Multiple machines single audio output

[hwttdz]

I think it would be clever to be able to sync multiple machines to play the same audio at the same time. The idea was sparked by this thread.

http://ubuntuforums.org/showthread.php?t=1328338

After reading I think it should be possible to do roughly the same with the audio out, such that two machines play the same audio at the same time (possibly with network delay, which should be minimal over a LAN). This would allow one to have the same music playing on two different machines synced up. Of course it would be better if one could limit it to only say rhythmbox. Any ideas?

[NoaHall]

Why would you want to? Too much network traffic, and one will become out of sync.

[hwttdz]

On a local network I think one might be able to avoid both these problems. And how will I ever know if I'm not able to test?

[Xbehave]

Pulseaudio is the tool your looking for, it's network transparent so what you want shouldn't be too hard.

[Grenage]

You can't limit it to rythmbox, but you could easily do it in the same way as that poster. I don't know why you'd want to, but you could.

[hwttdz]

Ok, given I can open an ssh connection and I have root access on both machines, how could I do this? i.e. what would the command be, would I necessarily need to clobber the audio on the receiving machine?

I would want to so that I could have the same music playing on two machines in sync, of course the real reason I would want to is because I'm curious.

[Dougie187]

You can't limit it to rythmbox, but you could easily do it in the same way as that poster. I don't know why you'd want to, but you could.

Maybe you dont have a home sound system, and you want to listen to music all over the house while you are cleaning. Also, you don't have a portable music player, or a sound system that is capable of being heard throughout the house.

This way you could seamlessly play your music all over the house at the same time, and as you clean you don't have to mess with multiple computers or any issues from the songs begin out of sync.

[Xbehave]

Why would you want to? Too much network traffic, and one will become out of sync.
you can stream audio of the internet so over lan shouldn't be a problem. A typical lan latency is 1ms that is virtually unoticable.


You can't limit it to rythmbox, but you could easily do it in the same way as that poster. I don't know why you'd want to, but you could.
I'm not pulseaudio expert but limiting it to a single app is exactly what PA can do!

[hwttdz]

remove

[Grenage]

You'd simply do what the other poster did. He/she is simply taking all audio from one machine to the other.

[Dougie187]

Ok, given I can open an ssh connection and I have root access on both machines, how could I do this? i.e. what would the command be, would I necessarily need to clobber the audio on the receiving machine?

I would want to so that I could have the same music playing on two machines in sync, of course the real reason I would want to is because I'm curious.

Well, assuming it works. You could go to the machine that you want to play the synced music. So imagine you have your laptop that has music on it, and a desktop that you want to hear the music from.

First, you need to make sure openssh-server is installed on your laptop.

So, on your desktop you would type this

ssh <username on laptop>@<ip to laptop> 'dd bs=1k if=/dev/audio' > /dev/audio

that should work as he said before, though you might have to play with either /dev/audio if it's in a different location, or the bs=1k if the block size needs to be larger.

[Grenage]

I'm not pulseaudio expert but limiting it to a single app is exactly what PA can do!

That is quite possible, but I've never used it for that purpose, so I cannot comment. It does sound like a better way to go, probably get better sound quality.

[hwttdz]

Just discovering paprefs, pulseaudio is getting to be pretty clever. Of course it would probably be moreso if I had any idea how to use it.

I think I've got the first possible solution of copying all audio using dd figured. It seems to me that there is in all likelyhood a pulseaudio solution that forwards all sound, and then further a solution that forwards only rhythmbox.

[Dougie187]

I'm not sure about this, but I think the biggest obstacle in these situations is the synchronization between the machines.

Did it work well with the ssh solution?

[Xbehave]

I never quite got the PA terminology, but you should be able to define a sink on all the clients (sound from these) that is the server (machine with speakers). On the server there will need to be a source (representin the netwrok) that goes to the real sink (the speakers)
Each of the clients can be set up to use the network sink as the output for just rythembox

http://wiki.archlinux.org/index.php/PulseAudio#PulseAudio_over_network
is a guide to setup PA to network all sound, it avoids using the words sink/source tho so i'm not sure how it relates to my description.


I'm not sure about this, but I think the biggest obstacle in these situations is the synchronization between the machines.
That is exactly what PA sorts out, beside a latency of 1ms is unnoticable and as it's simply sound to server, ther is no need for syncronisation.

[Dougie187]

I never quite got the PA terminology, but you should be able to define a sink on all the clients (sound from these) that is the server (machine with speakers). On the server there will need to be a source (representin the netwrok) that goes to the real sink (the speakers)
Each of the clients can be set up to use the network sink as the output for just rythembox

http://wiki.archlinux.org/index.php/PulseAudio#PulseAudio_over_network
is a guide to setup PA to network all sound, it avoids using the words sink/source tho so i'm not sure how it relates to my description.

