From moritz at bunkus.org Fri Apr 1 09:27:33 2011 From: moritz at bunkus.org (Moritz Bunkus) Date: Fri, 1 Apr 2011 09:27:33 +0200 Subject: [Matroska-devel] weba audio format In-Reply-To: References: <201103311652.05410.mconstable@gmail.com> Message-ID: <201104010927.33475.moritz@bunkus.org> Hey, On Thursday 31 March 2011 22:20:14 Steve Lhomme wrote: > That could be investigated, because right now one of the big problem > in WebM is the size of the Vorbis init packet. It would be very to add > an extension to Matroska to have it compressed. In fact being a codec > private it would just take calling the codec "A_VORBIS2" and treating > the codec private as zlib compressed rather than raw. Adding support > in all OSS players will be very fast and easy. Matroska (unlike WebM) supports compressing codecprivate already via the ContentEncoding element and its children (with (ContentEncodingScope & 2) == 2). It would make much more sense to use ContentEncoding with WebM as well (support for basic zlib compressed subtitles is already present in many hardware players) than to invent arbitrary new codecIDs just for this minor difference. Not that compressing would actually gain all that much as this discussion showed. Regards, Mosu -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part. URL: From stebbins at jetheaddev.com Fri Apr 1 21:00:27 2011 From: stebbins at jetheaddev.com (John Stebbins) Date: Fri, 01 Apr 2011 12:00:27 -0700 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster Message-ID: I've noticed that mkvalidator flags a warning for clusters that do not start with a video keyframe and that mkclean --remux will "fix" this (and error out if the video has no keyframes). I've been searching through the specification at http://www.matroska.org/technical/specs/index.html and can not find where this requirement is documented. Can someone please point me to where this is documented? The requirement has a some "interesting" side effects. First, it limits the max keyframe interval that an application can use to < ~65 seconds (assuming default timescale). A corollary to this is that matroska does not support progressive intra refresh video (video that has no keyframes). Is this intended, or is there some loophole that would allow this? Again, pointers to where this is specified would be useful. Thanks John From moritz at bunkus.org Fri Apr 1 21:39:51 2011 From: moritz at bunkus.org (Moritz Bunkus) Date: Fri, 01 Apr 2011 20:39:51 +0100 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: References: Message-ID: <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org> Hey, mkvalidator emits only a warning. It's not a spec violaton, and mkvalidator assumes that you have normal video with more or less regular intervals between key frames. The specs definitely do not require a cluster to start with a key frame. It's optimal if they do so that less seeking is required. But that's why it's only a warning. I have argued with Steve Lhomme (mkvalidator's author) that people will take warnings by a "validation tool" very serious, and that this is not worth a warning but maybe only a recommendation (or whatever such a info level might be called). But that's my personal opinion. Regards, mosu From computacao.rafael at gmail.com Fri Apr 1 22:03:59 2011 From: computacao.rafael at gmail.com (Rafael) Date: Fri, 1 Apr 2011 17:03:59 -0300 Subject: [Matroska-devel] Sony XDCAM EX1 format Message-ID: Hi there, I was looking for a good MP4 splitter when I found the Haali Media Splitter, and started doing some tests with it. However, one of the formats I'm working with, wich source is the Sony XDCAM EX1 camera (now one of the hotshots of the TV industry), didn't work. Actually, the splitter didn't find the audio stream. I've tested many samples from many different sources using this camera and all had the same results: a video stream (Mpeg2Video) and no audio stream, when there should be a PCM there. Thanks for you attention, Rafael. -------------- next part -------------- An HTML attachment was scrubbed... URL: From stebbins at jetheaddev.com Fri Apr 1 22:15:14 2011 From: stebbins at jetheaddev.com (John Stebbins) Date: Fri, 01 Apr 2011 13:15:14 -0700 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org> References: <4D9620CB.3060409@jetheaddev.com> <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org> Message-ID: On 04/01/2011 12:39 PM, Moritz Bunkus wrote: > Hey, > > mkvalidator emits only a warning. It's not a spec violaton, and mkvalidator assumes that you have normal video with > more or less regular intervals