search

AlbrechtL / welle.io

A DAB/DAB+ Software Defined Radio (SDR)
https://www.welle.io
Other
625 stars 117 forks source link

BBC National re-config - Frankenstein stream #587

Open andimik opened 4 years ago

andimik commented 4 years ago

BBC National regularly uses re-configuration of the mux. It has been described first in https://www.rundfunkforum.de/viewtopic.php?f=11&t=55891&start=3330 (and following English language pages).

The issue is called Frankenstein stream due to a very strange output message (see below).

It's impossible to listen to the news (because the re-config is done exactly at the beginning of the hour) on some channels.

A raw file is see below (thanks to user DanDAB)

These services play without problems in welle.io

BBC Radio 1 (0xc221) Pri subch= 1 start= 0 CUs= 96 PL=uep 3 bitrate=128 BBC Radio 2 (0xc222) Pri subch= 2 start= 96 CUs= 96 PL=uep 3 bitrate=128 BBC Radio 3 (0xc223) Pri subch= 3 start=192 CUs=140 PL=uep 3 bitrate=192 (it's playing also with 160 kBit compared to Andi's DAB player)

There is no sound after re-config for

BBC Radio 4 (0xc224) Pri subch= 4 start=332 CUs= 96 PL=uep 3 bitrate=128 BBC Radio 5 Live (0xc225) Pri subch= 5 start=428 CUs= 58 PL=uep 3 bitrate=80 BBC Radio 1Xtra (0xc22a) Pri subch=10 start=582 CUs= 96 PL=uep 3 bitrate=128 BBC AsianNetwork (0xc236) Pri subch= 7 start=486 CUs= 48 PL=uep 3 bitrate=64 BBC WorldService (0xc238) Pri subch= 9 start=534 CUs= 48 PL=uep 3 bitrate=64 BBC Radio 4Extra (0xc22c) Pri subch=12 start=774 CUs= 58 PL=uep 3 bitrate=80 BBC Radio 6Music (0xc22b) Pri subch=11 start=678 CUs= 96 PL=uep 3 bitrate=128

See logfile attached welleio.log

AlbrechtL commented 4 years ago

Thanks for reporting. Please upload your file again.

andimik commented 3 years ago

Sorry for the delay, did not read your answer.

Update: I had provided the file for you on my homepage, but this was in year 2020.

AlbrechtL commented 3 years ago

Thanks for the file. Now I can reproduce the issue.

Do you know when the BBC will switch to DAB+?

nickpiggott commented 3 years ago

The reconfiguration process is entirely normal and part of the standard. In the run-up to reconfiguration, the new configuration is sent interleaved with the current configuration (for at least 15 frames prior to the change), and there's a frame counter to tell you exactly which frame to change over on. I wouldn't call it Frankenstein.

The BBC has never made any comment on a timeline for changing to DAB+. It's possible they'll be the last to change and only change some of their services.

AlbrechtL commented 3 years ago

@nickpiggott Thanks for your explanation. Because I'm living in Germany I'm not effected and thus this bug doesn't have high priority on my list.

But I will leave this issue open because we shall fix it.

Pull requests are welcome!

AlbrechtL commented 1 week ago

Does the BBC still broadcasts via DAB?

andimik commented 1 week ago

Yes, it's still DAB (old), but in Central Europe it's no longer possible to receive the BBC DAB feed, as they are on the Spot Beam on 28.2 East. So I cannot provide current samples ...

AlbrechtL commented 1 week ago

The reconfiguration process is entirely normal and part of the standard. In the run-up to reconfiguration, the new configuration is sent interleaved with the current configuration (for at least 15 frames prior to the change), and there's a frame counter to tell you exactly which frame to change over on.

@nickpiggott Can you point me to that direction where to find this information within the frame?

nickpiggott commented 1 week ago

Hello

It's in section 6.5 (Multiplex Reconfiguration) of EN 300 401.

The C/N flag on the relevant FIGs indicates if that FIG refers to the Current or Next configuration of the multiplex. For example, the ordering of subchannels and the start of each channel may change at a reconfiguration.

AlbrechtL commented 1 week ago

Thanks for that. I just saw that the backend already detects the reconfiguration but is not handling it :-(. https://github.com/AlbrechtL/welle.io/blob/330ee0a16318ccb9cb2f70d69cf325feaa143ccb/src/backend/fib-processor.cpp#L144-L157

Maybe an option is simply retuning to the service. This is not nice but should work.

AlbrechtL commented 1 week ago

OK, I added a workaround in https://github.com/AlbrechtL/welle.io/commit/c0434be71fe45c1ca8da9946dc44b5b828e56c84.

It is not a nice solution because it just restarts the decoder, which creates a short audio drop. But it is better than nothing :-).

Please test https://welle-io-nightlies.albrechtloh.de/20241009_c995c85b_Windows_welle-io-setup_x64.exe.

beermad commented 1 week ago

Thanks for this attempt. Unfortunately it seems not to work.

Log shows output from a test run, with audacious streaming the audio on one channel. I hope this gives some useful clues; if there's any more information you need, just ask - is there a debug mode that would give more useful data?

welleTest.log

Log from latest production welle-cli at the same time (fortunately I have two dongles). Note that timestamped lines come from my monitor/restart daemon. wellePrimary.log

AlbrechtL commented 1 week ago

Thanks for the test. I reopened the issue.

Since I'm living in Germany, I can't try this by my own. I did the test with the recording by @andimik.

Can you provide me a RAW file recording? In the expert mode you can do a RAW file recording.

beermad commented 1 week ago

Can you provide me a RAW file recording? In the expert mode you can do a RAW file recording.

Will do. Is that from the option "Enable Tll decoding to console log"? Or is some other option I haven't noticed needed?

andimik commented 1 week ago

No, please have a look at https://www.welle.io/devices/rawfile

AlbrechtL commented 1 week ago

You can also try the experimental RAW file recorder grafik

But you have to catch the reconfiguration. So the best would be to have a sample with +-5 minutes around the reconfiguration.

beermad commented 1 week ago

No, please have a look at https://www.welle.io/devices/rawfile

Great, thanks. The next reconfiguration I can confidently predict will be late on Monday afternoon. I'll record that and upload it.

beermad commented 5 days ago

Raw DAB dump (uploaded to cloud as it's a bit big to do in browser) https://filedn.com/lnJp6X7QX8y063Wuq1QkInV/20241014_165850_12B.iq

Output from welle-cli -f 20241014_165850_12B.iq -p "BBC Radio4" (using the welle-cli built from your updated files). log.txt

AlbrechtL commented 5 days ago

Thanks, I got the file!

andimik commented 5 days ago

but the SNR is very bad, hard to get errorfree audio.

beermad commented 5 days ago

but the SNR is very bad, hard to get errorfree audio.

I was surprised how poor the sound quality was after recording the stream using your dab_raw_record.sh script. Though something seemed a bit odd about it, as it ignored my -t argument and I had to kill it manually after the sync failure had occurred.

I wonder if the hard-coded sample rate might need changing, though when I tried doubling the rate it didn't work.

I may try again, dumping from welle-io when the problem recurs tomorrow evening if necessary.

andimik commented 4 days ago

You could also use

beermad commented 4 days ago

Another attempt at a dump, this time using welle-io. https://filedn.com/lnJp6X7QX8y063Wuq1QkInV/welle-io-record.iq

I think it stopped recording pretty much as soon as the sync was lost, so just have to hope it managed to get what you need. Much better quality this time round.

andimik commented 2 days ago

Yes, much much better (21 dB)!