Eliminating a single point of failure scenario involves implementing redundancy in all system components, which implies as a minimum a duplication of all individual systems and ensuring that any telephony switches routing calls into the IVR will fail over automatically upon failure detection.
As implementing failover detection in Telco telephony switches that deliver analog services is not usually done for smaller deployments like this one. The simplest approach for analog systems is to have two systems actively running, taking calls in a "round-robin" fashion - with alternating calls sent to different systems. This way the failure of one system results in service degradation (with not all calls being answered) but the system continues to operate.
Also,any services used by the IVR would need to have similar design considerations, with clustering used for any databases etc.
It is not possible to use telephony cards on a visualized system. Two separate complete systems must be deployed, each with its own telephony card and its own license.
Having said all that, VoiceGuide IVR deployments regularly achieve 99.99% uptime or more, with many systems never requiring any downtime year round - so achieving 99.999% +, so you may want to consider whether in your case it would be necessary to still pursue deploying a "no single point of failure" system.
(systems that achieve 'five-nines' 99.999% uptime are usually disconnected from internet and have automatic Windows updating and restarts disabled)
Please let us know if you have any questions,