between key frames. The specs definitely do not require a cluster to start with a key > frame. It's optimal if they do so that less seeking is required. But that's why it's only a warning. > > I have argued with Steve Lhomme (mkvalidator's author) that people will take warnings by a "validation tool" very > serious, and that this is not worth a warning but maybe only a recommendation (or whatever such a info level might be > called). But that's my personal opinion. > > Regards, > mosu > Thanks for the clarification. FYI, it wasn't mkvalidator that prompted me to take this more seriously. We have a user that is claiming that the twonky dlna server "restarts" on streams that have clusters that do not begin with a keyframe. I adjusted our code to place keyframes at the beginning of clusters and had the user retest. His problem went away. So it appears that other application developers are taking this to be a requirement and are creating broken demuxers. Also, if it is just a warning, mkclean shouldn't error out on streams that have no keyframes. John From moritz at bunkus.org Fri Apr 1 22:57:32 2011 From: moritz at bunkus.org (Moritz Bunkus) Date: Fri, 01 Apr 2011 21:57:32 +0100 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: References: <4D9620CB.3060409@jetheaddev.com> <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org> Message-ID: <85d114c13ee2154a6dcdc1cead7a24b7@www.bunkus.org> Hey, On Fri, 01 Apr 2011 13:15:14 -0700, John Stebbins wrote: > We have a user that is > claiming that the twonky dlna server "restarts" on streams that have > clusters that do not begin with a keyframe. That's definitely a bug in twonky. There are situations in which it's imply impossible to start clusters with a key frame. Each cluster can contain only so many packets (e.g. due to timecode scale and the relative timecode field in the packets being a signed 16bit integer). > So it appears that other application > developers are taking this to be a requirement and are creating > broken > demuxers. Unfortunately demuxer writers have often simply used the common cases (e.g. files produced by mkvmerge) to be the only target they test against resulting in various bugs/missing features that the specs allow for but that the files they're testing with didn't contain. > Also, if it is just a warning, mkclean shouldn't error out on streams > that have no keyframes. That's a bug in mkclean then. Steve? Regards, mosu From slhomme at matroska.org Sat Apr 2 13:31:02 2011 From: slhomme at matroska.org (Steve Lhomme) Date: Sat, 2 Apr 2011 13:31:02 +0200 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org> References: <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org> Message-ID: On Fri, Apr 1, 2011 at 9:39 PM, Moritz Bunkus wrote: > Hey, > > mkvalidator emits only a warning. It's not a spec violaton, and mkvalidator > assumes that you have normal video with more or less regular intervals > between key frames. The specs definitely do not require a cluster to start > with a key frame. It's optimal if they do so that less seeking is required. > But that's why it's only a warning. > > I have argued with Steve Lhomme (mkvalidator's author) that people will take > warnings by a "validation tool" very serious, and that this is not worth a > warning but maybe only a recommendation (or whatever such a info level might > be called). But that's my personal opinion. At the end of the validation mkvalidator says if the file is valid or not. If there are only warnings then it is said as valid. Now a warning is a sign could be muxed better or there may be something that is hard to detect but may be investigated. On the other end if a file is not valid it really should be fixed (and the best way is to fix the muxer). -- Steve Lhomme Matroska association Chairman From slhomme at matroska.org Sat Apr 2 13:33:35 2011 From: slhomme at matroska.org (Steve Lhomme) Date: Sat, 2 Apr 2011 13:33:35 +0200 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: <85d114c13ee2154a6dcdc1cead7a24b7@www.bunkus.org> References: <4D9620CB.3060409@jetheaddev.com> <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org> <85d114c13ee2154a6dcdc1cead7a24b7@www.bunkus.org> Message-ID: On Fri, Apr 1, 2011 at 10:57 PM, Moritz Bunkus wrote: > Hey, > > On Fri, 01 Apr 2011 13:15:14 -0700, John Stebbins wrote: >> Also, if it is just a warning, mkclean shouldn't error out on streams >> that have no keyframes. > > That's a bug in mkclean then. Steve? Yes and no. I will probably not work with --remux set. But how is it possible for a stream not to have any keyframe