[Ffmpeg-devel-irc] ffmpeg.log.20121108

[burek]

[00:00] <shroomM> lol, nice!
[00:00] <shroomM> i think that might just be it
[00:00] <shroomM> will try
[00:00] <shroomM> thanks!
[00:20] <shroomM> so my source file is like this ...
[00:20] <shroomM> Duration: 00:42:59.20, start: 0.000000
[00:21] <shroomM> then I encode like so...
[00:21] <shroomM> ffmpeg -i input.mxf -pix_fmt yuv420p -vcodec libx264 -preset veryfast -bf 0 -vf scale=480:272 -ac 2 -c:a libfdk_aac -b:a 96k -shortest test_shortest.mp4
[00:22] <shroomM> and running ffmpeg -i test_shortest.mp4 yields this..
[00:22] <shroomM>  Duration: 00:42:59.24, start: 0.042667
[00:22] <shroomM> any ideas why and how to fix this ?
[00:28] <burek> everything is fine with that
[00:30] <shroomM> why doesn't it start at 0 and have the same lenght as the original?
[00:30] <shroomM> length*
[00:34] <JEEBsv> shroomM: probably to accomodate for the aac encoder delay
[00:34] <JEEBsv> all encoders have that, a variable amount of it
[00:34] <JEEBsv> in the end the decoded result is the same
[00:34] <JEEBsv> you should be happy ffmpeg is at least setting it :D
[00:34] <shroomM> hehe
[00:35] <JEEBsv> (as in, noting in the container that there is an encoder delay9
[00:35] <JEEBsv> because the stream doesn't and you're left to play the guessing game
[00:35] <shroomM> what is the encoder delay anyhow?
[00:35] <shroomM> I mean ...
[00:36] <shroomM> the samples go in, the samples go out, why is there a delay
[00:36] <shroomM> and why does it influence the output
[00:36] <JEEBsv> comes from the dct-baseness of audio formats, empty samples in the beginning
[00:36] <JEEBsv> AAC doesn't note in the bit stream how many samples of delay there will be
[00:37] <JEEBsv> (at least I /think/ the leading samples/encoder delay come from the DCT-baseness with AAC)
[00:37] <shroomM> aha
[00:37] <shroomM> ok, np
[00:37] <shroomM> explains why everyhing is the same if I encode just the audio
[00:37] <JEEBsv> in any case, it is there in any encoder, and the actual bit stream doesn't contain it
[00:38] <shroomM> but openning the resulting mp4 in premiere, I believe the number of frames it shows is different
[00:38] <shroomM> different from the source, that is
[00:38] <shroomM> like, 2 frames more than the original
[00:39] <JEEBsv> could be that it's adding them instead of just cutting the audio?
[00:39] <JEEBsv> dunno
[00:39] <shroomM> I actually tried opening it in virtualdub with this plugin
[00:39] <shroomM> http://code.google.com/p/ffinputdriver/downloads/list
[00:39] <shroomM> and got the same issue
[00:39] <shroomM> yeah, I guess that could be it
[00:40] <JEEBsv> you should make sure it is compiled with a new enough libavformat/codec
[00:40] <JEEBsv> also no idea how those things handle that kind of info
[00:40] <JEEBsv> all I know is that ffmpeg now at least sets the encoder delay into the output mp4 when possible
[00:41] <shroomM> I see
[00:41] <shroomM> ok
[00:41] <shroomM> any idea since when?
[00:41] <shroomM> approx
[00:43] <JEEBsv> relatively lately, also the ffmpeg's decoding at one point just assumed random encoder delay for AAC in MOV/MP4/etc.
[00:44] <shroomM> lol
[00:44] <JEEBsv> (before that it just decoded all samples that were in the AAC stream, and afterwards it was fixed to be all samples by default, and if there was a faac header found then faac encoder delay was used -- and support for reading it from the container was added)
[00:45] <shroomM> alright
[00:45] <shroomM> i then need a good mp4 video editor, heh
[00:45] <JEEBsv> aviutl + L-SMASH Works I guess? Not fully sure but I would guess that would be relatively OK, given the amount of love MP4_maniac has put into it
[00:46] <shroomM> aviutl?
