Building Digital Human - Part I

Problem Statement:

While current chatbot services undeniably offer quick responses to customers, they often leave users feeling like they're engaging with soulless machines devoid of any human touch. These chatbots struggle to foster genuine conversations or recall past interactions, resulting in answers that fall far short of user expectations. Their inability to discern the context of questions leads to responses that miss the mark, and their brevity can be attributed to rigid rule-based structures inherent in chatbot services. In essence, while they excel in speed, the human element remains conspicuously absent in these interactions

Proposed Solution:

Communication in plain text lacks the nuance of emotion, so any effective solution should closely replicate human interaction. Users should experience a dialogue akin to speaking with a person, complete with emotional responses and facial expressions. This improved approach should enable users to interact through voice or text, retain context from prior conversations, comprehend user preferences like language and personal preferences, and deliver responses with utmost politeness. In essence, the proposed solution should bridge the gap between human and chatbot interactions, making the experience seamless and emotionally engaging.

Challenges:

Technically, it is challenging to provide a solution that aligns with the proposed solution. The problem can be divided into five parts:

• Human Look – Provide a face to conversation

We need an avatar, which can be a unique human face tailored to the brand's identity. This will give the user a sense that they are interacting with another human while reinforcing the brand's image. The avatar size can be adjusted to display only the head portion, half body, or the full body, depending on the available space.

• Ease of Conversation – Accept inputs via voice or text

Users should have the capability to speak to the avatar. Such solutions should function seamlessly in both web browsers and on mobile devices, ensuring a smooth and effortless conversation experience for the user.

• Expression Builder – Show human like expressions on Avatar face

The avatar's face must vividly convey human-like expressions, encompassing a range of emotions such as joy, surprise, and empathy. These expressions should include subtle changes in eye movements, like blinking or widening, nuanced head nods, not just for agreement but also for attentiveness, and natural hand movements that emphasize points or convey gestures, all working together to enhance the interactive and lifelike experience.

• Intelligence and Context Building – Give a memory to Avatar

The avatar must retain previous conversations with the user and utilize this history to create a memory of their discussions. It should possess the capability to comprehend and connect user queries to the context established in prior interactions. Users and avatars can engage in discussions on various topics, and transitioning between these topics should be seamless and effortless.

• Business Process Integration – Talk business and much more

The avatar should have the capacity to respond to business-related inquiries, but it should not be limited solely to these queries. It should also be able to engage in conversations with the user on various non-business topics, including subjects like weather, politics, and more. This versatility will make the interaction more engaging and user-friendly.

Technical Stack:

• Utilizing WebRTC technology, we can effortlessly capture user audio via the connected microphone on a computer, without the need for any additional software. This functionality can be seamlessly integrated within WebRTC-compliant browsers. The captured voice data can then be transmitted to a Natural Language Understanding (NLU) service, which can effectively interpret and comprehend the user's questions or input.
• The conversion of audio into text enables us to discern the user's question or input accurately. This textual representation must then undergo further processing to formulate an appropriate reply or response.
• We can leverage WebGL technology to render expressions on the avatar. Depending on the response to a question, we can dynamically control the avatar's movements, lips, and facial expressions to create a more interactive and emotionally engaging conversation experience.
• To endow the Avatar with memory and expand its knowledge base, we have the option of employing custom-built AI models or utilizing publicly available AI models and services. To facilitate context building, pivotal conversation points from previous interactions can be stored in a database, enabling the Avatar to draw upon this information for more coherent and informed responses in subsequent conversations.
• For business process integration, we can use existing chatbot technologies like Dialog flow, RASA etc.

Interested in building Digital Human? Contact SpringCT:

At SpringCT, we've created a remarkable Digital Human with a lifelike human face, capable of engaging users in conversations on a wide array of topics and delivering intelligent responses. When combined with Chatbot functionality, this Digital Human not only excels in addressing business-related queries but does so with a genuine human touch. What sets it apart is its ability to convey answers accompanied by incredibly realistic facial expressions, to the extent that interacting with it feels like conversing with a fellow human being.

Our cutting-edge solution is driven by the power of Artificial Intelligence (AI), Natural Language Processing (NLU), and WebRTC technologies.

SpringCT stands as a renowned outsourced product development company, headquartered in Pune, India, with a rich history of over 15 years in the Unified Communications (UC) arena. As pioneers in adopting WebRTC technology, we see Chatbots and Digital Humans as natural progressions within the UC domain.

The Digital Human solution holds immense potential to revolutionize customer-agent interactions across diverse sectors. Its applications are limitless, spanning industries of all kinds.

If the Digital Human concept intrigues you and you're keen to develop a bespoke Digital Human for your business, don't hesitate to reach out to us. We're eager to explore how our technology can benefit your specific needs.

Contact us at:

https://springct.com/webRTC/index.html#contactus.