at all ? Technically the track will just never play. -- Steve Lhomme Matroska association Chairman From slhomme at matroska.org Sat Apr 2 14:03:00 2011 From: slhomme at matroska.org (Steve Lhomme) Date: Sat, 2 Apr 2011 14:03:00 +0200 Subject: [Matroska-devel] weba audio format In-Reply-To: <201104010927.33475.moritz@bunkus.org> References: <201103311652.05410.mconstable@gmail.com> <201104010927.33475.moritz@bunkus.org> Message-ID: On Fri, Apr 1, 2011 at 9:27 AM, Moritz Bunkus wrote: > Hey, > > On Thursday 31 March 2011 22:20:14 Steve Lhomme wrote: > >> That could be investigated, because right now one of the big problem >> in WebM is the size of the Vorbis init packet. It would be very to add >> an extension to Matroska to have it compressed. In fact being a codec >> private it would just take calling the codec "A_VORBIS2" and treating >> the codec private as zlib compressed rather than raw. Adding support >> in all OSS players will be very fast and easy. > > Matroska (unlike WebM) supports compressing codecprivate already via the > ContentEncoding element and its children (with (ContentEncodingScope & > 2) == 2). It would make much more sense to use ContentEncoding with WebM > as well (support for basic zlib compressed subtitles is already present > in many hardware players) than to invent arbitrary new codecIDs just for > this minor difference. > > Not that compressing would actually gain all that much as this > discussion showed. Well, the size of the CodecPrivate for Vorbis is an important issue in WebM. In the case of adaptive streaming you need to download the next part of the stream right after the first one is finished. And the extra data overhead between the 2 Clusters is as much time/bandwidth wasted to achieve the target bandwidth. That's why a tight container is important there and also the smallest header (+trailer) possible as well. Unfortunately the Vorbis init is making the WebM case worse than it should. If we can reduce it by half that would be really welcome. -- Steve Lhomme Matroska association Chairman From bastik.public.mailinglist at gmx.de Sat Apr 2 14:38:16 2011 From: bastik.public.mailinglist at gmx.de (Sebastian G. ) Date: Sat, 02 Apr 2011 14:38:16 +0200 Subject: [Matroska-devel] LICENSE as official tag? Message-ID: <4D9718B8.5060101@gmx.de> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hello, I'd like to suggest "LICENSE" to be official an official tag. "COPYRIGHT" and "TERMS_OF_USE" may not reflect the license something is put under. I'm aware that anyone can use any tag, but strong recommendations are pretty helpful when it comes to supporting tags or using tags probably anybody else is using. IMO "LICENSE" is something that should be covered by those strong recommendations. "COPYRIGHT" is most likely to be used for the copyright holder. "TERMS_OF_USE" may could hold the license, but is not limited to it, since it can contain further restrictions which are not covered by the license. Best Regards, bastik 02 Apr 2011, 14:38 (+0200) -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (MingW32) Comment: Download at http://www.gnupt.de iF4EAREIAAYFAk2XGLgACgkQ31deHVTmOFAofQD/YTqpnq85PPYUgiD36Yy18r/S CxnTKYGnE+mWWvwewpUBAIR0E/K9K19UcZWXEyyJhgFfRCXG3Rfl3TJl15gSkLQw =OOKr -----END PGP SIGNATURE----- From stebbins at jetheaddev.com Sun Apr 3 03:45:43 2011 From: stebbins at jetheaddev.com (John Stebbins) Date: Sat, 02 Apr 2011 18:45:43 -0700 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: References: <4D9620CB.3060409@jetheaddev.com> <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org> <85d114c13ee2154a6dcdc1cead7a24b7@www.bunkus.org> Message-ID: On 04/02/2011 04:33 AM, Steve Lhomme wrote: > On Fri, Apr 1, 2011 at 10:57 PM, Moritz Bunkus wrote: > But how is it possible for a stream not to have any keyframe at all ? Technically > the track will just never play. > Periodic Intra Refresh. It is a mode that x264 supports. There are no keyframes, but there are sei recovery points. These specify that if you start decoding from that point, the frame will fully refresh with X frames. From moritz at bunkus.org Sun Apr 3 11:11:42 2011 From: moritz at bunkus.org (Moritz Bunkus) Date: Sun, 03 Apr 2011 10:11:42 +0100 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: References: "\"<4D9620CB.3060409@jetheaddev.com> <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org>" " <85d114c13ee2154a6dcdc1cead7a24b7@www.bunkus.org> Message-ID: <922e4bd5f549807f976dabbaecc0c1ca@www.bunkus.org> Hey, On Sat, 02 Apr 2011 18:45:43 -0700, John Stebbins wrote: > Periodic Intra Refresh. It