IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
opus-tools 0.1.7 out
NullC
post Sep 9 2013, 01:28
Post #1





Group: Developer
Posts: 200
Joined: 8-July 03
Member No.: 7653



Here is the Opus-tools 0.1.7 announcement:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

We've tagged a new release of the opus-tools command line utility suite.

New in this release are

- - numerous build fixes.
- - update to opus 1.0.3 surround encoder api.
- - support for encoding from flac files.
- - better metadata support, including copying tags from flac input.

Source code is available from

http://downloads.xiph.org/releases/opus/op...ls-0.1.7.tar.gz
https://ftp.mozilla.org/pub/mozilla.org/opu...ls-0.1.7.tar.gz

I've made binary builds against prerelease opus 1.1 (today's git master)


https://ftp.mozilla.org/pub/mozilla.org/opu...0.1.7-win32.zip

https://ftp.mozilla.org/pub/mozilla.org/opu....7-macos.tar.gz

SHA-256 checksums:

e0f08d301555dffc417604269b5a85d2bd197f259c7d6c957f370ffd33d6d9cd
opus-tools-0.1.7.tar.gz
2094f7e41829e326dac0dbcf37f2f1a2072ae4b5ba743a4ea454347ee4aee424
opus-tools-0.1.7-macos.tar.gz
7ee311f790d2419c47793906b7946e778052b6c3ff9fb70c2f6934555fc8b02c
opus-tools-0.1.7-win32.zip

Thanks to everyone who contributed!

-r
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.14 (Darwin)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJSLOXsAAoJEEcAD3uxRB3vQzgH/iAzIai+MJ89C0SSPm9GOVrV
G1HgnRJwdBGxKGDvzLfY4RlpA1K98xRJFZl+iV2+wDC8B+2J8Hj2/+p40S21xSX2
nN9yk0KBFU++r9She+6MMqVla6YGx+o2Sro4pPw+xzerEHWcjE6DYeuWJSVek3gV
sYbBCJ3/4+4UuYEtUifcNB2kCCYzkv3k+KX1Rh0JHc9KoPLn+nGVuKAeo+GnmkPZ
KboSYIdCwBFQegRjUxGJQu1XsG6TANmQbYLLq7vEwsaNDnNL4P6LH2DOTgiTAprW
KSolgQkuPKwD0IBiD6Ljw2TuOPVqvVFuFSpBACnYQmZ4F7vlV0zg/xT4+zF3fFc=
=ULpg
-----END PGP SIGNATURE-----
Go to the top of the page
+Quote Post
eahm
post Sep 9 2013, 02:38
Post #2





Group: Members
Posts: 1058
Joined: 11-February 12
Member No.: 97076



Nice, thank you for this.


--------------------
/lwAsIimz
Go to the top of the page
+Quote Post
Anakunda
post Sep 9 2013, 06:45
Post #3





Group: Members
Posts: 458
Joined: 24-November 08
Member No.: 63072



Thank U, I'm a bit curious if prerelease opus 1.1 has some differences from opus 1.1 beta?
Go to the top of the page
+Quote Post
o-l-a-v
post Sep 9 2013, 09:49
Post #4





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



Anyone know if there ever will be x64 builds? Should be prioritized on such a new modern codec imo.
Go to the top of the page
+Quote Post
2012
post Sep 9 2013, 13:30
Post #5





Group: Members
Posts: 67
Joined: 7-February 12
Member No.: 96993



QUOTE (NullC @ Sep 9 2013, 02:28) *


Was a shorter release cycle for the library ever considered?
Speex should tell you very long release cycles are far from optimal.
Go to the top of the page
+Quote Post
o-l-a-v
post Sep 9 2013, 23:10
Post #6





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



With 0.1.7 I can easily ABX attached sample 250kbps (Complexity=10) vs original FLAC. Thats worse than MP3 (Lame 3.99.5) !
(ABX was last thing i did before i went to bed. Not optimal conditions)
CODE

foo_abx 1.3.4 report
foobar2000 v1.2.9
2013/09/10 00:03:29

File A: C:\Users\Olav\Documents\Stuff\Samples\ABX\Zatox - 2012 - Italian Hardstyle 024 - 01 - My Life.flac
File B: C:\Users\Olav\Desktop\TAC-OUT\Zatox - 2012 - Italian Hardstyle 024 - 01 - My Life_6.opus

