HydrogenAudio

Hydrogenaudio Forum => Validated News => Topic started by: john33 on 2003-09-07 19:00:30

Title: OggVorbis updates
Post by: john33 on 2003-09-07 19:00:30
Following some CVS updates to the libraries and the addition of the FLAC reading module to the standard oggenc, I've done some rationalising of the encoders offered at Rarewares.

The only version of oggenc now on offer is oggenc 2.3. This is oggenc 2.2 + oggencFLAC + oggencUNICODE. It is available as Post 1.0 CVS standard compile, P4 optimised compile and GT3b1 standard compile.

I've also updated the oggdropXPd complies to reflect the library changes.

Title: OggVorbis updates
Post by: rjamorim on 2003-09-07 19:06:51
Quote
oggencUNICODE

Does it work on Win9x?
Title: OggVorbis updates
Post by: john33 on 2003-09-07 19:12:31
Quote
Quote
oggencUNICODE

Does it work on Win9x?

Someone still running Win9x will have to tell me!!  However, this is activated as an option, so the message is - if it doesn't work, don't use it!!!
Title: OggVorbis updates
Post by: rjamorim on 2003-09-07 19:19:07
Yeah, indeed, it doesn't seem to work.

On my virtualized Win98:

Code: [Select]
C:\>oggenc2 --utf8 Chariots.wav
ERROR: Cannot open input file "Chariots.wav": Bad file descriptor


Thanks for the news
Title: OggVorbis updates
Post by: unplugged on 2003-09-08 09:48:17
Quote
Following some CVS updates to the libraries...

have you found any new additions or fixes on the Vorbis side?
Title: OggVorbis updates
Post by: john33 on 2003-09-08 10:07:37
I think you'll find the answer to that here: http://www.hydrogenaudio.org/forums/index....topic=13022&hl= (http://www.hydrogenaudio.org/forums/index.php?showtopic=13022&hl=)
Title: OggVorbis updates
Post by: kjoonlee on 2003-09-08 11:19:53
Quote
have you found any new additions or fixes on the Vorbis side?

Probably incomplete list, quoted from xiph.org's wiki (http://wiki.xiph.org/Release101):
Quote
libvorbis/libvorbisfile/libvorbisenc 1.0.1
  • Fixes to errors in the Vorbis specification documentation.

  • Specification converted to DocBook format.

  • Added timebase conversion call to libvorbis.

  • Added timebase conversion call to libvorbis.

  • Fixes ov_time_tell() bug for non-seekable streams.

  • Various decoding fixes.

  • Addition of ov_crosslap() which allows smooth transitions between two vorbis segments.

  • Numerous seeking fixes and optimizations.

  • Fixed bugs affecting decode of one-audio-page samples.

  • Added ov_halfrate() which allows a "free" conversion from the source sample rate to half of the source rate.

  • Improved handling of quiet signals in low bitrate modes
Title: OggVorbis updates
Post by: rjamorim on 2003-09-08 16:39:04
Hrm... I believe John33's compiles aren't 1.0.1. AFAIK, 1.0.1 hasn't even been announced yet. I think they are partially 1.0.1
Title: OggVorbis updates
Post by: kjoonlee on 2003-09-08 16:44:23
Oops, I should have made that clear. Thanks for pointing that out.
Title: OggVorbis updates
Post by: LCtheDJ on 2003-09-08 22:26:26
For me, Oggenc23.exe P4 optimised (Brit spelling?) version encodes slower than the non-P4 optimised version running on a P4.

I've found this with the Oggenc2.2 also.  I encoded from the same wav file using Speek's multi frontend on a P4 with Windows 2000.  Here's the command line switches I used: "-q 1 --advanced-encode-option lowpass_frequency=19".  The non-P4 version finishes about 2 seconds sooner and makes a file about 70 bytes smaller.

I don't know why the version optimised for a Pentium 4 processor runs slower than the version not optimised, even though running on a P4.

Has anyone else gotten results like this?
Title: OggVorbis updates
Post by: john33 on 2003-09-08 22:51:25
Comment noted, thanks.  I'll wait until there are any other takers on this. And, yes, Brit spelling!! 

