emoSyn (acronym for emotion-synthesizer) is a program that converts speechsignal-descriptions into parameter-values usable with a klatt-style formant synthesizer.
It's part of a project funded by the German Research Community (DFG) on emotional speech resp. my ph.d thesis about acoustical correlates of emotional speech.
The rules to express emotional arousal are mainly derived by analysis of a german emotional speech-database.
The phonetic knowledge comes for voiced sounds from a diphone-database and for unvoiced ones from rules. I don't believe that pure diphone-concatenation is the best choice for german formant-synthesis, but this concept is so widespread, that I use it as a starter.
It provides input for the freely available Iles & Simmons synthesizer and the commercial version from Sensimetrics .

examples of copy-synthesis with emoSyn can be found here.
examples of rule based synthesis with emoSyn can be found here.

Up to now there is no database, so I generated two samples of the german sentence "In sieben Stunden wird es soweit sein." ("In seven hours it will happen.") to test the sound of unrestricted synthesis: diphone-concatenation(34 kB wav-file), syllable-concatenation(36 kB wav-file)

All examples of emoSyn-output have a samplerate of 10 kHz.

The system was used in perception experiments. The article Verification of Acoustical Correlates of Emotional Speech using Formant-Synthesis can be downloaded here.