[00:46] <shroomM> http://www.divx-digest.com/software/aviutl.html ?
[00:46] <shroomM> ah, no, prolly this ... http://spring-fragrance.mints.ne.jp/aviutl/
[00:46] <JEEBsv> yes, but you sure want to grab the newest version from its official site
[00:47] <JEEBsv> and L-SMASH Works then is a test plugin for AviUtl that uses the L-SMASH demuxer
[00:47] <JEEBsv> the problem is that most documentation is in Japanese lol
[00:48] <shroomM> yeah, so I see :)
[00:48] <shroomM> i like vritualdub
[00:48] <shroomM> but getting mp4s in there
[00:48] <JEEBsv> well it's a VFW app :P
[00:48] <shroomM> yeah ... ><
[00:49] <JEEBsv> the next thing could be something based on vapoursynth, but getting timestamps out of there currently means you have to go through all frames' durations
[00:49] <JEEBsv> maybe in a year or two
[00:50] <shroomM> VS is the next AVS ?
[00:50] <JEEBsv> ye
[00:50] <shroomM> w/ python
[00:50] <shroomM> i saw
[00:50] <shroomM> how far along is it
[00:50] <JEEBsv> yeah, the official bindings are for python
[00:50] <JEEBsv> it's getting along pretty well
[00:51] <JEEBsv> we have a reader for mpeg-ts/ps native, we have win32 avs plugin support, there are inverse telecine filters already ported...
[00:51] <shroomM> nice
[00:51] <shroomM> should be useful
[00:51] <shroomM> sometime in the future
[00:51] <JEEBsv> Firesledge should get his resizer package ready soon'ish
[00:51] <JEEBsv> (vs by default uses libswscale resizing)
[00:52] <shroomM> aha
[00:52] <shroomM> nice
[00:52] <shroomM> sorry for jumping the topics ... the ffinputdriver for virtualdub was released jan2012, ffmpeg 0.9.1..
[00:52] <shroomM> i have no idea what to search for
[00:53] <shroomM> to see if this is a recent enough version :S
[00:53] <shroomM> of ffmpeg
[00:53] <JEEBsv> let's just say it isn't
[00:53] <JEEBsv> you want something a month or so old at max
[00:53] <shroomM> ah
[00:53] <shroomM> ok
[00:54] <JEEBsv> could be this... not sure http://git.videolan.org/?p=ffmpeg.git;a=commit;h=14fd34d73bb0bcb227b62506fb51d63751c9b190
[00:54] <shroomM> doesn't really matter ... just needed to know whether this was added "sometime this year" or "in the last month" :)
[00:54] <shroomM> thanks!
[00:55] <JEEBsv> (I know that audio encoders should output negative pts for stuff like encoder delay)
[00:55] <shroomM> ah, i see
[00:55] <shroomM> ok, thank
[00:55] <shroomM> s
[00:55] <shroomM> gotta run, too sleepy
[00:56] <shroomM> bye
[00:56] <shroomM> thanks again!
[00:56] <JEEBsv> np
[05:34] <jbr> Why would a video not play progressively *after* running qt-faststart on it?
[05:41] <Vaevictus> o/ okay... my ffmpeg is complaining because my temp file has no extension, if i use -f it gives a different error, is there a better flag for forcing the output type?
[05:43] <sacarasc> What errors are you getting, Vaevictus?
[05:44] <Vaevictus> sacarasc: lets see: unable to find a suitable output format for '/tmp/mt_trascode_JVYFNW'
[05:47] <sacarasc> And when you use -f?