The optimisation is only with the ICL 7.1 compiler using the P4 specific options.
Title: OggVorbis updates
Post by: LocustFurnace on 2003-09-10 11:22:08
Quote
I don't know why the version optimised for a Pentium 4 processor runs slower than the version not optimised, even though running on a P4.

I am running Win98 on an Athlon 900, using the UNICODE switch does not work either. the 2 seconds slower also was due to more accuracy in the encoder. Since I had posted a message about it being slower when John compiled the POST CVS version a couple months ago. He had said it was due to more acuracy in the encoder. and i am not using the P4 optimisations either.

oggenc2.3  -q3.02 # Oggenc2.3 v1.0 (libVorbis Post 1.0 CVS)
File length:   4m 52.0s
Elsp:      0m 56.0s
Rate:      5.2301
ABR:      109.2kb/s

oggenc1.0 -q3.02
File length:   4m 52.0s
Elsp:      0m 54.0s
Rate:      5.4238
ABR:      109.0kb/s
Title: OggVorbis updates
Post by: john33 on 2003-09-10 12:12:56
In the latest CVS, the library info date, as added into the headers, has now been changed to read "Xiph.Org libVorbis I 20030909". This looks to me as though 1.0.1 is now here, although it has not been formally announced.

I have recompiled everything on the basis that it is only the announcement that is lacking, maybe!!! 

Do you people want these new compiles uploaded, or do you want to wait a little for the actual announcement?
Title: OggVorbis updates
Post by: Slo Mo Snail on 2003-09-10 12:31:54
Code: [Select]
{"Xiph.Org libVorbis I 20030919", "1.0.1"}


so it will take at least 9 days until it gets announced...
Title: OggVorbis updates
Post by: Atlantis on 2003-09-10 12:35:12
I'm curious by nature, so I want the new compiles. 

But, I think is better not to upload them to rarewares, as vorbis 1.01 is not announced and those compiles may confuse people.

bye & thanks
Title: OggVorbis updates
Post by: kjoonlee on 2003-09-10 12:36:18
Quote
Code: [Select]
{"Xiph.Org libVorbis I 20030919", "1.0.1"}


so it will take at least 9 days until it gets announced...


That's the 1.0.1 identifier in the ogginfo source, but the libs have 20030909 set as the vendor string.

I suspect one of them is a typo, and IMHO, I think the ogginfo source is to blame.
Title: OggVorbis updates
Post by: john33 on 2003-09-10 12:59:14
Quote
I suspect one of them is a typo, and IMHO, I think the ogginfo source is to blame.

I'm sure you're right.

@Atlantis: I'll upload them to my own site and post the links a little later.
Title: OggVorbis updates
Post by: john33 on 2003-09-10 14:50:16
The following (1.0.1?) compiles are now available:

My oggdec: http://homepage.ntlworld.com/jfe1205/oggdec101.zip (http://homepage.ntlworld.com/jfe1205/oggdec101.zip)
oggdropXPd V1.7.2: http://homepage.ntlworld.com/jfe1205/oggdr...dV1.7.2-101.zip (http://homepage.ntlworld.com/jfe1205/oggdropXPdV1.7.2-101.zip)
oggenc 2.3: http://homepage.ntlworld.com/jfe1205/oggenc2.3-101.zip (http://homepage.ntlworld.com/jfe1205/oggenc2.3-101.zip)
oggvorbis dlls: http://homepage.ntlworld.com/jfe1205/oggvo...bis-dlls101.zip (http://homepage.ntlworld.com/jfe1205/oggvorbis-dlls101.zip)
libVorbis dll: http://homepage.ntlworld.com/jfe1205/libVorbis101.zip (http://homepage.ntlworld.com/jfe1205/libVorbis101.zip)
Old vorbis dll: http://homepage.ntlworld.com/jfe1205/vorbis-dll101.zip (http://homepage.ntlworld.com/jfe1205/vorbis-dll101.zip)
VorbisGain 0.33: http://homepage.ntlworld.com/jfe1205/vorbi...in-0.33-101.zip (http://homepage.ntlworld.com/jfe1205/vorbisgain-0.33-101.zip)

Title: OggVorbis updates
Post by: Atlantis on 2003-09-10 15:02:48
Thanks John 
Title: OggVorbis updates
Post by: Peter Harris on 2003-09-11 03:45:35
Quote
Code: [Select]
ERROR: Cannot open input file "Chariots.wav": Bad file descriptor

Bad file descriptor? Of all the possible errors, that's one I wasn't expecting.

