GML Gesture Markup Language

This paper discusses emerging problems of VHML – the Virtual Human Markup Language. VHML is XML based and allows interactive Virtual Humans to be directed so that human-virtual human interaction is more effective. The language addresses issues related to Facial and Body Animation, Dialogue Management, Text to Speech production, Emotional Representation plus Hyper / Multi Media information. Variants of VHML are being used and evaluated in several Talking Head applications as well as a Mentor System. VHML is an evolving language that aims for standardisation. To meet the rigorous criteria for this, VHML must be open and frank in its (self) evaluation, it must meets the needs of its targeted audience, it must be easy to learn, use and extend, etc. Peer discussions about VHML as well as a first implementation of the language from the specification have revealed problems, inconsistencies and deficiencies that need to be addressed before VHML can be seen as a stable markup language for wide spread use. This paper details some of these problems so as to promote open discussion. It also indicates the current state of VHML at specification level 0.4. The VHML development and implementation is part of a 3 year EU 5 Framework project called InterFace.