00:03:29 : Test started.
00:03:58 : 01/01 50.0%
00:04:15 : 02/02 25.0%
00:04:32 : 03/03 12.5%
00:04:47 : 04/04 6.3%
00:05:02 : 05/05 3.1%
00:05:12 : 06/06 1.6%
00:05:25 : 07/07 0.8%
00:05:37 : 08/08 0.4%
00:05:54 : 09/09 0.2%
00:06:09 : 09/10 1.1%
00:06:36 : 10/11 0.6%
00:06:46 : 11/12 0.3%
00:06:58 : 12/13 0.2%
00:07:10 : 13/14 0.1%
00:07:21 : 14/15 0.0%
00:07:30 : 15/16 0.0%
00:07:41 : 16/17 0.0%
00:07:54 : 17/18 0.0%
00:08:05 : 18/19 0.0%
00:08:14 : 19/20 0.0%
00:08:22 : 20/21 0.0%
00:08:27 : Test finished.

----------
Total: 20/21 (0.0%)



I see some more problems as well:
- The higher vbr value I set for conversion, the more the output actual bitrate misses, always ending up with a lower bitrate.
- The total length of the song gets shortened by almost a second on some samples. Attached one for instance. Original: 00.26.873. Opus: 00.26.006

This post has been edited by o-l-a-v: Sep 9 2013, 23:15
Attached File(s)
Attached File  Zatox_-_2012_-_Italian_Hardstyle_024_-_01_-_My_Life.flac ( 3.79MB ) Number of downloads: 86
Attached File  Zatox_-_2012_-_Italian_Hardstyle_024_-_01_-_My_Life_6.opus ( 767.11K ) Number of downloads: 85
 
Go to the top of the page
+Quote Post
kabal4e
post Sep 10 2013, 02:50
Post #7





Group: Members
Posts: 8
Joined: 10-March 13
From: Waikato, NZ
Member No.: 107144



QUOTE (Anakunda @ Sep 9 2013, 17:45) *
Thank U, I'm a bit curious if prerelease opus 1.1 has some differences from opus 1.1 beta?

Tried Opus 1.1 prerelease and Opus 1.1 beta posted here a month ago on a big test sample.
CODE
Differences found in 1 out of 1 track pairs.

Comparing:
"...\test_1.1b_vbr064.opus"
"...\test_1.1b21_vbr064.opus"
Differences found: 109638 sample(s), starting at 207.1135000 second(s), peak: 0.0018016 at 533.1586459 second(s), 2ch

peak: 0.0018016
So, there are differences, but I'd be really surprised if someone could ABX that.

QUOTE (o-l-a-v @ Sep 10 2013, 10:10) *
- The higher vbr value I set for conversion, the more the output actual bitrate misses, always ending up with a lower bitrate.

I'd suggest that Opus doesn't have to hit the exact bitrate, the bitrate should be considered as a 'target quality'. And I also noticed that on some very loud and fully synthetic samples, like the one you've provided, it's not unusual that Opus uses lower bitrate than the desired one. Try some more 'classical' instruments such as piano or guitar and you'll get higher rates than expected.

QUOTE (o-l-a-v @ Sep 10 2013, 10:10) *
- The total length of the song gets shortened by almost a second on some samples. Attached one for instance. Original: 00.26.873. Opus: 00.26.006

Personally tried 64, 128, 192, 256 & 512 kbps and they all came exactly 0:26.873 as per original sample. Doublecheck your coding tool.

Cheers, A

This post has been edited by kabal4e: Sep 10 2013, 02:58
Go to the top of the page
+Quote Post
Measly Twerp
post Sep 12 2013, 11:44
Post #8





Group: Members
Posts: 6
Joined: 12-September 13
Member No.: 110044



The support for converting flac files is much appreciated, however something is amiss with the playback gain, after conversion all of my music is extremely quiet. What I've found is that the playback gain tends to be very low when compared to the replay gain value for the same track.

With replay gain on the original:
replaygain_track_gain=-7.76 dB
replaygain_album_gain=-8.50 dB
Playback gain: -13.5 dB

Without replay gain:
Playback gain: 0 dB

I'm not sure what is happening exactly, but it seems I either need a way to disable the playback gain calculation, or I have to remove all replay gain information from my flac archive.
Go to the top of the page
+Quote Post
NullC
post Sep 12 2013, 16:58
Post #9





Group: Developer
Posts: 200
Joined: 8-July 03
Member No.: 7653



QUOTE (Measly Twerp @ Sep 12 2013, 03:44) *
The support for converting flac files is much appreciated, however something is amiss with the playback gain, after conversion all of my music is extremely quiet. What I've found is that the playback gain tends to be very low when compared to the replay gain value for the same track.
I'm not sure what is happening exactly, but it seems I either need a way to disable the playback gain calculation, or I have to remove all replay gain information from my flac archive.


