Yes, the extremely short audio buffers in HamSphere (< 30ms) can make small gaps in audio if you have a little bigger ping times to the server. It is also a result of Internet congestions as well as the overall load of the system.
This is not really affecting speech that much. But CW however is sometimes a bit trickier, but usually OK to copy. SSTV on the other hand is definately useless. SSTV is totally synchronized, and the smallest delay will create a picture shift.
This is probably a crazy feature request but I'll put it out there anyway :)
A user-defined length buffer for CW would be super handy!
The server itself wouldn't need to change buffer sizes, but the client, during CW mode, could perhaps delay playback by maybe 500ms, and buffer a number of packets before playing back. Something like this might reduce the occurrence of breaks in the stream.
I agree with this one. The Hamsphere breaks are unlike copying CW under difficult QRN conditions on the air, where your ear and brain can work out the underlying CW. In this case, the breaks change the Morse symbol being sent; there's nothing that your brain can fix. The result is garbled copy.
If a buffer length change would fix this problem, it would certainly be a help. Right now, CW on Hamsphere is really clumsy, and it really takes the fun out of it. If it could be smoothed out, I'd surely appreciate it, since I really enjoy CW operation.
Another solution, which is purely cheating, would be to send the text to the server, and have the server send it to the other clients, who then generate the audio at the client side. So there is no stream at all :P