OP25 OP25 AMBE Tone Feature Missing

[squirrel]

I have done a fresh install on a Dell E4310 laptop (Core i5 M650 @ 2.67Ghz, 8GB Ram) with Ubuntu 18.04.1 LTS and latest version of OP25. I am still getting the audio delay / cutoff issue. I have checked the log and it looks like there were about 15 PCM underruns reported in 24 hours. My startup command is below. Just like on the Pi, if I go back to a version prior to the tones being added everything works fine.

./rx.py --args 'rtl' -N 'LNA:47' -S 57600 -f 772.9812e6 -o 25000 -q 0 -U -T trunk.tsv -V -2 2> stderr.2

Am I missing something else? I was hoping that dumping the Pi and going to a higher power CPU would have solved this issue.

Thanks

 

[boatbod]

I have done a fresh install on a Dell E4310 laptop (Core i5 M650 @ 2.67Ghz, 8GB Ram) with Ubuntu 18.04.1 LTS and latest version of OP25. I am still getting the audio delay / cutoff issue. I have checked the log and it looks like there were about 15 PCM underruns reported in 24 hours. My startup command is below. Just like on the Pi, if I go back to a version prior to the tones being added everything works fine.

./rx.py --args 'rtl' -N 'LNA:47' -S 57600 -f 772.9812e6 -o 25000 -q 0 -U -T trunk.tsv -V -2 2> stderr.2

Am I missing something else? I was hoping that dumping the Pi and going to a higher power CPU would have solved this issue.

Thanks

Very puzzling. Perhaps you could try adding some additional debug to sockaudio.py to make sure the flush is being detected & acted upon. I'm not aware of anyone else who has reported anything remotely similar with the new code.

 

[squirrel]

Very puzzling. Perhaps you could try adding some additional debug to sockaudio.py to make sure the flush is being detected & acted upon. I'm not aware of anyone else who has reported anything remotely similar with the new code.

I have added some logging around the def dump(self): If you can confirm that is what does the flush or correct me.

In the meantime I set a cron job up to restart OP25 every 2 hours and that has seemed to worked around this issue I am having until I can figure this out.

 

[boatbod]

I have added some logging around the def dump(self): If you can confirm that is what does the flush or correct me.

In the meantime I set a cron job up to restart OP25 every 2 hours and that has seemed to worked around this issue I am having until I can figure this out.

Actually you should look at the drain() method invoked at line 310. That in turn calls snd_pcm_drain() at line 228 of sockaudio.py

 

[squirrel]

Actually you should look at the drain() method invoked at line 310. That in turn calls snd_pcm_drain() at line 228 of sockaudio.py

Ok I actually did that first but the drain seemed to be getting hit constantly so I thought I was in the wrong area. I added logging around those 2 lines and I have attached my log file. It appers the self.pcm.drain() is being constantly hit. when a transmission comes in it stops, but then when it completes the drain resumes. The log is only about 2 minutes of running. Not sure if this is normal behavior?

 

[boatbod]

Ok I actually did that first but the drain seemed to be getting hit constantly so I thought I was in the wrong area. I added logging around those 2 lines and I have attached my log file. It appers the self.pcm.drain() is being constantly hit. when a transmission comes in it stops, but then when it completes the drain resumes. The log is only about 2 minutes of running. Not sure if this is normal behavior?

Yes, that's pretty normal; any time there is a TDU (ph1), MAC_HANGTIME (ph2) , or MAC_END_PTT (ph2) the system should flush the pcm because it's considered the "end of a conversation". Some systems send multiple end messages so the pcm can be drained numerous times consecutively.

What you're looking for is a scenario where there is no drain when the voice ends, then new audio starts up and unavoidably plays whatever was left over in the buffer from the previous conversation.

 

[boatbod]

I've pushed a change that will likely cure the problem, but if not, it should make it easier to track down.