last day (15 days later) » 

12:03 PM
Are you still online?
 
izx
Yep, still here.
I replicated your scenario -- as well as another one that has been reported (flash videos pause 7 seconds in)
 
Hey there. Sorry, I just saw you comment asking to chat. I'm actually away from the problematic computer, and I'm now on my laptop. I didn't want to be rude and igore you, though.
 
izx
No worries -- do you have a few minutes?
(to talk -- don't need access)
 
Yeah, I appreciate the help!
My problem is a little different than what you just described, though.
I don't get a pause seven seconds in.
 
izx
Your is the stuttering, right?
yes, I know
 
12:05 PM
Yeah, stuttering.
Oh, okay.
Cool - so what did you have in mind.
 
izx
Your computers are both at home, right?
 
Yes.
They're actually within a metre of each other, under my desk.
 
izx
Well, the usual PA stream is about 180 Kbytes/sec -- so it's pretty small in the grand scheme of things.
But my experiments showed me that over a Wifi network, I got the problem you had.
however when the two computers were bridged -- basically almost infinite bandwidth, things worked fine.
So I suspect it is something to do with how the PulseAudio tunnel from server to client works.
 
"Bridged"?
I can believe that.
 
izx
think direct cable from one to the other, gigabit
 
12:08 PM
In other words, not have the router between them?
 
izx
Now, do you route all sound to the comp with the speakers, or do you still play something locally?
No, there should be no need for that, just an analogy :)
 
I route all sound to the computer with the speakers. It should be noted, just to be sure, that if I play a song in Banshee, the sound is perfect. I just mention that because that, to me, makes ot seem strange that it would just be a connection/speed issue.
 
izx
The current system with paprefs, etc. looks like:
App --> PA server server --> network --> PA client server --> speakers
no, it's something to do with how Flash packetizes its audio -- maybe it's using lots of very small chunks to break it up, so it somehow overwhelms something in the chain although the total rate is the same as banshee/rhythmnbox, etc.
 
Shouldn't that be: "App --> PAclient --> network --> PA server --> speakers"?
Ah, that makes sense.
Okay, I'm on board with your theory so far.
 
izx
Sorry, you're right.
Well, the solution is to remove the tunnel, and do this instead:
App --> Network --> PA server --> speakers
 
12:11 PM
Interesting.
How is that acheived?
 
izx
and that works perfectly for me over the previously problematic wireless
 
Cut out the middle man, so to speak.
 
izx
Exactly
On the server (with speakers)
you disable the RTP receiver, but leave the rest on -- this basically creates a PulseAudio socket listening on port 4713 TCP
on the client, you test by starting an app with e.g.
PULSE_SERVER=SERVER_IP:4713 firefox &
 
Can you be a little more specific about "disable the RTP receiver"?
 
izx
so that flash videos in there, etc. will talk directly to PulseAudio on the server
 
12:14 PM
Is that an option in an interface somewhere?
 
izx
I'll write it out in an answer, with screenshots, etc. to test -- basically uncheck one of the boxes in paprefs
yes
The downside to this method is
that if someone wanted to say, halfway through, decide to switch the sound to local speakers, they'd have to restart firefox
obviously that's not something you'd be doing
 
Oh, if that's the only downside, I can totally live with :)
 
izx
and you can force all apps to default to that by setting PULSE_SERVER in /etc/environment
 
My goal is to have all sound centralized so that I don't have to do any extraneous switching or whatever based on what computer and app is playing.
 
izx
Well, most of the common apps work great -- if you run any legacy apps that dont understand PA and demand ALSA or ESD, there's a workaround for that
 
12:16 PM
This sounds very promising. The Flash/packet theory has a lot of weight to it, as it matches all the observed conditions.
 
izx
OK, so I'll write things out and hopefully you can test it (and it works!)
 
Awesome! I'll try it out as soon as I see it (and I'm at home).
 
izx
It might be possible to capture packets and figure out why Flash is problematic, but that's something for another day (besides Flash is closed-source...)
 
I really appreciate you taking the time. That's really big of you.
 
izx
Great, expect it within the next hour or so.
 
12:17 PM
Okay! Thanks again. Over and out. :)
 

last day (15 days later) »