Are you just looking at the numbers? The playback gain there _is_ the album gain. OggOpus is specified to use the EBU reference level, so the numbers are offset from the replay gain ones. Any playback tool which is applying replay gain should play the flac and the opus at the same level.

Do you have a tool which isn't?
Go to the top of the page
+Quote Post
Measly Twerp
post Sep 12 2013, 21:54
Post #10





Group: Members
Posts: 6
Joined: 12-September 13
Member No.: 110044



QUOTE (NullC @ Sep 13 2013, 01:58) *
QUOTE (Measly Twerp @ Sep 12 2013, 03:44) *
The support for converting flac files is much appreciated, however something is amiss with the playback gain, after conversion all of my music is extremely quiet. What I've found is that the playback gain tends to be very low when compared to the replay gain value for the same track.
I'm not sure what is happening exactly, but it seems I either need a way to disable the playback gain calculation, or I have to remove all replay gain information from my flac archive.


Are you just looking at the numbers? The playback gain there _is_ the album gain. OggOpus is specified to use the EBU reference level, so the numbers are offset from the replay gain ones. Any playback tool which is applying replay gain should play the flac and the opus at the same level.

Do you have a tool which isn't?


Just tried playback in three players, two Quod Libet and Rhythmbox both play the FLAC at desired volumes and the opus file very quietly. However Banshee played them both at the correct level.

This leads me to think that Banshee is assuming a reference level that is different (and correct) to the other two. Originally the replay gain levels where calculated by Quod Libet.

After calculating the replay gain levels in foobar2000 I've got exactly the same problem playing back in Quod Libet and Rhythmbox.

Also, when using quick preview to play a file (no replay gain support) the audio levels are just as mismatched as when played in either of the above.
Go to the top of the page
+Quote Post
lazka
post Sep 13 2013, 01:03
Post #11





Group: Members
Posts: 58
Joined: 11-August 03
Member No.: 8316



QUOTE (NullC @ Sep 13 2013, 01:58) *
QUOTE (Measly Twerp @ Sep 12 2013, 03:44) *
The support for converting flac files is much appreciated, however something is amiss with the playback gain, after conversion all of my music is extremely quiet. What I've found is that the playback gain tends to be very low when compared to the replay gain value for the same track.
I'm not sure what is happening exactly, but it seems I either need a way to disable the playback gain calculation, or I have to remove all replay gain information from my flac archive.


Are you just looking at the numbers? The playback gain there _is_ the album gain. OggOpus is specified to use the EBU reference level, so the numbers are offset from the replay gain ones. Any playback tool which is applying replay gain should play the flac and the opus at the same level.

Do you have a tool which isn't?


I see the same problem with GStreamer 1.0.10, opus-tools 0.1.6, gst-launch-1.0 playbin..

the flac has the following tags:

replaygain_album_gain:-8.44 dB
replaygain_album_peak:1.0000
replaygain_track_gain:-8.82 dB
replaygain_track_peak:1.0000

after opusenc I get:

R128_TRACK_GAIN=-97
+ in the stream header: "Playback gain: -13.4414 dB"

and the opus file plays quieter by that difference (doing replaygain calc on the opus file using gstreamer gives me +4.68db)
Go to the top of the page
+Quote Post
Gainless
post Sep 13 2013, 14:25
Post #12





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



QUOTE (o-l-a-v @ Sep 10 2013, 00:10) *
With 0.1.7 I can easily ABX attached sample 250kbps (Complexity=10) vs original FLAC. Thats worse than MP3 (Lame 3.99.5) !
(ABX was last thing i did before i went to bed. Not optimal conditions)
CODE

foo_abx 1.3.4 report
foobar2000 v1.2.9
2013/09/10 00:03:29

File A: C:\Users\Olav\Documents\Stuff\Samples\ABX\Zatox - 2012 - Italian Hardstyle 024 - 01 - My Life.flac
File B: C:\Users\Olav\Desktop\TAC-OUT\Zatox - 2012 - Italian Hardstyle 024 - 01 - My Life_6.opus


