Puppy Linux Discussion Forum

Pulseaudio for Fatdog64-811 public test

This thread concerns a public test of a working PulseAudio for Fatdog-811, in response to recent requests.

The current test set comprises:

• PulseAudio server (pulseaudio 13.99.2)
• canonical Volume Control GUI (pavucontrol 4.0)
• contributed PulseAudio system tray GUI (pasystray 0.7.1-13)

Scripts:

• install script for Fatdog test (install-pulseaudio.sh 2020-11-24)
• tray icon GUI for Fatdog test (pa-tray.sh 2020-12-10)
• standalone Volume Control tray icon (pulseaudio-tray.sh 2020-11-26)

Application support

See post #2

How you can help

See post #2.

Getting started

The whole test set can be downloaded from this shared folder.

• Download the contents of the shared folder to an empty local folder, and make the scripts executable (script names end with .sh).
• Start a terminal window in the local folder.
• Run ./pa-tray.sh &. This will start a green tray icon labelled "PA".
• Right-click the PA tray icon to select from the menu:
  • Back up ALSA audio
  • Start PulseAudio
  • Volume Control

Now the PulseAudio server should be running and a volume control window should be open and show three tabs. Now you can venture in the unfamiliar world of PulseAudio. To end your venture right-click the PA tray icon and select from the menu:

• Exit apps
• Stop PulseAudio
• Restore ALSA audio
• Exit

What happened behind your back

The first time you click "Start PulseAudio" the server and its dependencies are loaded by install-pulseaudio.sh. The test set comprises SFS files and shell scripts only; nothing is installed permanently until you choose to do so. Therefore, if your system's audio gets messed up, you can reboot and it should work again. However, remember that this is a test; you must accept that not everything works as expected.

The direct dependencies that get loaded enable 70% of the functionality available in this test set. What is left disabled is pasystray's equalizer and the ability to toggle the volume control window open/closed by left-clicking the PA tray icon. To enable the equalizer install python3 or load the devx SFS. To enable window toggling install package wmctrl.

Once you become confident that PulseAudio works in your system, you could skip steps 2 and 3 and just click the pa-tray.sh file icon to get started.

Scripts

pa-tray.sh is useful to get started with testing but not at all necessary for stable use.

install-pulseaudio.sh loads the test set. Eventually, this script will be superseded by actual packages that you will be able to install with Gslapt Package Manager.

pavucontrol-tray.sh provides a stand-alone tray icon for pavucontrol in case you want one and are not running pa-tray.sh or pasystray.

The shared download folder includes more auxiliary scripts.

Support

I will do what I can to support you but keep in mind I'm not familiar with PulseAudio beyond what I have learned while creating this test set. Also I am not an audio expert. I play very little audio on my system, mostly just youtube and some local mp3 files. My audio hardware is limited to the stock audio card inside my PC and a USB webcam. I'm not even sure yet that PulseAudio has a rightful place in Fatdog-811 but it's interesting to find out through this test effort. How? See post #2.

If you want to help - hey! this is for all Fatdog users after all - just say!

What works - how to help

Fatdog64-811 applications are not compiled with PulseAudio support. Therefore they will not work with PulseAudio. Without applications PulseAudio is nearly useless for Fatdog64. Therefore, in order to move forward, qualifying useful applications becomes very important.

If you want to help with testing applications that work read the Helping section at the end of this post.

Command Line

• paplay, parecord 2020-11-26

GUI

• pavucontrol, pasystray 2020-11-26
• AnyDesk - remote desktop product - supports remote-to-local audio only 2020-11-26
• NoMachine - remote desktop product - overall nicer than AnyDesk - I couldn't notice any audio lag - 2021-01-26
• firefox 83.0 2020-12-09

How-to

• Streaming application audio over home LAN 2020-12-10

Helping

You can help testing by qualifying external applications that work with this pulsaudio test implementation for Fatdog.
Report your findings to this thread.

Suggested sources of possible candidates:

• Web browsers compiled with pulseaudio support
• shell scripts found on GitHub and similar sites
• PulseAudio aware programs in source form -- you will need to compile them
• PulseAudio aware binary programs packaged as AppImage files

Your post should follow some guidelines:

• Applications must run on a vanilla Fatdog64-811 without a savebox
• Include all instructions necessary to install and set up the application running and its dependencies
• Include download links
• If an application works partially, state what works and what doesn't.
• Do not post about applications that do not work, or only work a little.

