Enhancing Customer Service Analytics with Call Center Speech Dataset.

Client's Challenge & Our Solution

A leading customer service analytics provider partnered with FutureBeeAI to develop a customer service analytics model, requiring custom recorded call center conversations in BFSI, Retail, Delivery, and Logistics domains, focusing on Hindi and Arabic languages. Despite preferring custom-recorded data, they insisted on incorporating domain-specific terminology within 100 hours of conversations.

In the consultation phase, we expanded the domain-topic-subtopic list and diversified the dataset by defining specific inbound-outbound, and sentiment (positive, negative, neutral) ratios. We then onboarded and trained over 200 participants, ensuring a diverse range of accents, and collected the entire dataset, including high-quality transcriptions.

Outcome & Features:

Collected 100 hours of call center conversations in Hindi and Arabic.

Integrated domain-specific terminology for improved model accuracy.

Delivered structured, transcribed data that significantly boosted the client's model performance.