[Discuss] suggestions for low latency Internet audio conferencing software suitable for performing music?

[Bill Ricker]

On Wed, Apr 1, 2020 at 7:35 PM Bill Bogstad <bogstad at pobox.com> wrote:

> This model works
> very poorly for people trying to play/sing music together due to
> latency.


True.

> Obviously software can't eliminate the latency caused by
> the network itself, but if you are trying to do this in a metropolitan
> area among people who normally meet in person, it seems like it might
> be possible.


If they're all on the same ISP or there's a near-enough peering point
between the incumbent Cable Co, the incumbent Telco, and the insurgents.
(Which you probably are.)
But i've seen packets from Dorchester to Harvard U traceroute through
peering in NYC some days ...

Any suggestions?
>

The fellow BBC Trending or Digital Planet (I forget which) interviewed on
this topic didn't find one for realtime, so he's distributing the backing
track (beats or karaoke instrumental) and folks are individually
overdubbing (on 2nd channel) their line and sending the file back, and
he'll align and mix down.  It gets a resulting choral performance, but not
a sense of community presence.

Depending on hardware/software what kind of latency can you get from
> VOIP systems?
>

I don't know if it's going to be practical other than on true analog POTS
copper pair in a TouchTone™ enhanced calling services Conference Call in
close enough proximity that Long Lines doesn't digitally multiplex the
calls. Assuming the Central Office is still an analog crossbar and all
participants kept their copper pair for E911 despite moving up from ADSL to
FioS or Cable+Cell. (Even I cut the copper so.)

A full duplex analog radio link could work, but alas ham radio forbids
music as that is deemed broadcasting. I don't think typical GMRS gear can
full duplex. Police/ Fire dispatch consoles do FD,  but that's about it
aside from antique radiotelephones. DIY Part15 devices won't get you much
more range than singing from the balcony with the neighbors.

One problem is that for latency you want a fast codec, but for music you
want a hi fi codec.  Assuming arguendo there are acceptable codecs ...

For choral singing you'd need not only very low latency but full-duplex,
which would *require* headphones (or the most fabulous adaptive
feedback-cancelling speakerphones) to avoid feedback. (Headphones are of
course needed for the dubbing model also. And all this presumes better than
cellphone / laptop bulletin mic's unless you want that tinny 78rpm sound.)

Jitsi (which BLU and Natick FOSS will be using starting tomorrow,  and
possibly Boston PM as well) is lower latency than many because it doesn't
decode/mix/recode at the server; it multiplexes the audio bitstreams from
server out to all the clients which do their own mixing as they decode.
This lowers the CPU demand on the central server as well as lowering
digital delay aka latency. I don't recall if i could hear the other
participants while i was talking or not ... it might work, if you set up a
local server. And since it's FLOSS, you could. Maybe Jerry or I will think
to note during Natick FOSS "meeting" on Thursday (tomorrow) if we hear each
other when we talk over each other. (That never happens :-P.) Might or
might not be tight enough for barbershop choral singing,  but might be
enough for call-and-response in a chorus at least?

After the reports from the disastrous west coast choir rehearsal,  you are
wise to omit in-person singing.  Studies show singing projects both dry and
wet virus as far as a cough,  6' minimum,  and short lived dry virus
father. Shouldn't be a surprise as you sing from the diaphragm to project!

I guess one option would have been to drive to a vacant parking lot and
form a widdershins lagger (counterclockwise circle) and sing out open
driver side windows.  Not best posture. But the shelter in place has
progressed to where that is a non essential trip.


-- 
Bill Ricker
bill.n1vux at gmail.com
https://www.linkedin.com/in/n1vux