In the interest of quality, posts that do not follow the above guidelines will be flagged.

To avoid duplicated efforts, post a short message to let other people know which application you are testing.

Applications that are reported to work will be added to the list in this post.

What does not work

To be fixed

2020-12-10: pasystray does not work for user spot.

No way

TBD

Currently testing:

• Firefox
• Network audio inside the LAN

Firefox works

last edit 2020-12-09

The official Firefox update for Fatdog64 works well.

Versions

I tested Firefox version 83.0 with scripts updated on 20201209.

Installing Firefox for the test

This is a one-time task. In the system menu select "Internet > Firefox Browser", which will start a script that downloads the latest official Firefox version, makes a package from the downloaded file, and finally installs the package.

Next download file firefox-pulseaudio.tgz from the shared folder and unpack it in the root / folder. It provides new start scripts and desktop file for firefox with PulseAudio. The scripts starting firefox without apulse, the PulseAudio emulation for ALSA library. When firefox starts without apulse it looks for the real PulseAudio for sound support. Note that the real PulseAudio and apulse are not compatible, they can't be used together.

How to run Firefox PulseAudio as spot

By default on Fatdog64 firefox runs as user spot, that is, firefox-spot. Before running firefox-spot, we need to start the PulseAudio server. The PulseAudio server must run as spot as well because on Fatdog64 PulseAudio is configured as a user's audio server.

For testing purposes, when the PulseAudio server is running start firefox from a terminal with firefox-pulseaudio-spot &. When the PulseAudio server isn't running use firefox-spot &. Let's do it.

As user root, open a terminal window in your PulseAudio test folder and run the following command:

./install-pulseaudio.sh && cp pa-tray.sh /tmp && run-as-spot /tmp/pa-tray.sh &

Wait for the shell prompt to come back then check that an orange square icon labelled "PA" sits in the tray. Hoover the mouse pointer over it and confirm that the tooltip says "Running as spot". Every command you will run from this icon will run as user spot. Right-click the icon and select "Start PulseAudio". After a few seconds check the tooltip again; it should say "spot's PA started". Click the icon to open Volume Control. If everything works as intended you should be staring at the Volume Control window and your audio equipment should be listed in the Output and Input
tabs.

Now start firefox:

firefox-pulseaudio-spot &

Go to a streaming site, youtube or whatever, and play some audio content. You should hear sound coming from your speakers and the Playback tab of Volume Control should show a "Firefox audioStream" live meter as picture below.

How to stop Firefox PulseAudio as spot

After exiting Firefox you can quit PulseAudio and its application test set as usual from the PA tray icon. Right-click it and select from the menu:

• Exit apps
• Stop PulseAudio
• Exit

Uninstalling Firefox

After uninstalling Firefox from your system via the package manager, you can remove the launch scripts by running the following command in a terminal window:

rm /usr/bin/firefox-pulseaudio* /usr/share/applications/firefox-pulseaudio.desktop

Streaming application audio over LAN

last edit 2020-12-10

Streaming audio between two computers in your LAN is easy. You need: an application that can play sound via PulseAudio; two PCs connected to your home LAN—let's say a desktop PC, further referred to as "Fat", and a notebook, "Slim", both equipped with working PulseAudio.

╭─ Fat upstairs ─╮         ╭── Slim downstairs ───────╮
│ playing sound  │─▶ LAN ─▶│ listening to Fat's sound │
╰────────────────╯         ╰──────────────────────────╯

Motivation

As a regular VNC user, I actually need audio streaming. Fat, my desktop PC, runs all heavy applications and Internet access. Slim, my notebook, views and controls Fat's screen via VNC. However, while video can transfer easily via VNC, audio cannot. Now with PulseAudio I can transfer Fat's sound to Slim. There are ways to transfer sound over LAN with ALSA alone, but I never bothered to learn.

Overview

The gist of transferring audio is for Slim to publish its local audio output—the sink—so that Fat—the source—can send the sink sound:

• Publishing takes place via "avahi zeroconf", which is built into Fatdog64 and can be enabled in PulseAudio by loading the zeroconf module.
• Sending audio takes place via "TCP network transfer", which can be enabled in PulseAudio by loading the native protocol TCP module.

