The aim is to automate with audio recognition. The solution includes the construction of a system that converts voice to text, the intention classification system given the user's text, using artificial intelligence techniques, contemplated in the area of natural language processing. In addition, this system will be designed to integrate with commercial IoT devices, such as smart light bulbs, humidifiers, heaters, among other actuators, which have an open communication interface via a local network and the Internet. Finally, the students will version the documents using the Git tool and make the complete solution available on GitHub.
To support this system, students will have to design it using the concepts of scalability, load balancing, fault tolerance, service-oriented architecture, management and development methodologies, to solve a real problem in the context presented by the selected market partner.
APPLICATION EXAMPLES:
- Voice recognition in hands-free scenarios (situations in which the user has a preference for hands-free interactions, such as access control after shopping in a market, while driving a car, while cooking at home);
- Personalized chatbots (adaptive responses according to the user's feelings);
- Voice recognition and conversational interfaces for accessibility applications.
MANDATORY INFRASTRUCTURE RESOURCES:
- Programming languages and technologies for the machine learning application, frontend, backend and cloud deployment: C, C++, Python, JavaScript, HTML, CSS and others that are relevant to the application developed;
- Access to a cloud platform with functionality for data engineering and machine learning pipelines;
- Version control system: Git with distribution and collaboration on the GitHub platform;
- Database: DynamoDB, SQLite, MySQL or PostgreSQL or Market Partner Database;
- Integrated Development Environment- IDE (VSCode);
- Services from cloud technology partners;
- Frameworks, libraries and/or software for integration with the IoT system provided by the institution;
- Access to a data set with annotations and of a size consistent with the training of machine learning models for the application in question (e.g., examples of sentences categorized with intentions, expected responses and/or expected actions).
COMPLETION AND DELIVERY OF THE PROJECT:
All the prototypes generated during the project are delivered at the end of the 10th week.
