Confidential Internal Document – Comparative Technology Report
For privacy purposes we have been authorized to present this confidential report as long as we don’t identify the name of the company that prepared it. Therefore, in order to protect their anonymity we have identified them as “Company X” in keeping with that agreement. Company X operates as a leading global security and aerospace company, which engages in the research, design, development, manufacture, integration, and sustainment of technology systems, products, and services. It operates through the following business segments: Aerospace Systems, Innovation Systems, Mission Systems, and Technology Services.
Trade Comparison of Licenses, Unity Avatar Support, Natural Language Processing, and AI Platforms
“The current needs of the organization’s design and production team, as it pertains to NLP, Avatars and AI in 3D accelerated applications, include:
- Non- monotonic reasoning for natural language AI
- Isolated or no network operation (operates without an internet connection)
- Text to speech phoneme marker parsing (Avatar speech systems related – Viseme Tool)
- Phoneme to viseme facial animation mapping (Avatar speech systems related – Viseme Tool)
- Viseme animation blending (Avatar speech systems related – Viseme Tool for Unity)
- Integrated support for Unity
Below is a trade comparison of popular natural language processing applications and how they relate to our development needs for our initiative.
IBM Watson
The IBM WATSON product is a natural language processing application that relies on connectivity to the internet to calculate use cost. The cost is based on something IBM calls “Natural Language Understanding” or NLU units, and requires a subscription and internet connection to be calculated. This means all language queries are collated and calculated on an IBM server, located on the public WAN. There is no support for speech. There is support for Unity, our primary 3D training development platform for ART, but no support for avatars in the virtual 3D space. This support would have to be developed by the ART Team.
Microsoft Bot Framework
The Microsoft Bot Framework product is a natural language processing application that relies on connectivity to the internet to calculate use cost. The cost is based on per word calculations, and requires a subscription and internet connection to be calculated. This means all language queries are collated and calculated on a Microsoft server, located on the public WAN. There is support for speech. There is no support for Unity, our primary 3D training development platform for ART, and no support for avatars in the virtual 3D space. This support would have to be developed by the ART Team.
Wit.ai
The Wit product is a natural language processing application that relies on connectivity to the internet to compute language and meaning exchanges. The cost is free. This means all language queries are collated and calculated on a Wit server, located on the public WAN. There is support for speech. There is no support for Unity, our primary 3D training development platform for ART, and no support for avatars in the virtual 3D space. This support would have to be developed by the ART Team.
Google Cloud Service Dialogflow
The Dialogflow product is a natural language processing application that relies on connectivity to the internet to calculate use cost. The cost is based on per query calculations, and requires a subscription and internet connection to be calculated. This means all language queries are collated and calculated on a Google server, located on the public WAN. There is support for VOIP. There is no support for Unity, our primary 3D training development platform for ART, and no support for avatars in the virtual 3D space. This support would have to be developed by the ART Team.
ChatterBot
The ChatterBot product is an open source natural language processing application that does not rely on internet connectivity to function. The cost is free based on BSD licensing. There is no support for speech. There is no support for Unity, our primary 3D training development platform for ART, and no support for avatars in the virtual 3D space. This support would have to be developed by the ART Team.
Cognitive Code – SILVIA
The SILVIA product is a natural language processing application that does not rely on connectivity to the internet. The cost is based on an annual license structure negotiated exclusively for our company by the senior IP counsel, and, Cognitive Code. This means all language queries can be confined to local computers and networks, and completely controlled. There is support for speech through an API which allows the use of any speech recognition software or speech solutions. There is support for Unity, our primary 3D training development platform for ART, and there is support for avatars in the virtual 3D space.
Conclusion
Cognitive Code Silvia offers the most complete support for the combined effort of Unity, avatar, and AI development for WISLA ART. Cognitive Code Silvia provides complete integration with our Unity development pipeline, and full support for real- time avatar speech and lip-syncing. It further supports our need to run on isolated systems, with no internet connectivity, and subject matter can be programmed visually through its UI.”