Fat:                                     Slim:
load-module module-zeroconf discover ◀─  load-module module-zeroconf publish
env PULSE_SERVER=Slim "sound-player"  ─▶ load-module module-native-protocol-tcp

That's it, really. It can be described tersely because PulseAudio is a client/server architecture by design. Its server component is called "Pulse". Pulse runs on every PC where PulseAudio is installed. In this application, Fat's and Slim's Pulses set up the audio transfer between themselves of behalf of Fat's "sound-player" application and Slim's sound speakers.

Note for VNC users: here the client/server roles are reversed with regard to audio/video transfers: Fat's VNC is a server, and "source-player" is a PulseAudio client; Slim's viewer is a VNC client, and Slim's Pulse is the audio server.

Set up Slim

Edit /etc/pulse/default.pa and add after the last line:

load-module module-zeroconf-publish
load-module module-native-protocol-tcp auth-anonymous=1
#load-module module-native-protocol-tcp auth-ip-acl=127.0.0.1;192.168.0.0/24

The first line makes Slim's sound server visible to other PC's in your LAN. Specifically, avahi-aware applications can search and find Fat's sound resources, in terms of sinks and sources. The next two lines are mutually exclusive. The auth- piece defines which computers are allowed to connect to Slim: anonymous=1 allows connections from any PC while auth-ip-acl only allows PCs inside your home router's jurisdiction, your home LAN. You need to change 192.168.0.0/24 to the actual IP subnet of your LAN if you are going to use auth-ip-acl. As is the text above enables anonymous access because # is the comment character.

To complete this setup start/restart Slim's PulseAudio via the pa-tray icon.

Set up Fat

Edit /etc/pulse/default.pa and add after the last line:

load-module module-zeroconf-discover

This makes Fat's Pulse search for Slim's Pulse. Results will be visible in Fat's Volume Control application (pavucontrol) in terms of additional sinks and sources.

To complete this setup exit all PulseAudio applications via the pa-tray icon, stop Fat's PulseAudio and exit the pa-tray icon.

Streaming youtube's audio from Fat to Slim

You already know how to run Firefox via PulseAudio on the local computer. Fat's user spot runs Firefox locally, which connects to Slim's Pulse. In a terminal window type the following commands:

./install-pulseaudio.sh && cp pa-tray.sh /tmp && run-as-spot /tmp/pa-tray.sh &

Now start PulseAudio from the pa-tray icon then in a terminal window run[:1]:

env PULSE_SERVER=Slim firefox-pulseaudio-spot &

[:1] you need to change "Slim" with the actual IP address or hostname of "your Slim" computer.

env PULSE_SERVER=Slim tells Firefox to connect to Slim's Pulse, whose default sink and sources are Slim's own. Therefore, when Firefox plays youtube, sound can be heard on Slim's default sink, most likely named "Built-in Audio Analog Stereo", comprising its output "ports": Speakers, Line-in and Headphones.

Now open the Output tab of Volume Control: you will see that at least two sinks are available: "Built-in Audio Analog Stereo" and "Built-in Audio Analog Stereo on root@Slim". The latter exists because the zeroconf discovery module is loaded. In this dialog you can set either sink to be the default output sink for all applications.

Cheatsheet

Run on Slim to start/restart root's PulseAudio:

./pa-tray.sh --stop-pulseaudio
./pa-tray.sh --start-pulseaudio

Run on Fat to start/restart spot's PulseAudio:

./install-pulseaudio.sh # once only as root
cp ./pa-tray.sh /tmp
run-as-spot /tmp/pa-tray.sh --stop-pulseaudio
run-as-spot /tmp/pa-tray.sh --start-pulseaudio

Run on Fat to find if Slim's PulseAudio is running (this command prints the hostnames of all published Pulses in your LAN):

avahi-browse -at | grep -Fv `hostname` | grep -F _pulse-source._tcp | cut -d: -f1 | cut -d@ -f2 | uniq

Run on Fat to start Firefox with audio redirected to Slim:

env PULSE_SERVER=$(avahi-browse -at | grep -Fv `hostname` | grep -Fm1 _pulse-source._tcp | cut -d: -f1 | cut -d@ -f2) firefox-pulseaudio-spot &

Further reading

This guide on ArchWiki

Separately downloaded all of the files/sfs's (quite a few of them, might be nice to also have a single sfs of the combined set)

I then used the pulse forwarding example as per the last posting in this post/thread (that starts with "The below example shows how to create a pulseaudio TCP tunnel to forward sound from computer alpha to computer beta ...")

https://askubuntu.com/questions/70556/h ... er-the-lan

that does not use Avahi Zero-configuration networking.

