Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Sample Shows LAME 3.97 -V2 (old vbr) better than -V2 --vbr-new (Read 7340 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Sample Shows LAME 3.97 -V2 (old vbr) better than -V2 --vbr-new

It’s been awhile since I’ve given LAME a critical listen, and today I came up with some interesting results with an old sample (GBTINC):

Link to GBTINC Sample


I can easily ABX this sample with LAME 3.97 with:  -V2 --vbr-new

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/11/04 00:24:59

File A: C:\gbtinc_397_v2new.mp3
File B: C:\gbtinc.wav

00:24:59 : Test started.
00:26:39 : 01/01  50.0%
00:27:07 : 02/02  25.0%
00:27:30 : 03/03  12.5%
00:28:37 : 04/04  6.3%
00:29:00 : 05/05  3.1%
00:29:58 : 06/06  1.6%
00:30:08 : 07/07  0.8%
00:30:19 : 08/08  0.4%
00:30:45 : 09/09  0.2%
00:31:14 : 10/10  0.1%
00:31:28 : 11/11  0.0%
00:31:36 : Test finished.

 ---------- 
Total: 11/11 (0.0%)



For curiosity’s sake, I tried the same sample but this time without  --vbr-new.  I thought it sounded better, and an ABX confirms:

Code: [Select]
foo_abx 1.3.1 report
foobar2000 v0.9.4.5
2007/11/04 00:39:17

File A: C:\gbtinc_397_v2new.mp3
File B: C:\gbtinc_397_v2.mp3

00:39:17 : Test started.
00:40:07 : 01/01  50.0%
00:40:31 : 02/02  25.0%
00:40:40 : 03/03  12.5%
00:40:56 : 04/04  6.3%
00:42:32 : 05/05  3.1%
00:42:42 : 06/06  1.6%
00:42:56 : 07/07  0.8%
00:43:18 : 08/08  0.4%
00:43:45 : 09/09  0.2%
00:44:12 : 10/10  0.1%
00:44:20 : 11/11  0.0%
00:44:28 : Test finished.

 ---------- 
Total: 11/11 (0.0%)



So... I guess with this sample at least --vbr-new isn't so great.  I'm curious as to what others think.

Sample Shows LAME 3.97 -V2 (old vbr) better than -V2 --vbr-new

Reply #1 –
I'm pretty sure it's already known that there are a few rare cases where vbr old is better than vbr new

P.S. Love the track that's sampled from, Sy & Unknown are great.

Sample Shows LAME 3.97 -V2 (old vbr) better than -V2 --vbr-new

Reply #2 –
Quote from: enVias on
I'm pretty sure it's already known that there are a few rare cases where vbr old is better than vbr new


You're 100% right on that.  Haha, I should have said "ANOTHER sample where vbr old is better than vbr new"

Quote from: enVias on
P.S. Love the track that's sampled from, Sy & Unknown are great.


Yeah, I'm a big fan too!  Gotta love the happy2bhardcore cd's  =P

Sample Shows LAME 3.97 -V2 (old vbr) better than -V2 --vbr-new

Reply #3 –
I found a sample a month ago that LAME 3.90.3 APS and -V 2 --vbr-old to be transparent while -vbr-new does poor even at -V 0 --vbr-new. Also I have found a few tracks on my music collection that perform better on LAME 3.90 then LAME 3.97.

V0 --vbr-new Killer Sample:

http://www.hydrogenaudio.org/forums/index....showtopic=58152
"I never thought I'd see this much candy in one mission!"

Sample Shows LAME 3.97 -V2 (old vbr) better than -V2 --vbr-new

Reply #4 –
I hope I'm not getting the feeling that people would rather tolerate slower encodes with --vbr-old rather than help improve the sound quality of --vbr-new.

Sample Shows LAME 3.97 -V2 (old vbr) better than -V2 --vbr-new

Reply #5 –
Quote from: /mnt on
I found a sample a month ago that LAME 3.90.3 APS and -V 2 --vbr-old to be transparent while -vbr-new does poor even at -V 0 --vbr-new. Also I have found a few tracks on my music collection that perform better on LAME 3.90 then LAME 3.97.

V0 --vbr-new Killer Sample:

http://www.hydrogenaudio.org/forums/index....showtopic=58152


That's a great sample!  My hope is that changes can be made from all these samples to improve upcoming lame builds.  How are you finding -V 2 --vbr-new overall with lame 3.97 (and the 3.98 betas)?  So far, with the exception of a few sporadic samples... I'm finding quality to be pretty good.

Quote from: sld on
I hope I'm not getting the feeling that people would rather tolerate slower encodes with --vbr-old rather than help improve the sound quality of --vbr-new.


I'd 100% rather help improve --vbr-new.  My hope is that finding problem samples for --vbr-new will help the lame developers to use those samples to make positive changes to --vbr-new.  From past listening tests on HA it appears that --vbr-new is already of better quality than --vbr-old.  If the relatively few problem samples for --vbr-new can be fixed, it will surely increase quality even more 

Sample Shows LAME 3.97 -V2 (old vbr) better than -V2 --vbr-new

Reply #6 –
Quote from: MrDrew on
Quote from: /mnt on

I found a sample a month ago that LAME 3.90.3 APS and -V 2 --vbr-old to be transparent while -vbr-new does poor even at -V 0 --vbr-new. Also I have found a few tracks on my music collection that perform better on LAME 3.90 then LAME 3.97.

V0 --vbr-new Killer Sample:

http://www.hydrogenaudio.org/forums/index....showtopic=58152


That's a great sample!  My hope is that changes can be made from all these samples to improve upcoming lame builds.  How are you finding -V 2 --vbr-new overall with lame 3.97 (and the 3.98 betas)?  So far, with the exception of a few sporadic samples... I'm finding quality to be pretty good.

Quote from: sld on
I hope I'm not getting the feeling that people would rather tolerate slower encodes with --vbr-old rather than help improve the sound quality of --vbr-new.


I'd 100% rather help improve --vbr-new.  My hope is that finding problem samples for --vbr-new will help the lame developers to use those samples to make positive changes to --vbr-new.  From past listening tests on HA it appears that --vbr-new is already of better quality than --vbr-old.  If the relatively few problem samples for --vbr-new can be fixed, it will surely increase quality even more 


Here is some tests I did that LAME 3.97 -V 2 --vbr did worse then LAME 3.90.3 and most of the time LAME 3.98b did the same or better then LAME 3.90 (APS) at --vbr-new (V2).

http://www.hydrogenaudio.org/forums/index....showtopic=58061
"I never thought I'd see this much candy in one mission!"