I've a media player that only understand srt subtitles, so I'm looking for a way to extract the dvb subtitles from a vdr ts recording, and then use some kind of ocr to convert it to srt. I asked google and I came empty handed (many people asking the same question with no working answer).
What I tried so far:
- ProjectX (fails because it doesn't like hd pixel coordinates) - avidemux (it has a ts->srt option but it only allows pids up to to 255, and the lucky few that managed to make it work say that it doesn't work with hd) - mencoder -vobsubout (it just produces garbage instead of good pictures, at least according to subrip)
Any other option?
Bye
Al 15/12/10 18:28, En/na Luca Olivetti ha escrit:
I've a media player that only understand srt subtitles, so I'm looking for a way to extract the dvb subtitles from a vdr ts recording, and then use some kind of ocr to convert it to srt. I asked google and I came empty handed (many people asking the same question with no working answer).
What I tried so far:
- ProjectX (fails because it doesn't like hd pixel coordinates)
- avidemux (it has a ts->srt option but it only allows pids up to to
255, and the lucky few that managed to make it work say that it doesn't work with hd)
- mencoder -vobsubout (it just produces garbage instead of good
pictures, at least according to subrip)
Any other option?
I found how to do it but it's way too cumbersom:
the cvs version of ProjectX supports HD subtitles, *but* the sup exported subtitles aren't readable neither by suprip (not subrip!) nor by BDSup2Sub, so I additionally export to sub format and use BDSup2Sub to convert it to a valid sup, that can then be processed by suprip.
Both ProjectX and BDSup2Sup, being java programs, work under Linux, but suprip is windows only.
Projectx: http://sourceforge.net/projects/project-x/
BDSup2Sub http://forum.doom9.org/showthread.php?t=145277
SupRip http://exar.ch/suprip/
Bye
Al 18/12/10 12:57, En/na Luca Olivetti ha escrit:
Al 15/12/10 18:28, En/na Luca Olivetti ha escrit:
I've a media player that only understand srt subtitles, so I'm looking for a way to extract the dvb subtitles from a vdr ts recording, and then use some kind of ocr to convert it to srt. I asked google and I came empty handed (many people asking the same question with no working answer).
[...]
I found how to do it but it's way too cumbersom:
the cvs version of ProjectX supports HD subtitles, *but* the sup exported subtitles aren't readable neither by suprip (not subrip!) nor by BDSup2Sub, so I additionally export to sub format and use BDSup2Sub to convert it to a valid sup, that can then be processed by suprip.
Both ProjectX and BDSup2Sup, being java programs, work under Linux, but suprip is windows only.
Projectx: http://sourceforge.net/projects/project-x/
BDSup2Sub http://forum.doom9.org/showthread.php?t=145277
SupRip http://exar.ch/suprip/
I found a simpler (i.e scriptable) way: I wrote a simple python script that uses ProjectX (cvs version) to extract the subtitles from the ts file, BDSup2Sub to convert the subtitles to png images, ImageMagick's convert to improve the resulting images and finally gocr to convert the images to text. It seems to work acceptably with recordings from the bbchd/bbcone-hd. It's available here:
http://ventoso.org/luca/vdr/vdrsubrip/
Both ProjectX and BDSup2Sub will spit a million warnings, but the final result is ok. As always YMMV.
Bye
Al 01/01/11 18:09, En/na Luca Olivetti ha escrit:
I found a simpler (i.e scriptable) way: I wrote a simple python script that uses ProjectX (cvs version) to extract the subtitles from the ts file, BDSup2Sub to convert the subtitles to png images, ImageMagick's convert to improve the resulting images and finally gocr to convert the images to text. It seems to work acceptably with recordings from the bbchd/bbcone-hd. It's available here:
http://ventoso.org/luca/vdr/vdrsubrip/
Both ProjectX and BDSup2Sub will spit a million warnings, but the final result is ok. As always YMMV.
Yesterday I noticed that the timing offset is not fixed but depends on the recording (e.g, with my previous experiments I hardcoded a 4 seconds delay for the bbc channels, while yesterday I needed a 2 minutes delay). I don't know the cause of this offset (ProjectX or the crappy media player of my tv) but the problem can be easily solved with Subtitles:
http://karasik.eu.org/software/
Bye
En/na Luca Olivetti ha escrit:
Yesterday I noticed that the timing offset is not fixed but depends on the recording (e.g, with my previous experiments I hardcoded a 4 seconds delay for the bbc channels, while yesterday I needed a 2 minutes delay). I don't know the cause of this offset (ProjectX or the crappy media player of my tv) but the problem can be easily solved with Subtitles:
I now know the cause of the offset, but I don't know how to (easily) fix it: Project-X for some reason (maybe a wrong option I gave?) doesn't extract the pts from the video stream, so it doesn't synchronize the subtitles to the video but it simply assumes the time starts with the first subtitle. If, e.g., the first subtitle appears 3 minutes into the recording, it will get a timestamp of 0 seconds so it will show 3 minutes early.