In my case alpha is my desktop PC system booted into OpenBSD, chrome running and playing a video with pulse audio as the audio server. Laptop (beta) running Fatdog with the pulseaudio sfs's/scripts loaded/running.

Image is for pavucontrol on my Fatdog/laptop i.e. it sees the OpenBSD servers pulseaudio server

(note that when installing openbsd I usually assign a name of obsd and a domain of .org ... and create a userid of spot, hence the OpenBSD boxes pulseaudio shows as being ... spot@obsd.org)

Not actually hearing any sound though, but that's because I've been messing around with the OpenBSD's sndio recently and pretty much crippled it. i.e. I've been meaning to reinstall the OpenBSD anyway to get sound back up and working again, but thought I'd give the Fatdog pulseaudio a trial run first.

-- moved -- to viewtopic.php?p=12653#p12653

rufwoof wrote: Tue Dec 15, 2020 11:18 am

Not actually hearing any sound though, but that's because I've been messing around with the OpenBSD's sndio recently and pretty much crippled it. i.e. I've been meaning to reinstall the OpenBSD anyway to get sound back up and working again, but thought I'd give the Fatdog pulseaudio a trial run first.

@rufwoof, thank you for engaging in this thread! On your laptop (beta) running Fatdog64 can you hear sound emitted from a local PulseAudio application, such as paplay? If so, then the faulty element could be your OpenBSD sound configuration, as you suspect, or the tunnel setup (I haven't used the module-tunnel module that the Ubuntu post suggests).

No sound locally, i.e. if I pacat /usr/share/sounds/bell.wav ... then I do see the 'silence' bar in pavucontrol pulse around (slider appearance), but no audible sounds heard.

I'm not really that familiar with pulseaudio, so likely a 'user configuration issue'.

Can XRDP (Fatdog repository) carry BOTH the video & audio on its 3389 port streams?

It offers VNC console and audio in the session.

Thus, if you have a Youtube browser active on FAT (the "server") your can both see and hear what is playing at the Slim (the client).
Thus, if you have a multimedia file playing in any desktop on the server, you'll get the full experience on the client as well.

Rephrased: Everything that you would get when sitting at the server's console would be EXACTLY the same when client is running the RDP session with the server.

The beauty is a single product XRDP providing a compressed full desktop experience at clients running connectons to the server. VNC has been a great product for mere desktop viewing for years. But it has "stated" that it has NO intention for change to support audio. So, a full desktop connection in this decade is needed!!! MS has had the full desktop remote experience for 25 years with RDP in their 1995 Windows.

I think I saw a thread where @jamesbond set XRDP up somewhere, but I cannot find it.

@rufwoof There is a learning curve, I know, me too. Try arecord to create a test file then play that with aplay. To exit arecord just hit Ctrl+C, your test file will still be created.

@Clarity Please stay on topic. This thread concerns a public test of PulseAudio for Fatdog64-811. Thank you for your understanding.

Thanks @step. I mistakenly thought that the topic was shifting from local sound management to remote sound direction from a server login desktop to an accessing client user.

My comments was about a facility that makes use of the desktop session in such a way as all that is running on the FATDOG server could be experienced at the client as if the client was at the server. That includes pulseaudio operation with any app, including browsers, that one enjoys at the server console to be experienced at the client PC.

My apology, and I agree, that bit of info expands on the topic and would/could be confusing.

step wrote: Thu Dec 10, 2020 8:12 pm

There are ways to transfer sound over LAN with ALSA alone, but I never bothered to learn.

