts: 1842
Joined: 30-November 06
Member No.: 38207



[on synchronizing tagsets]

QUOTE (greynol @ Apr 12 2013, 17:27) *
That sounds like a problem with your choice in file management software.


Hm? Any hint to share?


--------------------
One day in the Year of the Fox came a time remembered well
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 - 15:10