Would be helpful if you could give a proper description of the artifatcs you're hearing. Nonetheless I've checked the sample either at 96 kbps and noticed some weird volume changes during the bass part, like from a limiter.
Go to the top of the page
+Quote Post
o-l-a-v
post Sep 13 2013, 16:48
Post #13





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



QUOTE (Gainless @ Sep 13 2013, 15:25) *
QUOTE (o-l-a-v @ Sep 10 2013, 00:10) *
With 0.1.7 I can easily ABX attached sample 250kbps (Complexity=10) vs original FLAC. Thats worse than MP3 (Lame 3.99.5) !
(ABX was last thing i did before i went to bed. Not optimal conditions)
CODE

foo_abx 1.3.4 report
foobar2000 v1.2.9
2013/09/10 00:03:29

File A: C:\Users\Olav\Documents\Stuff\Samples\ABX\Zatox - 2012 - Italian Hardstyle 024 - 01 - My Life.flac
File B: C:\Users\Olav\Desktop\TAC-OUT\Zatox - 2012 - Italian Hardstyle 024 - 01 - My Life_6.opus


Would be helpful if you could give a proper description of the artifatcs you're hearing. Nonetheless I've checked the sample either at 96 kbps and noticed some weird volume changes during the bass part, like from a limiter.


In my ears, the "crunch" and richness/ resolution is missing from the bass kicks + the melody is weaker after conversion. Thats how i can distinguish flac from opus.
Go to the top of the page
+Quote Post
Measly Twerp
post Sep 15 2013, 02:52
Post #14





Group: Members
Posts: 6
Joined: 12-September 13
Member No.: 110044



Is it possible to remove the playback gain information altogether? Nearly had a heart attack playing music last night, I thought I'd gotten my sound levels to balance out by applying replay gain again, but I jumped from one album to another and suddenly the volume was twice what it should've been and my amplifier shut itself off (at 500wrms per channel that's quite alarming I can assure you).
Go to the top of the page
+Quote Post
db1989
post Sep 15 2013, 02:58
Post #15





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



ReplayGain info can be removed by mp3tag, foobar2000, and probably many other programs.

Is there some reason you cannot simply instruct your player to ignore the RG info? Alternatively, tag all your files with it, so there will be no unadjusted tracks to give you an unpleasant surprise by being too loud, if I have interpreted your series of posts correctly.
Go to the top of the page
+Quote Post
Measly Twerp
post Sep 15 2013, 03:35
Post #16





Group: Members
Posts: 6
Joined: 12-September 13
Member No.: 110044



Playback gain != Replay gain.

But actually, on reflection the playback gain values are balanced, Black Sabbath is the same volume as much newer music, however the overall volume is about 1/3 of what it should be. Thankfully all of my music goes through a mixer before it goes to my amplifier so I can just turn the gain up 1/4, however I don't think that's a viable solution for most people.
Go to the top of the page
+Quote Post
lithopsian
post Mar 8 2014, 17:49
Post #17





Group: Members
Posts: 171
Joined: 27-February 14
Member No.: 114718



QUOTE (Measly Twerp @ Sep 15 2013, 03:35) *
Playback gain != Replay gain.

But actually, on reflection the playback gain values are balanced, Black Sabbath is the same volume as much newer music, however the overall volume is about 1/3 of what it should be. Thankfully all of my music goes through a mixer before it goes to my amplifier so I can just turn the gain up 1/4, however I don't think that's a viable solution for most people.

It has been explained why, but I'll try to clarify. The "playback gain" is deliberately targetted to a level about 5-6dB lower than the replaygain tags. Thus noticeably quieter. Some players will give you the ability to adjust to this so that music sounds the same even with different gain tag systems, but many just blindly play back what they see. Most players have the ability to completely ignore replay gain tags when they're being problematic, and some can remove or edit them.

Opus has its own gain system based on the "lower volume" R128 standard with a header gain (not a visible tag) that is supposed to always be applied (but obviously music player may decide not to), and a track gain tag (R128_TRACK_GAIN). Although this is the recommended gain for the codec, wide support for the replaygain* tags means that those may get placed in with an Opus stream. and can result in confusion. It is entirely undefined, up to each player, what happens in that situation. The Opus specification specifically says don't do it.
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: 29th August 2014 - 00:50