[05:48] <Vaevictus> -f mp4 gives me: muxer does not support non seekable output]
[05:48] <Vaevictus> and
[05:48] <Vaevictus> Could not write header for output file #0
[05:49] <Vaevictus> but if i just change the input to .mp4, it works perfectly. sadly, i'm not in control of the tmp file name for the process i'm working on
[05:49] <Vaevictus> oh hmm
[05:49] <Vaevictus> nm
[05:49] <Vaevictus> shit
[05:49] <Vaevictus> i'm writing to a pipe
[05:50] <Vaevictus> apparently i can't do that
[05:50] <Vaevictus> even with -y
[05:50] <sacarasc> Nope, you can't use mp4 with a pipe.
[05:50] <Vaevictus> so i guess that's a non-bug and a no-go
[05:50] <sacarasc> You could use -f mkv, maybe?
[05:51] <Vaevictus> i don't know whether my android phone and/or ps3 could play matroskas
[05:53] <sacarasc> MP4 isn't very good for streaming, if that's what you're doing. You have to have the whole file complete before you can write the index usually. Unless you write lots of microfiles.
[05:53] <Vaevictus> what's a good option then?
[05:53] <Vaevictus> mpeg2
[05:54] <jbr> got any thoughts on my issue sacarasc? :)
[05:54] <sacarasc> Nope, sorry.
[05:57] <Vaevictus> sacarasc: got a handy ffmpeg line for mpeg2?
[05:59] <sacarasc> I've not encoded MPEG2 for a long time, so I don't know how to make it look good without throwing a LOT of bitrate at it.
[05:59] <Vaevictus> got a better option that's compatible with ps3 / my phone ? :D
[06:00] <sacarasc> I don't know what they can take.
[06:01] <kline> trying to stream a video from linux, but vlc is throwing errors it says are because of ffmpeg. error is here: http://pastebin.com/nSVZ30Ku , and ffmpeg details are here: http://pastebin.com/8FMzrv9L
[06:01] <Vaevictus> mp4, dunno what else
[06:01] <kline> distro is debian 6 - anyone have tips?
[06:02] <sacarasc> kline: "If you don't know how to fix this, ask for support from your distribution."
[06:03] <kline> yep, but asking here first because A) im bad at reading, and B) you guys probably have more experience than generic debian dudes
[06:03] <kline> ill be asking both debian and vlc channels soon
[06:03] <sacarasc> Don't ask the VLC channel, it specifically says not to.
[06:03] <Vaevictus> kline: no output file specified
[06:03] <kline> theyre the last port of call
[06:03] <kline> Vaevictus: thats just to get the version info
[06:04] <Vaevictus> no, its not
[06:04] <sacarasc> And AFAIK, you'll probably have to compile ffmpeg and VLC yourself, so maybe try #debian first? :D
[06:04] <Vaevictus> or you just did --version?
[06:04] <kline> Vaevictus: user at host:~$ ffmpeg
[06:05] <kline> sacarasc: ok, thanks for your time
[06:05] <Vaevictus> looks like an X related issue
[06:05] <kline> theres no x on this server
[06:06] <Vaevictus> hmm i see now
[06:06] <Vaevictus> your ffmpeg doesnt support mp3 lol
[06:07] <Vaevictus> thats debian fir you
[06:07] <Vaevictus> debian hates proprietary stuff
[06:08] <Vaevictus> definately your deb package
[06:13] <Vaevictus> sacarasc: http://pastebin.com/e6m2eqyp
[06:14] <Vaevictus> which one should i take?
[08:56] <cptmorgan> i built ffmpeg with --enable-libspeex but im getting this error when trying to use it: Unknown encoder 'libspeex'. im running rhel6 and have speex 1.2 installed
[09:13] <cptmorgan> also, for rhel6 should i be using 0.6 or 0.7 for the old api/abi or the latest version?
