A Jamulus-Sonobus bridge without audio adapter?

[ve3meo]

Hi Kevin,
Is it possible via Patchage or Qjackctl to connect the i/o of the Jamulus client to the o/i of the Sonobus client in order to bridge between the two domains? If that's possible, I'd like to operate the Sonobus at 16kHz sample rate because the client does not work at 48kHz on many Android devices and to lower network traffic.

I know that there will be added delay between clients of one type and the other but it's the only way I can think of to bring online a few singers in my chorus who do have Internet but not a suitable computer nor the desire to spend money.

Thanks for your thoughts.

Tom

P.S. I started this on my XP that controls JamBox but its Chrome browser stuck at v49 apparently does not like the Category selection so I finished it on the JamBox but had to copy and paste through a Gmail draft because I couldn't paste directly through VNC :-)

[ve3meo]

I should add that if a solution does require a USB audio adapter, so be it. That might be essential for debugging and monitoring.

[kdoren]

@ve3meo That idea is is so crazy it just might work!
Unfortunately the delay may be too high to be useful. Jamulus is much less fun when delay > 50 ms, and this will be a lot higher.
You'll need wired headphones/earbuds, because bluetooth adds too much delay.
The RPi running the bridge will need a good internet connection, especially if there are multiple clients.
The setup is a bit tricky (SonoBus replaces the audio interface), but here's what it looks like (for mono setup to minimize bandwidth):

1, restart jack, launch Jamulus & SonoBus without using the start scripts:

sudo apt upgrade sonobus
sudo systemctl restart jack
SonoBus &   # "&" = send process to background
jamulus

In bridge Jamulus client:

1. connect to your jamulus server.
2. Settings (to minimize bandwitdh):

• Audio Channels: Mono
• Audio Quality: Normal

In bridge SonoBus client:

1. set "Monitor" all the way to the left ("-inf")
2. in Settings -> Audio:

• Audio Device Type: Jack
• Input: Jamulus
• Output: Jamulus
• Active Input Channels: Input Left
• Active Output Channels: Output Left
• Audio Buffer Size: 64 (or higher if needed for stability)

3. in "Options" set "Default Send Quality" to "16kbps/ch", "Change all connected"
4. "Connect" to a private group, name the group and your bridge client.

On smartphone SonoBus clients:

1. start with "Monitor" all the way to the left ("-inf").

• This will let you listen to the mix from the server, but delay will be very high.
• You can turn it all the way up to hear yourself sooner - see below - but then you will be off time.

2. in Settings -> Audio:

• Sample Rate: 48000 (only choice on iPhone)
• Audio Buffer Size: 64 (or higher if needed for stability)

3. in "Options" set "Default Send Quality" to "16kbps/ch", "Change all connected"
4. "Connect" to a private group, name the group and your bridge client.

Possible mitigation for long delay:

1. on bridge Jamulus client: Under the slider for SonoBus client, -> "Mute"

• This will take the SonoBus audio out of the mix coming back from Jamulus

2. on smartphone SonoBus clients, set "Monitor" slider all the way to the right

• This will cause smartphone user to hear themselves only direct.

3. But it will still be impossible to sing in correct time, they will be late.

FYI you can copy/paste with noVNC, but it's a 2-stop operation. You need to go through the clipboard that pulls out from the left side of your browser window.

Please report back your results.

[ve3meo]

Kevin, so crazy that it did work! I can say that this was an unmitigated SUCCESS! OK, maybe mitigated... I'll need to spend some more time with it and get someone not on my LAN onto SonoBus but I had a 2-way conversation with a Jamulus user via the Bridge to SonoBus on my Samsung phone for a few tens of minutes until the phone SB spontaneously disconnected. He said the sound quality was good, no discernible artefacts until shortly before the disconnect when we had bubbling. I heard a moderate clicking on the phone throughout.

I'll have to play some more with channel settings (so many options in SB!) and would like to go stereo. BTW, I also had a separate stereo bridge between Jamulus and Zoom on the Jamulus server at the same time but I didn't have Zoom running. My upstream bandwidth of 10Mbps could become hard pressed with all this activity.