is a mode that x264 supports. There are > no keyframes, but there are sei recovery points. These specify that > if you start decoding from that point, the frame will fully refresh > with X frames. mkvmerge marks SEI recovery points as key frames at the container level. From the perspective of the container a key frame is one that a decoder can start decoding from. If you don't mark any frame as a key frame then most players will have a hard time seeking to anywhere in the file. Regards, mosu From slhomme at matroska.org Sun Apr 3 18:47:53 2011 From: slhomme at matroska.org (Steve Lhomme) Date: Sun, 3 Apr 2011 18:47:53 +0200 Subject: [Matroska-devel] weba audio format In-Reply-To: References: <201103311652.05410.mconstable@gmail.com> <201104010927.33475.moritz@bunkus.org> Message-ID: I just added support for Codec Private zlib compression in mkclean 0.8.2. It doesn't do anything on WebM files as compression is not supported. So you need to use "--doctype 2 --optimize" to have compressed Vorbis private data (will have no effect on other codec where the codec private is too small). In the end we save 1.3 KB out of 4 KB (about a third of the size in general). That's about one TCP packet so it's nice for streaming startup time, and it's very easy to handle on the playback side. Unfortunately most players probably don't support it yet (Haali's splitter does). As said before it could be useful for adaptive streaming so maybe should be considered for WebM. Steve On Sat, Apr 2, 2011 at 2:03 PM, Steve Lhomme wrote: > On Fri, Apr 1, 2011 at 9:27 AM, Moritz Bunkus wrote: >> Hey, >> >> On Thursday 31 March 2011 22:20:14 Steve Lhomme wrote: >> >>> That could be investigated, because right now one of the big problem >>> in WebM is the size of the Vorbis init packet. It would be very to add >>> an extension to Matroska to have it compressed. In fact being a codec >>> private it would just take calling the codec "A_VORBIS2" and treating >>> the codec private as zlib compressed rather than raw. Adding support >>> in all OSS players will be very fast and easy. >> >> Matroska (unlike WebM) supports compressing codecprivate already via the >> ContentEncoding element and its children (with (ContentEncodingScope & >> 2) == 2). It would make much more sense to use ContentEncoding with WebM >> as well (support for basic zlib compressed subtitles is already present >> in many hardware players) than to invent arbitrary new codecIDs just for >> this minor difference. >> >> Not that compressing would actually gain all that much as this >> discussion showed. > > Well, the size of the CodecPrivate for Vorbis is an important issue in > WebM. In the case of adaptive streaming you need to download the next > part of the stream right after the first one is finished. And the > extra data overhead between the 2 Clusters is as much time/bandwidth > wasted to achieve the target bandwidth. That's why a tight container > is important there and also the smallest header (+trailer) possible as > well. Unfortunately the Vorbis init is making the WebM case worse than > it should. If we can reduce it by half that would be really welcome. > > -- > Steve Lhomme > Matroska association Chairman > -- Steve Lhomme Matroska association Chairman From stebbins at jetheaddev.com Sun Apr 3 23:21:44 2011 From: stebbins at jetheaddev.com (John Stebbins) Date: Sun, 03 Apr 2011 14:21:44 -0700 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: <922e4bd5f549807f976dabbaecc0c1ca@www.bunkus.org> References: "\"<4D9620CB.3060409@jetheaddev.com> <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org>" " <85d114c13ee2154a6dcdc1cead7a24b7@www.bunkus.org> <922e4bd5f549807f976dabbaecc0c1ca@www.bunkus.org> Message-ID: On 04/03/2011 02:11 AM, Moritz Bunkus wrote: > Hey, > > On Sat, 02 Apr 2011 18:45:43 -0700, John Stebbins wrote: > >> Periodic Intra Refresh. It is a mode that x264 supports. There are >> no keyframes, but there are sei recovery points. These specify that >> if you start decoding from that point, the frame will fully refresh >> with X frames. > > mkvmerge marks SEI recovery points as key frames at the container level. From the perspective of the container a key > frame is one that a decoder can start decoding from. If you don't mark any frame as a key frame then most players will > have a hard time seeking to anywhere in the file. > > Good to know. I was wondering if that might be the right way to handle such video. Thanks John From stebbins at jetheaddev.com Sun Apr 3 23:28:17 2011 From: stebbins at jetheaddev.com (John Stebbins) Date: Sun, 03 Apr 2011 