I think it would be the other way in your description. Sink would be on the server (with the music you want to hear), and Source would be on all the clients (machines you want synced with the server).

But I may have just misunderstood what you meant.


That is exactly what PA sorts out, beside a latency of 1ms is unnoticable and as it's simply sound to server, ther is no need for syncronisation.

That's good to know. But yeah, 1ms is not going to be noticeable.

[Xbehave]

I think it would be the other way in your description. Sink would be on the server (with the music you want to hear), and Source would be on all the clients (machines you want synced with the server).

Client
Source=Rythembox
Sink=Network

Server
Source=Network
Sink=Speakers

but I may be wrong

[Dougie187]

Client
Source=Rythembox
Sink=Network

Server
Source=Network
Sink=Speakers

but I may be wrong

I think your Sources and Sinks are right, but Client and server are wrong. You want the server to be the rhythmbox feed, and the clients to have the speakers. At least from what I understand.

Just because you only want one computer to be the one playing music, and we could call that the server. And then multiple computers will be playing that music through their speakers so you could hear it anywhere, and you could call these the clients.

[Xbehave]

I think your Sources and Sinks are right, but Client and server are wrong. You want the server to be the rhythmbox feed, and the clients to have the speakers. At least from what I understand.

Just because you only want one computer to be the one playing music, and we could call that the server. And then multiple computers will be playing that music through their speakers so you could hear it anywhere, and you could call these the clients.
Ah i think i misunderstood what OP wanted.

[Dragonbite]

Wouldn't setting up a machine, like a server, as a shoutbox server do the same thing?

Otherwise making a shortcut that runs the ssh .. command mentioned might work. Again, if it is a machine acting like a server then any desktops on the network can access it whenever they want.

[gnomeuser]

Ironically given all the misdirected hatred it is the subject of, this is one of the situations PulseAudio handles extremely easily.

[hwttdz]

gnomeuser, you don't happen to know how to do it using pulseaudio, i.e. what files I need to edit and which commands I need to run?

[gnomeuser]

gnomeuser, you don't happen to know how to do it using pulseaudio, i.e. what files I need to edit and which commands I need to run?

Something akin to this would work.

http://wiki.archlinux.org/index.php/PulseAudio#PulseAudio_over_network

[hwttdz]

Of course ubuntu defaults to per user pulseaudio sessions, so it would instead be in the user file instead of the system file. I can't even manage to get started with that because I can't get even the first steps of pulseaudio going from the command line.


user 14:59:48 @ texian ~> pulseaudio -C
E: socket-server.c: bind(): Address already in use
E: module.c: Failed to load module "module-esound-protocol-unix" (argument: ""): initialization failed.
E: main.c: Module load failed.
E: main.c: Failed to initialize daemon.

When pulseaudio isn't running, due to stale locks. I've tried deleting /temp/pulse-*, but to no avail. This is a reported bug, and will probably be ironed out in time.

I realize that theoretically this is easy especially with the design of pulseaudio, but there doesn't seem to be much useful documentation, and that which exists seems to run into the reality not working the same as the documentation.

Edit: pulseaudio 0.9.19

[hwttdz]

A first step:
killall pulseaudio && sudo rm -rf /tmp/pulse-* && pulseaudio -nC
# -n means don't read the startup file
# -C gets you a pulseaudio command prompt

And to play around without screwing up pulse audio for everyone (and to have a convenient backup):
cp /etc/pulse/daemon.conf ~/.pulse/daemon.conf
cp /etc/pulse/default.pa ~/.pulse/default.pa
cp /etc/pulse/client.conf ~/.pulse/client.conf

[Dougie187]

I was just curious if you got this to work, and if you did how well it works. With either of the two methods.

[hwttdz]

Nope, not yet, and it may be some time. I'm a little confused as to how the gui interacts with the configuration files. And if the gui makes changes to the user config files or the /etc/ config files.

[Dougie187]

Well, I would assume that the gui made user specific changes, but I'm also not really sure that it matters for your test cases.

I think everything you want to do has to be done in the configuration files. As that's where you would define new sinks and sources. But I am not completely sure, because I have never done it myself.

[hwttdz]

But it gets tricky because there are a few ways to configure pulseaudio
1) system wide config (not recommended)
2) per user config, reading files from /etc/pulse (ubuntu default)
3) per user config reading files from ~

I'm guessing the gui makes changes to the file processed in 2.