These are the options I'm giving to Project-X (via its ini file):
# Application Application.Agreement=1 Application.OutputDirectory=/temporary/word/directory
# CollectionPanel CollectionPanel.CutMode=0
# SubtitlePanel SubtitlePanel.SubpictureColorModel=UkFreeview(mc) SubtitlePanel.enableHDSub=1 SubtitlePanel.exportAsVobSub=1
ExportPanel.Streamtype.MpgAudio=0 ExportPanel.Streamtype.Ac3Audio=0 ExportPanel.Streamtype.PcmAudio=0 ExportPanel.Streamtype.Teletext=0 ExportPanel.Streamtype.Subpicture=1 ExportPanel.Streamtype.Vbi=0
Any hint on what option I could try to get the video pts (but only the pts, I don't want to extract the video stream)? Or simply Project-X cannot do it with h264 streams?
Bye
Al 04/01/11 10:01, En/na Luca Olivetti ha escrit:
En/na Luca Olivetti ha escrit:
Yesterday I noticed that the timing offset is not fixed but depends on the recording (e.g, with my previous experiments I hardcoded a 4 seconds delay for the bbc channels, while yesterday I needed a 2 minutes delay). I don't know the cause of this offset (ProjectX or the crappy media player of my tv) but the problem can be easily solved with Subtitles:
I now know the cause of the offset, but I don't know how to (easily) fix it: Project-X for some reason (maybe a wrong option I gave?) doesn't extract the pts from the video stream, so it doesn't synchronize the subtitles to the video but it simply assumes the time starts with the first subtitle. If, e.g., the first subtitle appears 3 minutes into the recording, it will get a timestamp of 0 seconds so it will show 3 minutes early.
These are the options I'm giving to Project-X (via its ini file):
Ok, I found there's an option to enable hd video demuxing (OptionPanel.enableHDDemux), *but* ProjectX crashes with it enabled, so I changed the script to use dvbsnoop to obtain the first pts of the video track and of the subtitles track, and use the difference as a time offset.
BTW: am I the only one needing this and/or wrestling with ProjectX to manipulate vdr recordings? I wasn't exactly overwhelmed with the replies...
Bye
Hi all,
VDR-1.7.16 with plugin osdteletext-0.9.0 crashes on channels with slovak or czech teletext.
You can try on this fta channel:
CT 24;CS Link:12525:VC34M2O0S0:S23.5E:27500:165=2:100=cze@4:52:0:8006:3:3014:0
Any help appreciated
Marek Hajduk
I just get this error:
sock_send: socket write error Segmentation fault
It seems, there is a problem with wrong choosing of character set on Czech or Slovak teletext pages.
How can I debug it?
Regards
Marky
-----Original Message----- From: vdr-bounces@linuxtv.org [mailto:vdr-bounces@linuxtv.org] On Behalf Of Marek Hajduk Sent: Sunday, January 02, 2011 11:40 PM To: 'VDR Mailing List' Subject: [vdr] Problem with osdteletext plugin 0.9.0
Hi all,
VDR-1.7.16 with plugin osdteletext-0.9.0 crashes on channels with slovak or czech teletext.
You can try on this fta channel:
CT 24;CS Link:12525:VC34M2O0S0:S23.5E:27500:165=2:100=cze@4:52:0:8006:3:3014:0
Any help appreciated
Marek Hajduk
_______________________________________________ vdr mailing list vdr@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/vdr
I'm successfully using osdteletext-0.8.2.cz from http://www.cssf.cz/showthread.php?35867-vdr-1.7.10-a-eHD with attached patch.
Regards Michal
On 01/05/2011 11:51 PM, Marek Hajduk wrote:
I just get this error:
sock_send: socket write error Segmentation fault
It seems, there is a problem with wrong choosing of character set on Czech or Slovak teletext pages.
How can I debug it?
Regards
Marky
-----Original Message----- From: vdr-bounces@linuxtv.org [mailto:vdr-bounces@linuxtv.org] On Behalf Of Marek Hajduk Sent: Sunday, January 02, 2011 11:40 PM To: 'VDR Mailing List' Subject: [vdr] Problem with osdteletext plugin 0.9.0
Hi all,
VDR-1.7.16 with plugin osdteletext-0.9.0 crashes on channels with slovak or czech teletext.
You can try on this fta channel:
CT 24;CS Link:12525:VC34M2O0S0:S23.5E:27500:165=2:100=cze@4:52:0:8006:3:3014:0
Any help appreciated
Marek Hajduk
vdr mailing list vdr@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/vdr
vdr mailing list vdr@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/vdr
Thanks Michal for your answer,
I tried it, but my vdr can't start now.