Fortunately (or unfortunately, depending on your perspective), I don't have a Win 9x box anywhere to test on, so I can't fix the problem myself. [I believed Microsoft when they said that 95 was to be the only transitional Windows between 3.1 and NT  , so I never had 98 or Me].

It should be fairly straightforward for somebody else to fix, though.
Title: OggVorbis updates
Post by: rjamorim on 2003-09-11 04:49:43
Quote
Fortunately (or unfortunately, depending on your perspective), I don't have a Win 9x box anywhere to test on, so I can't fix the problem myself. [I believed Microsoft when they said that 95 was to be the only transitional Windows between 3.1 and NT  , so I never had 98 or Me].

I don't have one either. Use an OS virtualizer like Connectix VirtualPC or VMWare. I use VirtualPC here, and that's a wonderful software for testing programs on Win9x or running programs that won't run on 2K/XP anymore.
Title: OggVorbis updates
Post by: LCtheDJ on 2003-09-11 16:57:03
@ john33,

I used oggenc 2.3 from http://homepage.ntlworld.com/jfe1205/oggenc2.3-101.zip (http://homepage.ntlworld.com/jfe1205/oggenc2.3-101.zip) with Speek's multi frontend and MAC.exe that is modified for stdout.

Oggenc2.3-101 won't encode from an ape file; neither will the previous oggenc23. Encoding from .ape works fine with oggenc22, so it's not the MAC.exe file.

When using Oggenc2.3-101, I get only this in the cmd box: "Opening with wav module: WAV file reader" then I get a popup error message: "oggenc2.3-101.e.exe (sic) has generated errors and will be closed by Windows. You will need to restart the program. An error log is being created. OK". After closing the popup, the cmd box adds "Error: 1001"

The Dr. Watson log shows "Exception number: c0000005 (access violation)",

Running Windows 2000 Pro SP4 on a 1.6 GHz P4.

(edited for spelling)
Title: OggVorbis updates
Post by: john33 on 2003-09-12 08:24:55
OK, I'll look into that. Using MultiFrontend, I get the same error here. (WinXP Pro SP 1)

However, you do realise that you don't need to use MultiFrontend for this? Providing MAC.exe is somewhere in the path, you can input ape files directly as oggenc2.3 will invoke the decoder in the background and the decoded samples are then piped directly into the encoder.
Title: OggVorbis updates
Post by: Peter Harris on 2003-09-12 16:07:52
Quote
I don't have one either. Use an OS virtualizer like Connectix VirtualPC or VMWare. I use VirtualPC here, and that's a wonderful software for testing programs on Win9x or running programs that won't run on 2K/XP anymore.

Are you offering to purchase for me a copy of VMWare, a copy of Windows Me, and another stick of RAM?

I didn't think so. For me, that's a lot of money to put out just to support (as a hobby, remember) a few people who are using an obsolete OS.

This is the point where somebody (you perhaps?) with 9x jumps in to offer to update the my patch to at least degrade gracefully on systems that don't support UNICODE properly. I'm happy to answer any questions you may have on the code.
Title: OggVorbis updates
Post by: idem on 2003-09-19 14:15:17
Quote
The following (1.0.1?) compiles are now available:

My oggdec: http://homepage.ntlworld.com/jfe1205/oggdec101.zip (http://homepage.ntlworld.com/jfe1205/oggdec101.zip)

I decoded file (encoded with OggEncGT3b1 -q6) with this decoder (--dither 0) and decoder from http://www.vorbis.com/files/1.0/windows/vo...s-1.0-win32.zip (http://www.vorbis.com/files/1.0/windows/vorbis-tools-1.0-win32.zip) (no options) and get different output - wav files same in size but not in content.
For me both wav's sound same, but I why files content different and whitch decoder best to use?

Sorry: it's only lsb difference, so it's only different rounding modes.