2 weeks
Architecture Design
2 weeks were spent designing a scalable and robust architecture for the consultant, ensuring it could handle a large number of users and integrate seamlessly with the client’s existing systems.
Summary
Many individuals in Asia and around the world experience profound feelings of isolation.
Our customer is creating a conversational agent designed to serve as a companion to users.
This dialog agent can retain user details and recognize emotions in both the user's speech and text content.
DRL has leveraged cutting-edge NLP technologies to develop a full-fledged Dialog Agent compatible with Alexa or Google devices (currently accessible through closed beta).
Business Impact
Project is in the active development phase, and currently is purely technical without any business part.
Tech challenges
The system must be capable of simulating various recognized voices, including that of Scarlett Johansson.
Voice generation should encompass 15 distinct emotions and levels of intensity.
The voice must be generated in less than 1/100th of a second, regardless of the audio duration.
The system should engage in conversations with users, demonstrate common sense, retain a vast amount of information, and function offline (distinct from Google Assistant).
The system is expected to successfully pass the Turing Test (Imitation game).
Timelines
1
2
3 weeks
Data Integration
Over 3 weeks, various data sources were integrated, including conversational datasets, emotional databases, and user preferences, to provide the AI with comprehensive data for accurate and personalized interactions.
3
4 weeks
Backend Development
4 weeks were spent developing the backend, focusing on the application’s core functionalities, such as natural language processing (NLP) and machine learning algorithms.
4
2 weeks
UI Design
The UI design phase took 2 weeks, during which an intuitive and user-friendly interface for the virtual girlfriend was created.
5
1 week
Integration
Integrating the application with the client’s existing platforms took 1 week
Case Study