Here is my log:
vdr: /data/install_dvb-s2/vdr/vdr/PLUGINS/lib/libvdr-osdteletext.so.1.7.16: undefined symbol: _ZN11cRenderPage18ReadTeletextHeaderEPh
Marky
-----Original Message----- From: vdr-bounces@linuxtv.org [mailto:vdr-bounces@linuxtv.org] On Behalf Of Michal Sent: Thursday, January 06, 2011 3:15 PM To: vdr@linuxtv.org Subject: Re: [vdr] Problem with osdteletext plugin 0.9.0
I'm successfully using osdteletext-0.8.2.cz from http://www.cssf.cz/showthread.php?35867-vdr-1.7.10-a-eHD with attached patch.
Regards Michal
On 01/05/2011 11:51 PM, Marek Hajduk wrote:
I just get this error:
sock_send: socket write error Segmentation fault
It seems, there is a problem with wrong choosing of character set on Czech or Slovak teletext pages.
How can I debug it?
Regards
Marky
-----Original Message----- From: vdr-bounces@linuxtv.org [mailto:vdr-bounces@linuxtv.org] On Behalf Of Marek Hajduk Sent: Sunday, January 02, 2011 11:40 PM To: 'VDR Mailing List' Subject: [vdr] Problem with osdteletext plugin 0.9.0
Hi all,
VDR-1.7.16 with plugin osdteletext-0.9.0 crashes on channels with slovak or czech teletext.
You can try on this fta channel:
CT 24;CS Link:12525:VC34M2O0S0:S23.5E:27500:165=2:100=cze@4:52:0:8006:3:3014:0
Any help appreciated
Marek Hajduk
vdr mailing list vdr@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/vdr
vdr mailing list vdr@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/vdr
Aaaaaaaargh, my mistake.
Michal thank you. Osdteletext-0.8.2 with cz patch and your patch Works.
Regards
Marky
-----Original Message----- From: vdr-bounces@linuxtv.org [mailto:vdr-bounces@linuxtv.org] On Behalf Of Marek Hajduk Sent: Thursday, January 06, 2011 6:33 PM To: 'VDR Mailing List' Subject: Re: [vdr] Problem with osdteletext plugin 0.9.0
Thanks Michal for your answer,
I tried it, but my vdr can't start now.
Here is my log:
vdr: /data/install_dvb-s2/vdr/vdr/PLUGINS/lib/libvdr-osdteletext.so.1.7.16: undefined symbol: _ZN11cRenderPage18ReadTeletextHeaderEPh
Marky
-----Original Message----- From: vdr-bounces@linuxtv.org [mailto:vdr-bounces@linuxtv.org] On Behalf Of Michal Sent: Thursday, January 06, 2011 3:15 PM To: vdr@linuxtv.org Subject: Re: [vdr] Problem with osdteletext plugin 0.9.0
I'm successfully using osdteletext-0.8.2.cz from http://www.cssf.cz/showthread.php?35867-vdr-1.7.10-a-eHD with attached patch.
Regards Michal
On 01/05/2011 11:51 PM, Marek Hajduk wrote:
I just get this error:
sock_send: socket write error Segmentation fault
It seems, there is a problem with wrong choosing of character set on Czech or Slovak teletext pages.
How can I debug it?
Regards
Marky
-----Original Message----- From: vdr-bounces@linuxtv.org [mailto:vdr-bounces@linuxtv.org] On Behalf Of Marek Hajduk Sent: Sunday, January 02, 2011 11:40 PM To: 'VDR Mailing List' Subject: [vdr] Problem with osdteletext plugin 0.9.0
Hi all,
VDR-1.7.16 with plugin osdteletext-0.9.0 crashes on channels with slovak or czech teletext.
You can try on this fta channel:
CT 24;CS Link:12525:VC34M2O0S0:S23.5E:27500:165=2:100=cze@4:52:0:8006:3:3014:0
Any help appreciated
Marek Hajduk
vdr mailing list vdr@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/vdr
vdr mailing list vdr@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/vdr
_______________________________________________ vdr mailing list vdr@linuxtv.org http://www.linuxtv.org/cgi-bin/mailman/listinfo/vdr
Am 06.01.2011 15:15, schrieb Michal:
I'm successfully using osdteletext-0.8.2.cz from http://www.cssf.cz/showthread.php?35867-vdr-1.7.10-a-eHD with attached patch.
This issue is tracked at vdr-developer.org bug tracking: http://projects.vdr-developer.org/issues/469
Cheers,
Udo
-
Luca Olivetti -
Marek Hajduk -
Michal -
Udo Richter