CELT 0.9.1 is out!, Help wanted Options

[LaserSokrates]

Hey jmvalin, thank you and the team behind CELT for such an amazing codec. I only use it for speech (mumble) and the quality is amazing, as well as the latency.

[Ljubo44]

In any case, since people asked... I create a win32 build of the 0.11.0 encoder and decoder tools: https://people.xiph.org/~greg/celt/celt011-win32.zip

Link not working. Can someone upload celt encoder .. anyplace. Please.

[Anakunda]

Hello there, tried to compile celtenc with library and no conversion is performed, anyone can say why??

Here's the binaries that don't work for me:

CODE

http://www.mediafire.com/?lioj5w9hi78vd26

Called with

celtenc test.wav test.oga

and nothing....

[2304p]

Hello there, tried to compile celtenc with library and no conversion is performed, anyone can say why??

Here's the binaries that don't work for me:

CODE

http://www.mediafire.com/?lioj5w9hi78vd26

Called with

celtenc test.wav test.oga

and nothing....

try this:

CODE

http://audiocoding.ru/celt.html

the current version is celt 0.11.2

[Anakunda]

Thanks much, that's work. I'd be interested how's CELT performing as for quality in contest with Vorbis@Q6, Lame@V2 and NeroAAC@Q0.55.
Would anyone consider co-ordinate a listening test at these bitrates

the current version is celt 0.11.2

Where did you find that version? Official site still reports 0.11.1: http://www.celt-codec.org/downloads/

This post has been edited by Anakunda: Jun 3 2011, 07:30

[klonuo]

You can get 0.11.2 also from here: http://listening-tests.hydrogenaudio.org/igorc/index.htm and see test results at 64 Kb/s

Do you have any idea what it takes to make listening-test?

[lvqcl]

version 0.11.4 available at http://git.xiph.org/?p=celt.git

[klonuo]

version 0.11.4 available at http://git.xiph.org/?p=celt.git

So, where is 'configure'? - Process it with autoconf
OK, but it throws errors. - So you are stucked with Windows?
No, I'm just sick of packages that can't compile on MSYS or Cygwin on first try. *sigh*

...dream on for MSVC solutions...

[Anakunda]

celt.c violates the C standard several times in the
ALLOC_STACK;
SAVE_STACK;
constructions (variable definition after statement), my compiler complains about that

could you please fix it

version 0.11.4 available at http://git.xiph.org/?p=celt.git

please for link to tar.gz archive, I don't have git installed

[lvqcl]

celt.c violates the C standard several times in the
ALLOC_STACK;
SAVE_STACK;
constructions (variable definition after statement), my compiler complains about that

Really? At least msvs2010 compiles it without problems.

please for link to tar.gz archive, I don't have git installed

It's there. Click on the first "snapshot" link.

[Anakunda]

celt.c violates the C standard several times in the
ALLOC_STACK;
SAVE_STACK;
constructions (variable definition after statement), my compiler complains about that

Really? At least msvs2010 compiles it without problems.

I think VS2010 uses less strict lexical rules, but perhaps it would be better to stick with the standard strictly so that the code is truely portable. I'm using ICL.

please for link to tar.gz archive, I don't have git installed

It's there. Click on the first "snapshot" link.

Thanks

[2304p]

version 0.11.4 available at http://git.xiph.org/?p=celt.git

Thanks!
I have downloaded this snapshot, how to compile celt-0.11.4.tar.gz to exe and dlls (32bit)?

[lvqcl]

Either I cannot compile CELT 0.11.4 correctly, or it is incompatible with current foo_input_celt.

[klonuo]

I have downloaded this snapshot, how to compile celt-0.11.4.tar.gz to exe and dlls (32bit)?