The one built-in to Fatdog is called rsound: https://github.com/Themaister/RSound but it is halfway broken, so I won't even explain how to use that. The other possibility is to use jack, but I've never tested that so I can't give you the instructions yet.

rufwoof wrote: Tue Dec 15, 2020 2:41 pm

Using sndio/sndiod, basically involves creating a sndiod userid and compiling the sndio package

Never heard of sndio ... Oh, it's BSD stuff, that's why.

adding that in run-as-spot ...etc. but no joy. Thinking that perhaps something like pcm.!default ... in .asoundrc needs to be set to point sound to sndio rather than alsa ???

You can't do that unless sndio comes with "alsa compatibility package". Rsound does have it, but as I said, it's halfway broken (as it, it works, but not fully, and not all the time, and not for all audio apps).

I'd better end it before it pollutes the thread even more and get schooled by @step I just want to say that actually I am quite interested to test this (curiosity reasons if nothing else), I have downloaded the SFS for weeks but it's sitting in my download folder collecting dust ... too many things to do, too little time ... ces't la vie.

jamesbond wrote: Thu Dec 17, 2020 4:53 pm

I have downloaded the SFS for weeks but it's sitting in my download folder collecting dust ... too many things to do, too little time ... ces't la vie.

No worries, we'll get there. When you find the time fo this project check the folder for updates. I don't think I updated the SFS files but definitely the scripts are newer.

-- moved -- to viewtopic.php?f=60&t=1664

To the risk of being unpopular, I'm determined to keep this thread on topic. Let me be clear by repeating myself from post #2:

Your post should follow some guidelines:

Applications must run on a vanilla Fatdog64-811 without a savebox
Include all instructions necessary to install and set up the application running and its dependencies
Include download links
If an application works partially, state what works and what doesn't.
Do not post about applications that do not work, or only work a little.
In the interest of quality, posts that do not follow the above guidelines will be flagged.

To avoid duplicated efforts, post a short message to let other people know which application you are testing.

Applications that are reported to work will be added to the list in [post #2].

I want for this thread to be focused and on topic mainly because I think that pulseaudio's aura of mistery, or stigma, can abate if we can show working results. Talks about what could be are fun but can lead to confusion, and confusion adds to the mistery/stigma.

I have module-rtp in my personal todo list for this thread, as an alternative to module-tcp, which I tested in this post. Rtp is interesting because it's multicast. I don't know when I'll have time to set up a test environment for module-rtp. If anyone wants to take the ball, please do!

PS Nothing personal, no bad feelings. We're playing here.

Thanks step

I am going to try this out.

stemsee

jamesbond wrote: Thu Dec 17, 2020 4:53 pm

step wrote: Thu Dec 10, 2020 8:12 pm

There are ways to transfer sound over LAN with ALSA alone, but I never bothered to learn.

The one built-in to Fatdog is called rsound: https://github.com/Themaister/RSound but it is halfway broken, so I won't even explain how to use that. The other possibility is to use jack, but I've never tested that so I can't give you the instructions yet.

@step
Sorry about posting off-topic...

@jamesbond
Did you ever look at this one?:
http://www.pogo.org.uk/~mark/trx/stream ... audio.html

Back on-topic, some of my PA deliberations might prove useful:
I briefly tried PulseAudio in EasyOS, had it working:

https://bkhome.org/news/202010/pulseaud ... asyos.html

More links on my blog:
https://bkhome.org/news/202010/problem- ... audio.html
https://bkhome.org/news/202010/easyos-d ... audio.html

Perhaps I should have persisted, but went back to plain ALSA. I know that I have a bit of a negative bias concerning PA, but the main motivation to consider PA was support for bluetooth audio, but my BluePup manager is handling audio just fine. Was having difficulty with configuring PA, so, I just went back to what I am familiar with.

A negative comment, but that is just my personal thing, and PA does bring some things to the table that might make it worthwhile persisting with getting it running nicely.

@BarryK thanks for joining in, if even for a quick visit!

These are my notes about some of the links you mentioned.

Your blog post mentions solving some ALSA/PulseAudio compatibility problems with Arch's pulseaudio-alsa package, which consists in a pulseaudio configuration file pulseaudio-default.conf. This configuration file seems to default ALSA output to the PulseAudio server.

• Could this configuration file help us overcome the shortage of PulseAudio-ready applications we face for our testing effort--cf. "Helping" in post #2? Note: our PulseAudio server isn't configured to start on system boot. You will need to do that before ALSA output can work with this configuration file.

