Teamspeak is a voice communication software, which is intended for using it parallel to online games. So it is developed to have a small cpu usage and to be used with low bandwidth. As listening to music while playing is common there is a new high quality audio codec (coming with TS 3.0) called Opus Music which allows enabling a higher bandwidth for music streaming. Therefore TS might be interesting for live performing music over the internet.
The software consists of two parts: a client software with GUIGraphical User Interface and a console based server software. Communication partners have to connect to a server to start talking. If you are connected to a server you can create channels, give them names, protect them with passwords and configure the audio settings for them. As there can be many users on one channel at the same time TS offers the possibility for the users to place the other users in a 3D map helping to locate the different voice messages.
TS offers three control settings for the input signal: push to talk, voice based activation and constant streaming.
You can also activate echo reducing and echo damping, noise reduction and automatic signal normalization, which are affecting the resulting signal.
If the server settings are supporting there are are five different codecs which can be chosen in ten different quality levels:
I wrote a Pd[[Pure Data]] a dataflow programming environment patch to test the distortion of the spectrum and the delay depending on the chosen codec and the settings for the input signal - you can download it here: Media:Test communication quality.pd (it might be useful for other communication software as well). To make use of it you need a audio routing software (jack or soundflower) and route the output of your communication software to the 3rd inlet of Pd[[Pure Data]] a dataflow programming environment and the inlet of your communication software to the 3rd outlet of Pd[[Pure Data]] a dataflow programming environment. The patch offers to functions: spectrum analysis and delay analysis.
As Jack and me have not found any time to work together I did the research on my own. So I have chosen a Teamspeak Server in California (to simulate the delay) and connected to it with two clients. In Jack I routed the output of the first client to the input of Pd[[Pure Data]] a dataflow programming environment and the output of Pd[[Pure Data]] a dataflow programming environment to the input of the second client (as described above).
These are the results of the spectrum and delay analysis:
median roundtrip latency: 330.8 ms
median roundtrip latency: 332.7 ms
median roundtrip latency: 340.0 ms
median roundtrip latency: 352.5 ms
median roundtrip latency: 323.8 ms
median roundtrip latency: 385.3 ms
This is short feedback improvisation using the testing sound of teamspeak. Depending on the current sound I activated and deactivated the possible input flags, adjusted the feedback strength and the current bandwidth (by opening loads of vimeo hd videos). The result is a rhythmic, slightly changing lofi sound.