Assuming you are on Windows with MSVC < 10:

1. open MSVS and make empty project for libcelt from files from libcelt folder and build it
2. download libogg and build it
3. make separate projects for celtenc and celtdec* with linker dependant on already build libcelt and libogg libraries
* from files in tools folder (read Makefile.am as a help)

If you setup everything in those steps correctly, you should have celtenc and celtdec executables

[2304p]

I have downloaded this snapshot, how to compile celt-0.11.4.tar.gz to exe and dlls (32bit)?

Assuming you are on Windows with MSVC < 10:

1. open MSVS and make empty project for libcelt from files from libcelt folder and build it
2. download libogg and build it
3. make separate projects for celtenc and celtdec* with linker dependant on already build libcelt and libogg libraries
* from files in tools folder (read Makefile.am as a help)

If you setup everything in those steps correctly, you should have celtenc and celtdec executables

i use minGW32

[klonuo]

run automation bash script provided and then as readme says just compile with ./configure; make

or forget about it

[2304p]

can you compile celt-0.11.4.tar.gz and upload as celt-0.11.4-win32.zip?

google doesn't help me further.

[Anakunda]

I can compile but for me it doesnot encode anything and exits.

[lvqcl]

Celt 2011-04-21: [attachment=6643:celt_2011_04_22.7z]

(re-uploaded: compiled with CUSTOM_MODES=1. Source and *.exe included)

This post has been edited by lvqcl: Aug 11 2011, 19:45

[2304p]

Celt 2011-04-21: [attachment=6540:celt_201...21_win32.7z]

thank you

[darkbyte]

Hi!

First, thanks for this truly amazing codec. I'm excited to see where this codec advances in the future

I'm curious if it's possible to implement a similar technology like bitrate peeling in Vorbis which was actually never used there. It would be so nice to use CELT for eg. web radio streaming where the client could decide according to it's current connection speed the bitrate it requests and switch between them gaplessly. Say, the main encoder encodes the audio @192kbps, but it can be peeled down to 96kbps or 48kbps without transcoding on the audio server by sending less data. When you listen to web radio on eg. your mobile phone while moving in your car you roam between cells but not all of them supports high speed connection for example, on EDGE the web radio player on the phone can revert to 48kbps so it can keep up with continous streaming. What do you think? As i read CELT scales very well so maybe it's not impossible to arrange encoded data in a peelable way.

This post has been edited by darkbyte: Nov 13 2011, 10:02

[NullC]

I'm curious if it's possible to implement a similar technology like bitrate peeling in Vorbis which was actually never used there. It would be so nice to use CELT for eg. web radio streaming where the client could decide according to it's current connection speed the bitrate it requests and switch between them gaplessly. Say, the main encoder encodes the audio @192kbps, but it can be peeled down to 96kbps or 48kbps without transcoding on the audio server by sending less data. When you listen to web radio on eg. your mobile phone while moving in your car you roam between cells but not all of them supports high speed connection for example, on EDGE the web radio player on the phone can revert to 48kbps so it can keep up with continous streaming. What do you think? As i read CELT scales very well so maybe it's not impossible to arrange encoded data in a peelable way.

For files encoded with Opus using the hybrid mode, they can be stripped to just the low frequency parts inexpensively. Otherwise, no. Part of the efficiency of the format comes from making the coding so right that 'pealing' is not possible.

On the other hand, our encoder is fairly fast— the fairly unoptimized reference code (No ASM, No SIMD) is >12x faster than libvorbis' managed mode encoding. You should also be able to switch clients between different bitrate sources on a packet by packet basis without breaking continuous streaming.

[darkbyte]

For files encoded with Opus using the hybrid mode, they can be stripped to just the low frequency parts inexpensively. Otherwise, no. Part of the efficiency of the format comes from making the coding so right that 'pealing' is not possible.

I see. And how about the other end? Making CELT scalable to lossless? That would very nice also. 96kbps for portable listening is already nice, but what about storing correction data? Like HD-AAC does.

This post has been edited by darkbyte: Nov 14 2011, 17:59

[Gainless]

Can someone re-compile the celt version 0.11.4 to an exe file?
I didn't really get through how to do it by myself...