• Barry settled on something different for EasyOS, a method to make ALSA and PulseAudio peacefully coexist. See this page.

As regards Bluetooth support, Fatdog64 comes with bluez-alsa, the ALSA background for Bluetooth, pre-installed and working. If anyone feels particularly interested in testing PulseAudio with Bluez5 (not me) keep in mind that bluez-alsa will conflict with Bluez5's built-in support for PulseAudio, and you will need to rebuild / reinstall / reconfigure your software appropriately.

.

.

step wrote: Sun Dec 20, 2020 9:29 am

@BarryK thanks for joining in, if even for a quick visit!

These are my notes about some of the links you mentioned.

Your blog post mentions solving some ALSA/PulseAudio compatibility problems with Arch's pulseaudio-alsa package, which consists in a pulseaudio configuration file pulseaudio-default.conf. This configuration file seems to default ALSA output to the PulseAudio server.

• Could this configuration file help us overcome the shortage of PulseAudio-ready applications we face for our testing effort--cf. "Helping" in post #2? Note: our PulseAudio server isn't configured to start on system boot. You will need to do that before ALSA output can work with this configuration file.

• Barry settled on something different for EasyOS, a method to make ALSA and PulseAudio peacefully coexist. See this page.

As regards Bluetooth support, Fatdog64 comes with bluez-alsa, the ALSA background for Bluetooth, pre-installed and working. If anyone feels particularly interested in testing PulseAudio with Bluez5 (not me) keep in mind that bluez-alsa will conflict with Bluez5's built-in support for PulseAudio, and you will need to rebuild / reinstall / reconfigure your software appropriately.

An extra clarification: I abandoned that PA and ALSA coexistence method. Now no PA at all in EasyOS, using bluez-alsa and apulse, like Fatdog.

@step @jamesbond
An update, I am back onto using Pulseaudio.

A significant difference from Puppy is that I am now running apps that access the Internet each as its own user and group. So firefox runs as user 'firefox' and group 'firefox, located in /home/firefox

So things will bog down if run an instance of pa in each of these. So there is just the one instance of the pa daemon, that users can access via a socket.

It works, see blog post:

https://bkhome.org/news/202111/disable- ... users.html

I am now going to experiment using TCP instead of Unix Domain Sockets.

I am also a pa newbie! The reason that I am looking at it again, is it will be a stepping stone to pipewire. Though, the more I look at pipewire, I am wondering if it is the "systemd of multimedia"!

Oh well, have gone down the rabbit hole, might as well keep going...

I have to agree with this direction that Barry is pursuing. He is taking LInux multimedia (sound+microphone+video) "all the way" to an EASY operations immersive experience.

This benefit is that in our future, we will manage all of these subsystem vehicles in a visual manner with enormously better and simpler utilities to "see and control" what is happening. In theory, it will reduce developer assistance to anyone with multimedia problems going forward.

Puppy LInux distros are maturing "rapidly". Thanks to this (and other) efforts as this project and the community bullets forward in the 21st Century along with the physical miniaturizations that are continuing to emerge.

The TCP method for pulseaudio, for multiple users and also in containers, is very simple:

https://bkhome.org/news/202111/pulseaud ... h-tcp.html

But, wish there was some way to have individual enable/disable.

BarryK wrote:

, the more I look at pipewire, I am wondering if it is the "systemd of multimedia"!

I too was looking at pipewire! Seemed to be very promising. There was another closely related project/fork/spinoff ... can't recall it just now!

Keep us posted on your progress!

I also have a question regarding bmixd and default soundcard script. If I want to connect directly to a bt mac soundcard fatdog doesn't play along unless I go through the default soundcard wizard. My app outputs a .asoundrc file, but this doesn't take effect. There must be a quick way to change the device by replacing the mac address!

Also bmixd .... is it that first you set the default bluetooth soundcard then run again selecting the bmixd?

regards
stemsee

@BarryK - we just went the other around, using ALSA sound server instead of pulseaudio. Perhaps not the most future-proof, but right now it works for us (for me specifically). But that's as far as I want to say here as I don't want to pollute this thread. Let this thread focus on PA only and we can discuss it elsewhere.

@stemsee, for the same reason, I will respond to your question, viewtopic.php?t=4398