[09:21] <erib> Hello. I can't cross compile ffmpeg 1.0 to android because of undefined reference to `ff_hqdn3d_row_8_x86`
[09:21] <erib> it's part of libavfilter (vf_hqdn3d)
[09:23] <erib> how can I disable hqdn3d ? there are no compile flags I could find for that
[09:23] <ubitux> --disable-filter=hqdn3d maybe
[09:23] <ubitux> but you should not have this bug in the first place
[09:24] <ubitux> cptmorgan: you likely need the development package
[09:24] <erib> --disable-filter=hqdn3d didn't work. I tried that
[09:25] <ubitux> cptmorgan: libspeex-dev, libxspeex-devel, or sth like that
[09:25] <ubitux> erib: grep HQDN3D config.h
[09:26] <ubitux> try to --disable-filter=mp maybe
[09:26] <ubitux> --disable-filter=hqdn3d,mp if you need both
[09:27] <ubitux> the hqdn3d from mplayer wrapper is removed in recent versions though, so you should not need that
[09:28] <erib> i'm working on the 1.0 tarball from the official downloads since I didn't want to mess with unstable stuff :)
[09:29] <ubitux> it's more likely not less stable
[09:30] <ubitux> we didn't do any bug fix releases since 1.0
[09:30] <ubitux> and thus git master has a lot of pending fixes :p
[09:59] <erib> ubitux: commenting out in config.h didn't help. the --disable-filter didn't work either, and it didn't work even with the latest git pull
[10:00] <erib> is that filter somehow related to ffprobe ?
[10:03] <ubitux> commenting? i didn't told you to comment it
[10:03] <ubitux> i asked to check if it was enabled
[10:05] <erib> yeah sorry. I mean I changed the flag from 1 to 0 in config.h (assuming config.h is not modified again  by some other tool)
[10:06] <erib> anyay you said that hqdn3d is removed in recent versions. how come I can still see this file in libavifilter/vw_hqdn3d.c ?
[10:13] <ubitux> the copy in libmpcodecs is removed
[10:13] <ubitux> not the native one
[10:13] <ubitux> the *old* copy
[10:13] <ubitux> also, "didn't work" doesn't help
[10:14] <ubitux> is it still the same message? did you try make distclean?
[10:15] <erib> yeah still the same message. im almost sure I did distclean. trying again now.
[10:19] <ubitux> pastebin your configure line and the make failure
[10:19] <ubitux> and eventually your configure.log too
[10:28] <erib> the error:http://pastebin.com/r0yeQkaQ the config line: http://pastebin.com/NZan5KC5
[10:57] <eedev> hi
[10:58] <eedev> I'm on Ubuntu 12.10, I'd like to build ffmpeg as a shared library because I'm building it to use it with ffmpeg-php for a website but I'm having trouble, if someone can help ?
[11:12] <eedev> nvm, found out :)
[11:15] <ubitux> erib: ./configure --enable-gpl --disable-filter=hqdn3d ’ #define CONFIG_HQDN3D_FILTER 0 in the config.h
[11:16] <ubitux> can you confirm you have "#define CONFIG_HQDN3D_FILTER 0" after your configure?
[11:18] <ubitux> it's strange that you have HAVE_YASM defined in your case
[11:18] <erib> why ? what's have YASM ?
[11:19] <ubitux> it likely means you are assembling x86 asm
[11:20] <erib> oh, well something is possible wrong with my config line. now I see I don't have other thigs I added
[11:21] <ubitux> it breaks after the first \n without '\'
[11:21] <erib> yep !!!! can't believe it/
[11:21] <ubitux> --enable-yasm  why do you do that?
[11:21] <ubitux> drop this.
[11:21] <ubitux> and you won't need --disable-filter=hqdn3d
[11:22] <ubitux> why all these enable flags?
[11:22] <ubitux> --enable-avcodec, avformat etc is not necessary
[11:22] <ubitux> --disable-shared --enable-static is not necessary either
[11:23] <ubitux> what's the reason for using --enable-runtime-cpudetect? it has almost no effect
[11:23] <ubitux> why the --enable-decoders, encoders, etc? it's the default as well.
[11:23] <erib> I need a binary running on android devices that capture screen from tegra gpu that doesn't give me access to fb0
[11:24] <ubitux> why --enable-ffmpeg and --enable-ffplay? it should be set
[11:24] <ubitux> how is that related to my questions?