Thanks for the clear instructions. Being a novice at Linux I was not sure what you meant by

SonoBus &
jamulus

The Terminal seemed to be waiting after starting SonoBus. I tried sudo systemctl restart SonoBus and that got an error message but this launched them both:
SonoBus & jamulus

I think there's a mistaken copy of some settings from the Bridge SonoBus client to the Phone SonoBus client under para 2.:

in Settings -> Audio:

• Audio Device Type: Jack
• Input: Jamulus
• Audio Buffer Size: 64
• Auto-Reconnect to Last Group

What I have is:

• Input: System Default
• Output: System Default
• Active Output Channels: Channels 1 & 2
• Sample Rate: 16000 Hz
• Audio Buffer Size: 192 samples (12.0 ms)

(that's the smallest on my phone but the delay diminishes with higher sampling frequency but the probability of a crash increases)

Thank you for guiding me so easily through what I expected to be a far more difficult proposition. If Jamulus apps for iOS and Android ever come out, there won't be any more need for the SonoJamBridge but until then I think it holds some promise worth exploring further.

Tom

[kdoren]

Congrats, happy to hear you got it working.
You're correct that I mistakenly copied some config setting (I have now edited the original post).

I am running iPhone client, which only offers sample rate 48000 Hz
If you phone can do it, I think it would be better to run sample rate > 16K, and use compression to lower bit rate (16kbps/ch).

In SonoBus settings, I was able to set audio input/output to "jamulus" directly, not "system", which I think is better.
FYI you can bring up patchage to see/change patches. I started to do that, but didn't need to.
You should update SonoBus if you haven't already: "sudo apt upgrade sonobus"

[ve3meo]

Ran a little test today with a friend in the same town:
.Me - Jamulus on my laptop on Cogeco Internet
..Jamulus server on a cloud host 110km away in Toronto
...Jamulus client on my JamBox on Cogeco Internet (same LAN)
...SonoBus client on same JamBox
..Him - SonoBus client on his laptop on Bell Fibe Internet

Because we are on different ISPs, our peer-peer connection is via the Toronto Internet Exchange so we saw SonoBus reporting an overall delay of ~60ms while Jamulus on the Pi reported ~40ms. Therefore his roundtrip latency was on the order of 100 ms. He could hear it on the foldback. We counted a tempo together and he started anticipating and we were in sync after about 4 bars.

So, yes, that's a lot of delay and is on the order that we would see between much more widely scattered Jamulus clients. But it may still be worthwhile to bring musicians together as if they were about 90 ft apart than just remaining solo.

[kdoren]

You might get a bit less delay by co-locating the sonobus - jamulus bridge with the jamulus server. That would give minimum delay/jitter between the jamulus bridge client and the jamulus server. You would want it to be on fiber internet though.

FYI, Cloud hosting for jamulus server is only as good as the server you are renting. I have one with dedicated CPUs that works consistently well, never a problem. I've run tests on less-expensive virtualized cloud servers that work well sometimes, and badly at others. And the issue is definitely the server ("noisy neighbor" problem) - I can do A/B comparison between server types which rules out client-side and other explanations.

Which suggests: if your friend has fiber internet, you might be better off with him running your jamulus server on Raspberry Pi at his house (it's a systemd service on jambox). Then run your bridge on a second raspberry pi sitting next to it (or maybe even same Pi to test).

[ve3meo]

If I could get a headless SonoBus and headless Jamulus client on my cloud virtual machine, that might be the ideal. Putting the JamBox at my friend's place as the bridge would be a definite advantage for all SonoBus clients on the same ISP and sub-net and possibly avoiding going through the Toronto Internet Exchange. The singers I'm aiming for are fairly scattered and on a mix of ISPs and even DSL and fiber on the same ISP. I think the best solution is that they get a JamBox but it may take a taste of the possible to get them to commit to it. Maybe my time would be better spent coming up with a loaner JamBox, audio adapter and disposable mic and earphones.

"Noisy neighbour" is a good metaphor. I've tended towards "rush hour" but that misses conveying the randomness and burstiness implicit in yours.