Problem with teletext subtitles special characters
|Status:||Need feedback||Start date:|
|Assignee:||John Törnblom||% Done:|
|Found in version:||3.2||Affected Versions:||3.2, 3.3|
I encounter problem with teletext subtitle display.
Tested using xbmc-pvr or vlc, same behavior.
Special characters like é,è,â are all displayed as # either in live streaming or recording.
The generated mkv file has bad characters too, testing with vlc shows them as #.
Second problem ( perhaps related ) , some lines are not displayed.
When this happens, the previous ttxt line stays on screen until the next line, instead of disapearing.
I've posted 2 samples, the recorded mkv file, and a full raw ts dump of the transponder here:
The tested pids for the raw dump are:
ttxt: 32 ( page : 888 or 889 )
#5 Updated by Adam Sutton 9 months ago
Do we actually know if this is still happening? Does it happen with your own recordings, which presumably have non-ascii in the stream.
Could it be that we're simply not setting the language descriptor in the MKV for subtitle streams? I know we weren't setting the lang for metadata fields, so although it goes through a different path maybe its a similar issue.
#6 Updated by Adam Sutton 9 months ago
Ok, I've just checked and we do store the language (at least that's what it looks like in the code, I've not checked a recording yet). So maybe not that (I also double checked 2.12).
But anyhow worth checking if this is actually still and issue as its an old report. Could even be lack of player support for MKV (i.e. not handling the language field, though I might be clutching at straws there).
#9 Updated by Hotlobster - 9 months ago
I took the time to make some dumps, and document them 2 years ago for this issue, but the file was never downloaded, neither an answer here.
As there were nobody answering on the other side I gave up TvHeadend for VDR/Xbmc combo. It was working well then.
Dump files were deleted few month ago when migrating the server.
I don't use DVB streams anymore, can't help you now, too late sir.
I would reopen this bug: here in Italy I got this bug since early versions of tvheadend. This affects Sky-it channels (encrypted) but also free channels (RAI).
I provides a link to two dumps made using tvheadend (3.2) recordings: one with in mkv format and one with pass-throught options (ts). They are from the same channel (RAI 1 - free) but in different moment (sorry).
I can see wrong chars in xbmc interface (also during live view) and in vlc on the recordings.
Let me know if you need other infos or if I have to open a new bug report.
#12 Updated by Adam Sutton 7 months ago
- Category changed from Demultiplex to PVR
- Status changed from Invalid to Need feedback
- Priority changed from High to Normal
- Found in version changed from 2.12 to 3.2
- Affected Versions 3.2, 3.3 added
Ok, I will re-open and hopefully John can take a look when he has a chance.
Just to clarify though, is the TS version correct? and the MKV version incorrect?
And are things shown correctly when playing streams live?
I'm not able to see the subtitles with VLC on the TS file. I provided it as "original stream" before the muxing of tvheadend. Maybe I'm wrong and I should capture the TS from /dev/adapters.
In the MKV file you can extract the wrong subtitles (I used mkvextract). Some examples of wrong chars in the subtitles:
- 'Bravi ragazzi perchc tua mamma' should be 'Bravi ragazzi perché tua mamma'
- 'mamma ú giovanissima' should be 'mamma è giovanissima'
- 'Questo cos'ú?' should be 'Questo cos'ì?'
I can provide other samples if you tell me how.
Could you kindly explain how to make the dump? dvbstream? dvbsnoop? With which options?
In alternative, if you live in Europe and your disc is on Hotbird you can find free transmissions with italian dvbtxt subtitles on channels RAI 1, RAI 2 or RAI 3 (mux at 10992.16 MHz).
I've switched my installation to the last git version of tvheadend and now I can provide the full dump of the mux in order to debug the problem with some chars in dvb-txt subs.
I hope the dump is correctly made: the sample comes from channel "RAI 1" on the mux at frequency 10,992,160V on satellite Hotbird 13°.
The dump should be available at the url: https://mega.co.nz/#!RppAxIBb!RcopqASrsntKeV3av1oCDkO1XccCxPOrM43XLJE973I
The text of the subs extracted from the mkv of the channel is available at the url: https://mega.co.nz/#!c0xgjYjS!U17Nv058HuZe7szjRw56LW3z-IZ4io16H5nJ4GIYY7A
Some errors are reported below:
00:00:09,154 --> 00:00:09,154
ú stato fatto
should be "è stato fatto"
00:00:23,722 --> 00:00:23,722
perchc ú uno di quelli che brilla
should be "perché è uno di quelli che brilla"
In the dump there are also chars with accent that are correctly reported; for example:
00:00:20,794 --> 00:00:20,794
ingrediente della sua personalitá
I hope it is enough.