[11:24] <erib> so what you see is after long hours of trial and error :)
[11:24] <ubitux> you can likely drop ¾ of your configure flags
[11:25] <ubitux> you can drop --disable-libxvid as well, it won't be autodetected
[11:25] <ubitux> just don't add random flags all over
[11:25] <ubitux> your --enable-yasm is breaking everything
[11:25] <ubitux> and most of the other flags are pointless
[11:26] Action: erib feeling dumb
[11:26] <ubitux> --enable-pic  is this necessary?
[11:27] <erib> this is actually inherited config script from here: https://github.com/guardianproject/android-ffmpeg
[11:27] <erib> here it is : https://github.com/guardianproject/android-ffmpeg/blob/master/configure_ffmpeg.sh
[11:28] <ubitux> it's looks stupid
[11:28] <ubitux> http://pastebin.com/dmTMfVK8
[11:28] <ubitux> i think you just need something like this.
[11:29] <ubitux> can be even more simplified.
[11:29] <ubitux> maybe this can do the trick: http://pastebin.com/mRNLaSZb
[11:30] <ubitux> --enable-pthreads should be autodetected as well
[11:30] <erib> I need to read from the gpu (not using fb0, using a pipe that I feed with png screenshots) and stream to rstp or multicast with mpegts
[11:31] <ubitux> i'm just dropping what is automatically enabled
[11:31] <erib> thus I had to try with image2pipe and the frame rate was very problematic
[11:34] <jonwil> can anyone tell me where the samples archive lives these days?
[11:34] <jonwil> google isnt showing useful results
[11:35] <ubitux> samples.ffmpeg.org ?
[11:35] <jonwil> ok, thanks
[11:44] <jonwil> ok, now I just gotta find the right bits I need to install in Cygwin to get FFMPEG compiling on my Windows box
[11:46] <jonwil> looks like the FFMPEG configure script will complain if I dont have what I need :)
[11:52] <jonwil> or not :(
[12:07] <jonwil> damn, looks like ffmpeg doesn't have what I need (encoder for ea_xas_adpcm and encoder for ealayer3)
[13:09] <jc5134> Hello. Is it possible to reencode a video using h264_vdpau as input decoder? When I try, I get: decoding to PIX_FMT_NONE is not supported
[13:10] <jc5134> I tried forcing output pixel format using -vf "format=yuvj420p" and -pix_fmt yuv420p to no avail
[13:10] <jc5134> any pointers or information would be much appreciated
[13:13] <jc5134> nevermind, actually -codec:v h264_vdpau does the trick. Now I get a segfault instead of the aformentioned error. Probably because I don't convert the pixel format afterwards
[13:13] <jc5134> sorry -pix_fmt vdpau_h264
[16:34] <gxk> need suggestion where to find ffmpeg's source code for streaming from TS file over UDP?
[18:35] <cptmorgan> ubitux: i sucesfully built ffmpeg. i get that encoder error when i try to use it. I built 0.7 version and it worked fine. 0.6 doesnt. i thought that 0.6 had to be used with el6 and was still being developed because of the old abi
[18:36] <ubitux> we are in 1.0 you know?
[18:37] <ubitux> there is a maintainer for 0.10, 0.11 and 1.0
[18:38] <ubitux> but no one bother about version as old as 0.5/0.6/..
[18:38] <mudkipz> what are 0.1 and 0.11?
[18:38] <ubitux> ?
[18:39] <JEEBsv> earlier releases?
[18:39] <ubitux> 0.5 ’ 0.6 ’ 0.7 ’ 0.8 ’ 0.9 ’ 0.10 ’ 0.11 ’ 1.0
[18:39] <mudkipz> oh right, of course.
[18:39] <mudkipz> just that for a second I was thinking it was like gstreamer who have two separate versions.
[18:39] <ubitux> 0.1 ` 0.10
[18:40] <ubitux> this is versionning, not math
[18:40] <mudkipz> yeaa, my mistake
[18:40] <mudkipz> I know that..usually :\
[18:40] Action: jonwil wonders how hard it would be to make ffmpeg support EALayer3 audio...
