Description
Real Talk is a conversational analytics platform designed to analyze customer-agent interactions in real-time or post-call. It utilizes advanced Natural Language Processing (NLP) to transcribe and derive key insights such as sentiment, agent performance, and customer satisfaction from call transcripts. The system also categorizes conversations based on topics, key phrases, and value-added actions taken by agents.
Practical Use Case and User Story
As a customer service manager, I want to analyze call transcripts for sentiment, intent, and themes using VADER and AWS Lambda, so I can assess agent performance and identify key issues at scale. The system processes transcripts, stores results in an AWS database, and visualizes agent metrics through Amazon QuickSight dashboards. This helps me monitor customer interactions and improve service quality efficiently.
Tech Stack Involved
- AWS (Lambda, S3, RDS, QuickSight)
- Python (NLP libraries: VADER, spaCy)
- SQL Server for database management
- Amazon Transcribe (for transcription)
