IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
Pre-beta builds
jmvalin
post May 27 2013, 06:11
Post #1


Xiph.org Speex developer


Group: Developer
Posts: 479
Joined: 21-August 02
Member No.: 3134



Greg posted a new win32 build of opus-tools at https://people.xiph.org/~greg/opus-tools_58d80ab.zip (updated build with bugfix)
This should normally be released as 1.1-beta in 1-2 weeks, so please report any issue you find with it. The main change in this build is the recent surround quality improvements, especially for 5.1. There were also speed optimizations and both the encoder and decoder should run faster.

This post has been edited by jmvalin: May 28 2013, 19:05
Go to the top of the page
+Quote Post
o-l-a-v
post May 27 2013, 09:55
Post #2





Group: Members
Posts: 126
Joined: 18-July 12
From: Norway
Member No.: 101545



I tell the encoder to use "VBR 128" and "complexity=10".
The outbut has a bitrate of 55.
VBR 144 -> 60 kbps

Whats up with that? I'm using TAudioConverter. Unless some commands have changed, there must be a bug with opusenc

This post has been edited by o-l-a-v: May 27 2013, 10:36
Go to the top of the page
+Quote Post
jmvalin
post May 27 2013, 11:08
Post #3


Xiph.org Speex developer


Group: Developer
Posts: 479
Joined: 21-August 02
Member No.: 3134



QUOTE (o-l-a-v @ May 27 2013, 04:55) *
I tell the encoder to use "VBR 128" and "complexity=10".
The outbut has a bitrate of 55.
VBR 144 -> 60 kbps

Whats up with that? I'm using TAudioConverter. Unless some commands have changed, there must be a bug with opusenc


It's not a bug, it's a feature :-) Seriously, what happens is that the encoder now reduces the rate when one speaker gets masked by another one. I haven't been able to re-calibrate because I'd need a wide variety of surround material. So for now, you pretty much have to use a higher rate.
Go to the top of the page
+Quote Post
o-l-a-v
post May 27 2013, 11:58
Post #4





Group: Members
Posts: 126
Joined: 18-July 12
From: Norway
Member No.: 101545



QUOTE (jmvalin @ May 27 2013, 12:08) *
QUOTE (o-l-a-v @ May 27 2013, 04:55) *
I tell the encoder to use "VBR 128" and "complexity=10".
The outbut has a bitrate of 55.
VBR 144 -> 60 kbps

Whats up with that? I'm using TAudioConverter. Unless some commands have changed, there must be a bug with opusenc


It's not a bug, it's a feature :-) Seriously, what happens is that the encoder now reduces the rate when one speaker gets masked by another one. I haven't been able to re-calibrate because I'd need a wide variety of surround material. So for now, you pretty much have to use a higher rate.


Interesting. So 60kbps should now be near transparent? (If thats the outcome of choosing 144kbps vbr smile.gif )

This post has been edited by o-l-a-v: May 27 2013, 12:07
Go to the top of the page
+Quote Post
jmvalin
post May 28 2013, 01:38
Post #5


Xiph.org Speex developer


Group: Developer
Posts: 479
Joined: 21-August 02
Member No.: 3134



QUOTE (jmvalin @ May 27 2013, 01:11) *
Greg posted a new win32 build of opus-tools at https://people.xiph.org/~greg/opus-tools_0fed074b.zip
This should normally be released as 1.1-beta in 1-2 weeks, so please report any issue you find with it. The main change in this build is the recent surround quality improvements, especially for 5.1. There were also speed optimizations and both the encoder and decoder should run faster.


Sorry, it appears we really messed up this build. PLEASE DON'T USE FOR NOW. We'll fix it.
Go to the top of the page
+Quote Post
IgorC
post May 28 2013, 14:43
Post #6





Group: Members
Posts: 1556
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803



From what I can see this build reduces bitrate significantly on quiet (low energy) parts. It has a great potential of reducing bitrate on low-freq/quiet content but also some issues on samples with quick variations of loudness.

Have tried Pink Floyd's Dark Side of the Moon at 96 kbps and some tracks go as low as 55-80 kbps without quality drops. Amazing.
But also have found a few other samples where it did quite bad.

This post has been edited by IgorC: May 28 2013, 14:45
Go to the top of the page
+Quote Post
SubV
post May 28 2013, 16:45
Post #7





Group: Members
Posts: 24
Joined: 27-November 08
From: Kursk, Russia
Member No.: 63333



Sorry, but encoding Dark Side of the Moon "at 96 kbps" sounds like blasphemy to me.
Go to the top of the page
+Quote Post
DonP
post May 28 2013, 17:38
Post #8





Group: Members (Donating)
Posts: 1471
Joined: 11-February 03
From: Vermont
Member No.: 4955



QUOTE (jmvalin @ May 27 2013, 20:38) *
Sorry, it appears we really messed up this build. PLEASE DON'T USE FOR NOW. We'll fix it.


It's good to hear some status even if it's a temporary one step back.
Go to the top of the page
+Quote Post
IgorC
post May 28 2013, 17:39
Post #9





Group: Members
Posts: 1556
Joined: 3-January 05
From: ARG/RUS
Member No.: 18803




Oh, You're right. Even Justin Bieber deserves higher bitrate.
Anyway, such a great way to celebrate 20th post.
Go to the top of the page
+Quote Post
jmvalin
post May 28 2013, 19:06
Post #10


Xiph.org Speex developer


Group: Developer
Posts: 479
Joined: 21-August 02
Member No.: 3134



QUOTE (jmvalin @ May 27 2013, 01:11) *
Greg posted a new win32 build of opus-tools at https://people.xiph.org/~greg/opus-tools_58d80ab.zip (updated build with bugfix)
This should normally be released as 1.1-beta in 1-2 weeks, so please report any issue you find with it. The main change in this build is the recent surround quality improvements, especially for 5.1. There were also speed optimizations and both the encoder and decoder should run faster.


OK, the problem should be fixed now and the build was updated. Please give this new build a try: https://people.xiph.org/~greg/opus-tools_58d80ab.zip
Go to the top of the page
+Quote Post
o-l-a-v
post May 28 2013, 20:25
Post #11





Group: Members
Posts: 126
Joined: 18-July 12
From: Norway
Member No.: 101545



I have a sample for you here, which i usually can abx with most lossy encoders and settings. Easily ABX'able at 144kbps opus (latest link), and I manage to ABX v0 vs v0.5 Lame 3.100. Seems like hard bass together with melody is hard for every encoder. The "resolution"/details of everything that is not bass, gets "reduced" compared to lossless.

Rock music, rap, pop i can't abx at v3-v4 (lame), but hardstyle and some house music i need v1-v0 to get satisfied with quality.

This post has been edited by o-l-a-v: May 28 2013, 20:33
Attached File(s)
 
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: 28th August 2014 - 05:27