[18:41] <JEEBsv> jonwil: do you know how the format works? if yes, welcome to implementing it :D
[18:42] <jonwil> its a derivitive of MP3 and there is code out there that can decode it. I dont know enough about MPEG audio or about the FFMPEG MPEG audio decoders to write code myself :P
[18:42] <JEEBsv> well, if there's a decoder it's better than having no idea about it whatsoever :)
[18:42] <JEEBsv> too bad it's not lossless so it's not the simplest thing unless it's mostly a MP3 copy
[18:43] <JEEBsv> could check the decoder's sources for possible info
[18:43] <JEEBsv> although I really wonder why you need to encode such a format :)
[18:52] <jonwil> The reason its worth en/decoding this format is that a number of EA game titles take audio in this format, i.e. its useful for extracting music from these games and also for adding new music to these games
[18:54] <JEEBsv> well, I generally completely agree with decoding such formats -- the encoding was kind of something I didn't fully understand. But I guess if you're making mods it makes sense?
[18:54] <jonwil> yes
[18:57] <jonwil> In this case EA is unable to release their own tools for any of the titles that use EALayer3 due to patent issues
[18:57] <jonwil> but FFMPEG doesn't need to care about the MPEG patents as much as EA does :P
[19:38] <xroberx> hi there
[19:40] <xroberx> I've got a problem: I can't read image sequences with the image2 demuxer on Android. I've compiled the same version of ffmpeg on both linux and Android, and compiled the same test code that simply tries to load an image sequences with filename "m%04d.jpg" and it works fine on Linux, but on Android I get ENOENT (-2) by avformat_open_input(). I've checked everyting, I ran the debugger, double checked the permissions of the directories,
[19:41] <xroberx> ENOENT (-2) is "No such file or directory"
[19:42] <xroberx> If I leav the filename out and just pass the directory I get "Is a Directory", which confirms that ffmpeg can actually can read the directory filename
[19:44] <xroberx> forgot to say that both version of ffmpeg are compiled with the same options (of course one for ARM[Android] and one for x86 [Linux] but that is the only difference). I've tried ffmpeg 1.0 and latest git
[20:59] <spaam> sooo
[20:59] <spaam> do ffmpeg have the -c option like avconv?
[21:00] <spaam> like -c copy ?
[21:01] <sacarasc> For codec?
[21:01] <spaam> yes
[21:02] <spaam> in avconv i can just type avconv -i super.ts -c copy file.mp4 to mux it from .ts to .mp4
[21:02] <sacarasc> Yes, I believe so.
[21:02] <spaam> good
[21:02] <sacarasc> I think.
[21:02] <sacarasc> Maybe.
[21:02] <sacarasc> I've not used ffmpeg in ages. :D
[21:03] <spaam> me either.. but there a lot of ppl doing it.....
[21:08] <sacarasc> I've not encoded any videos for a long time.
[21:15] <miggyb> Hi room. Having difficulty getting ffmpeg compiled with --enable-nonfree on OpenBSD 5.2 on an AMD64 system. First it was complaining about libgsm being missing, but I changed the config file so it would look at /usr/local/include/gsm.h instead of gsm/gsm.h and that seems to be fixed now
[21:16] <miggyb> now it compiles but gives the following errors:
[21:16] <miggyb> /tmp//cc2U7u2h.s:15695: Error: `-1(%r10d)' is not a valid 64 bit base/index expression
[21:16] <miggyb> when I try to 'gmake install' it
[21:17] <miggyb> Googling says that it might be an issue with gcc or as, that I have the wrong version or something along those lines
[21:18] <miggyb> should it compile fine with gcc (GCC) 4.2.1 20070719 and GNU assembler 2.15?
[21:27] <llogan> spaam: yes, it uses "-c copy"
[21:27] <llogan> might want to add "-map 0" to that
[21:28] <spaam> why?
