Hi Jesse & all.
I'd like to use Sonobus for an ensemble of some ten musicians, some of them having limited bandwidth internet-connections.
Latency is of importance. Minimizing latency pushes us in the direction of using PCM, to bypass any codecs and the inevitable associated algorithmic delay the use of any codec introduces.
However, 10+ incoming 16bit PCM connections is far too much for the internet-connection of some of the members of our ensemble; using the "limited bandwidth" options of Sonobus therefore seems a logical choice.
From what I understand, Sonobus has the Opus-codec "under the hood". This I believe uses CELP for the higher audio-bandwidths. From what I've read the frame size used by CELP determines the algorithmic delay, the delay introduced by the codec. And 20msec frames introducing 22.5 msec algorithmic delay are the default.
(This implying a 45msec algorithmic delay for a connection; compression at one end of the connection + decompression at the other end).
This 45msec algorithmic delay in combination with the inevitable transmission-path delay proves to be too much for the plucked-instrument players of our ensemble to be able to play in-sync.
Can anyone tell me what frame size Sonobus uses when invoking CELP?
Can this frame size (and thus the algorithmic delay involved) be configured in any way by the end-user?
Again, from what I've read usage of a 2.5msec frame size could reduce the algorithmic delay to as little as 5msec (at the cost of CPU I guess plus an increase of the bandwidth).
Any help/feedback appreciated!
Kind regards,
Christiaan