Configure ST's input to be wherever Ots is sending its output, then set ST's output to where you want it, which must obviously be different from the input.
From your screenshot, you don't seem to have enough independant audio connections so you'll probably need to add another sound card or at least another driver such as a virtual audio cable.
If you use a virtual cable, point Ots at its input, and ST's input to the cable's output. Then you can point ST's output to your speakers.
Okay, here it might get a little tricky. Normally I run Stereo Tool as a VST plugin or as a microphone processor. Sadly it doesn't appear
that OTS is capable of audio plugins at the moment(?) But I've only tinkered with the trial. (EDIT: However I see a Winamp folder and a Plugins subfolder in the installation directory)
I'm not experienced with most virtual audio cables, such as VAC. However I think the Breakaway Pipeline tool
should do the job, since it works for me when I need to route audio to or from other applications. I can't remember how exactly the initial setup goes, and I'm unable to upload screenshots in this subforum.
If I'm not mistaken, it should install 1 or 2 pipelines (aka cables or virtual audio devices). You then can open Stereo Tool and select the Input Device ID as "Line 1 (Breakaway Pipeline 1)". I'm not certain which interface method is best, but I use Kernal Streaming.
Then for your output, you would probably want to choose Normal Output and select your listening device, or what ever your streaming application is looking for. Then you can adjust your "Buffer size / Diversity Delay" to something more comfortable if you need lower latency. I believe the default is one second.
EDIT2: After more tinkering, it appears to accept the 32 bit Winamp DSP plugin as a Shoutcast plugin. Sadly it doesn't appear to affect the output audio (???). So this would be handy if you wanted to apply Stereo Tool to your streaming output, I take it. So yeah, try the pipeline method.