[21:28] <llogan> so it copies all streams from input 0
[21:28] <llogan> i'd guess it would apply to the fork stuff too
[21:45] <siganderson> does a way to overlap a jpg to a video for all its length exist?
[22:12] <Takyoji[laptop]> Any recommended streaming protocol and configuration for streaming Opus or Speex?
[22:12] <Takyoji[laptop]> Compatibility isn't a concern; most folks will very likely be receiving with VLC, etc
[22:13] <Takyoji[laptop]> Because I'm having trouble using Speex, the error is '[rtp @ 0x10edfe0] Unsupported codec'
[22:13] <Takyoji[laptop]> unsupported codec 15026*
[22:14] <Takyoji[laptop]> Command line arguments are: avconv -ac 1 -ar 32k -f alsa -i plughw:1,0 -acodec libspeex -re -f rtp rtp://localhost:14313
[22:15] <Takyoji[laptop]> I'm assuming the case is that RTP doesn't have a way to describe the stream is Speex, correct?
[22:15] <klaxa> do you really need -re?
[22:16] <Takyoji[laptop]> Perhaps not
[22:16] <Takyoji[laptop]> Anyway; any recommendations?
[22:17] <Takyoji[laptop]> Encoding needs to be fairly light on the processor; it's being encoded on ARMv7 at 700MHz
[22:18] <Takyoji[laptop]> The input is a single-side band radio stream, which is very narrow in bandwidth; so quality isn't a big concern.
[22:18] <klaxa> sounds like raspberry pi, you could try using plain udp instead of rtp
[22:19] <Takyoji[laptop]> It is, yes
[22:19] <klaxa> or http even... which would... hmm... you would need ffserver to redistribute that, because http is unicast
[22:20] <Takyoji[laptop]> There's only going to be probably 5 connections at most
[22:21] <Takyoji[laptop]> As what would be the relevant command line arguments for plain UDP?
[22:22] <klaxa> remove -f rtp and replace rtp://localhost by udp://localhost
[22:22] <klaxa> at least that's what i recall to have done at some point in the past
[22:22] <llogan> siganderson: see the overlay filter
[22:22] <klaxa> i'm afraid it won't work because it's missing headers or something when entering midstream, although i'm not sure how roboust speex is against that kind of data loss
[22:22] <Takyoji[laptop]> 'unable to find a suitable output format for 'udp://localhost'
[22:22] <klaxa> hmm...
[22:23] <Takyoji[laptop]> Well, it can be any codec really
[22:23] <klaxa> well... try vorbis then :P
[22:23] <Takyoji[laptop]> Same complaint
[22:24] <Takyoji[laptop]> I'm looking at the StreamingGuide and see: -f mpegts udp://10.1.0.102:1234
[22:26] <klaxa> if i use libmp3lame and -f rtp rtp://localhost2
[22:26] <klaxa> *rtp://localhost:1234
[22:26] <klaxa> i can listen to it when pointing ffplay at rtp://localhost1234
[22:27] <klaxa> vlc can play it too, for some reason mplayer2 fails
[22:27] <klaxa> mplayer2 has been failing for me a lot recently...
[22:29] <siganderson> llogan, ok I'm going to try
[22:34] <klaxa> Takyoji[laptop]: so the only things that worked and i have tested so far are either mp3 (libmp3lame) or vorbis (libvorbis) with an .sdp file (never used them though, haven't tested it just now either)
[22:38] <Takyoji[laptop]> For some reason I can't connect to it.
[22:38] <Takyoji[laptop]> It is properly encoding and 'streaming'
[22:38] <Takyoji[laptop]> But when I open rtp://192.168.0.10:1234 in VLC, it gives an error
[22:40] <klaxa> what codec did you try?
[22:42] <Takyoji[laptop]> libmp3lame
[22:42] <Takyoji[laptop]> The error is regarding connecting
[23:02] <darkstarbyte_> Mavrik, I tried what you said to try with strace -c
[23:02] <darkstarbyte_> and through the 4 minutes of encoding one call to a total of 6 seconds
[00:00] --- Fri Nov  9 2012