14:28:17 -0700 Subject: [Matroska-devel] Questions about keyframe requirement at start of cluster In-Reply-To: <4D98E4E8.9040605@jetheaddev.com> References: "\"<4D9620CB.3060409@jetheaddev.com> <75f9d03459e74007e0aaa5b2ff4a343d@www.bunkus.org>" " <85d114c13ee2154a6dcdc1cead7a24b7@www.bunkus.org> <922e4bd5f549807f976dabbaecc0c1ca@www.bunkus.org> <4D98E4E8.9040605@jetheaddev.com> Message-ID: On 04/03/2011 02:21 PM, John Stebbins wrote: > On 04/03/2011 02:11 AM, Moritz Bunkus wrote: >> Hey, >> >> On Sat, 02 Apr 2011 18:45:43 -0700, John Stebbins wrote: >> >>> Periodic Intra Refresh. It is a mode that x264 supports. There are >>> no keyframes, but there are sei recovery points. These specify that >>> if you start decoding from that point, the frame will fully refresh >>> with X frames. >> >> mkvmerge marks SEI recovery points as key frames at the container level. From the perspective of the container a key >> frame is one that a decoder can start decoding from. If you don't mark any frame as a key frame then most players >> will have a hard time seeking to anywhere in the file. >> >> > > Good to know. I was wondering if that might be the right way to handle such video. > > Thanks > John > Thinking about this some more. If you are using regular old Block's instead of SimpleBlock, the method of marking it as a keyframe is to *not* add the ReferenceBlock element. Since all frames reference other frames, this is technically incorrect. I suppose it's a distinction that makes no difference if nobody uses the ReferenceBlock element for anything except keyframe detection. Is it likely anybody uses it for any other purpose? From computacao.rafael at gmail.com Mon Apr 4 13:19:15 2011 From: computacao.rafael at gmail.com (Rafael) Date: Mon, 4 Apr 2011 08:19:15 -0300 Subject: [Matroska-devel] Sony XDCAM EX1 Message-ID: Hi there, I was looking for a good MP4 splitter when I found the Haali Media Splitter, and started doing some tests with it. However, for the outputs from Sony XDCAM EX1 camera, wich has a mp4 container, the splitter didn't find the audio stream. I've tested many samples from many different sources using this camera and all had the same results: a video stream (Mpeg2Video) and no audio stream, when there should be a PCM there. Thanks for you attention, Rafael. -------------- next part -------------- An HTML attachment was scrubbed... URL: From slhomme at matroska.org Mon Apr 4 22:08:40 2011 From: slhomme at matroska.org (Steve Lhomme) Date: Mon, 4 Apr 2011 22:08:40 +0200 Subject: [Matroska-devel] LICENSE as official tag? In-Reply-To: <4D9718B8.5060101@gmx.de> References: <4D9718B8.5060101@gmx.de> Message-ID: I agree on this. Anyone has objections/comments ? On Sat, Apr 2, 2011 at 2:38 PM, Sebastian G. wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA256 > > Hello, > > I'd like to suggest "LICENSE" to be official an official tag. > > "COPYRIGHT" and "TERMS_OF_USE" may not reflect the license something is > put under. > > I'm aware that anyone can use any tag, but strong recommendations are > pretty helpful when it comes to supporting tags or using tags probably > anybody else is using. > > IMO "LICENSE" is something that should be covered by those strong > recommendations. > > "COPYRIGHT" is most likely to be used for the copyright holder. > "TERMS_OF_USE" may could hold the license, but is not limited to it, > since it can contain further restrictions which are not covered by the > license. > > Best Regards, > bastik > > 02 Apr 2011, 14:38 (+0200) > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v1.4.11 (MingW32) > Comment: Download at http://www.gnupt.de > > iF4EAREIAAYFAk2XGLgACgkQ31deHVTmOFAofQD/YTqpnq85PPYUgiD36Yy18r/S > CxnTKYGnE+mWWvwewpUBAIR0E/K9K19UcZWXEyyJhgFfRCXG3Rfl3TJl15gSkLQw > =OOKr > -----END PGP SIGNATURE----- > _______________________________________________ > Matroska-devel mailing list > Matroska-devel at lists.matroska.org > http://lists.matroska.org/cgi-bin/mailman/listinfo/matroska-devel > Read Matroska-Devel on GMane: http://dir.gmane.org/gmane.comp.multimedia.matroska.devel > -- Steve Lhomme Matroska association Chairman From slhomme at matroska.org Mon Apr 4 22:18:16 2011 From: slhomme at matroska.org (Steve Lhomme) Date: Mon, 4 Apr 2011 22:18:16 +0200 Subject: [Matroska-devel] =?iso-8859-1?q?Please_add_alaw/=B5law_mapping_in?= =?iso-8859-1?q?_matroska?