Deciphering Call-log Intent using Advanced Speech-To-Text and NLP Algorithms for a Major US Telecom Provider

About the client

A US-based Telecom Major

Problems:

The client a telecom major was looking for a solution that would enable them to understand their customer grievances and preferences better. They wanted to draw inferences from recorded call logs to understand customer sentiment and intent to drive personalization, engagement, and loyalty.

• The client was receiving innumerable calls pertaining to their services and was keen to understand the reasons behind these calls
• Unable to gain insights into the innumerable telecalls the client could not prioritize the concerns
• They wanted a solution that could quickly and accurately decipher calls at scale
• The client wanted a speech-to-text-to-insight solution that is dependable and repeatable

Solution:

Innover designed a comprehensive solution that converted audio logs into text and drew valuable insights from the available text using machine learning and natural language processing algorithms.

• The solution was equipped with a transfer learning model for understanding the context of the call
• Unsupervised model to auto-generate “keywords” from the context
• Created a corpus of disposition codes along with agents that act as a foundation for benchmarking the entire process
• Highlighted the keywords and context to actual call logs to drive actions