A voice processing complex has a plurality of host machines, each host
machine supporting one or more voice applications, at least one host in
the complex including telephony hardware for providing access to a
plurality of telephone lines. Such a host provides telephony functions to
the complex. This is achieved by maintaining a mapping of voice
applications to telephone lines, and creating a call object in response
to an incoming call on one of the lines. It is then determined which
voice application to associate with the incoming call on the basis of
this mapping, and an identifier to the call object is then passed to the
determined voice application. Thereafter the call object is responsive to
requests from this voice application for providing telephony functions
for the call. This includes receiving a media object from the voice
processing application which represents desired output over a telephone
line. This media object is then processed into a plurality of output
components in accordance with one or more presentation criteria, and
these components are then output over the telephone line.