= In-Reply-To: <1301902577.2427.2.camel@e6500> References: <1301902577.2427.2.camel@e6500> Message-ID: Hello Nicola, If it's supported in avimux/demux then it can be supported in Matroska via A_MS/ACM. Steve On Mon, Apr 4, 2011 at 9:36 AM, Mailing List SVR wrote: > Hi, > > would be useful to add alaw/?law mapping so we can use the matroska > container with this audio codec without inventing our own mapping, > > please take a look at this gstreamer enhancement request too > > https://bugzilla.gnome.org/show_bug.cgi?id=646567 > > thanks > Nicola > > -- Steve Lhomme Matroska association Chairman From roman at alax.info Wed Apr 13 00:37:49 2011 From: roman at alax.info (Roman Ryltsov) Date: Wed, 13 Apr 2011 01:37:49 +0300 Subject: [Matroska-devel] Deadlock in Haali Media Splitter filter Message-ID: <4DA4D43D.6070507@alax.info> Hi, I have a problem with Haali Media Splitter filter, which under intensive seeking locks dead. I wonder if anyone is interested in feedback and details on the issue, such as minidumps etc. And/or if there are any program database files available to use with debugger and better understand the problem to possibly work it around. Thanks, Roman From chou at real.com Wed Apr 13 09:08:24 2011 From: chou at real.com (Changyu Hou) Date: Wed, 13 Apr 2011 00:08:24 -0700 Subject: [Matroska-devel] How to implement streaming mkv files in my player? Message-ID: <55A2D694BAFF824EABCF05271B3084FC742D6C812A@SEAMBX.corp.real.com> Dear experts, Since Matroska is streamable, I want to implement the streaming of mkv file in my media player. I used the http protocol to stream Matroska. Would anyone share with me the policy or steps to implements the streaming functionality of Matroska? Other types of media like avi and rv can now stream pretty good in my player. Thanks for your answer and any suggestions is highly appreciated! :) Thanks & Regards, Kevin -------------- next part -------------- An HTML attachment was scrubbed... URL: From k.a.h.v.leeuwen at tue.nl Wed Apr 13 09:56:50 2011 From: k.a.h.v.leeuwen at tue.nl (Ton van Leeuwen) Date: Wed, 13 Apr 2011 07:56:50 +0000 (UTC) Subject: [Matroska-devel] Consistent error in appending Vobsub tracks Message-ID: L.S., I have a consistent problem when appending two mkv files with Vobsub tracks. Although both individual mkv files play perfectly, in the combined file the subtitles are drastically out of sync after the "merging point", order of magnitude 10 seconds. The individual mkv files have been produced by Handbrake 0.9.5, the merging has been done with mkvmerge (mkvtoolnix 4.6.0, using mmg). Same behavior under previous mkvtoolnix version. I extracted all tracks from both input files and the output file with mkvextract, and examined the idx files of the vobsub tracks. The idx file of the combined file shows the shift: I reason that the timestamp of the first sub after the merging point in the combined file should simply be equal to the (maximum) track length of the first mkv file + the timestamp of the first sub in the second file. Instead, this timestamp was (in the example) 11 seconds earlier. By the way, it is not something as simple as using the timestamp of the last sub in the first file's track instead of the length of the total mkv file, since that would lead to a much larger error. Am I doing something wrong, or is there indeed a bug in mkvmerge? Any suggestion appreciated! From slhomme at matroska.org Wed Apr 20 19:05:12 2011 From: slhomme at matroska.org (Steve Lhomme) Date: Wed, 20 Apr 2011 19:05:12 +0200 Subject: [Matroska-devel] How to implement streaming mkv files in my player? In-Reply-To: <55A2D694BAFF824EABCF05271B3084FC742D6C812A@SEAMBX.corp.real.com> References: <55A2D694BAFF824EABCF05271B3084FC742D6C812A@SEAMBX.corp.real.com> Message-ID: Are you talking about live streaming or just a file streaming over the network ? For the latter you have AVI already working there's nothing different to support Matroska if you can play local files. On Wed, Apr 13, 2011 at 9:08 AM, Changyu Hou wrote: > Dear experts, > > > > Since Matroska is streamable, I want to implement the streaming of mkv file > in my media player. I used the http protocol to stream Matroska. Would > anyone share with me the policy or steps to implements the streaming > functionality of Matroska? Other types of media like avi and rv can now > stream pretty good in my player. Thanks for your answer and any suggestions > is highly appreciated! J > > > > Thanks & Regards, > > Kevin > > _______________________________________________ > Matroska-devel mailing list > Matroska-devel at lists.matroska.org > http://lists.matroska.org/cgi-bin/mailman/listinfo/matroska-devel > Read Matroska-Devel on GMane: > http://dir.gmane.org/gmane.comp.multimedia.matroska.devel > -- Steve Lhomme Matroska association Chairman