multilingual-call-summariser

📞 Call Transcript Summarizer

This repository contains summarization model trained to generate concise and meaningful summaries from call transcripts of indian agent customer calls. The model is trained to extract key insights, helping businesses analyze conversations efficiently.

🚀 Model Overview

• Model Type: Transformer-based Summarization Model
• Architecture: mBART
• Training Data: Preprocessed call transcripts
• Use Case: Customer support, sales calls

Using the Model with transformers

from transformers import pipeline

# Load the model
summarizer = pipeline("summarization", model="research-nobroker/multilingual-call-summariser")

# Example Call Transcript
call_transcript = """
customer: Hello, this is John from ABC Company. I was calling to check on the status of my order.
agent: Sure, John. Your order was shipped yesterday, and the tracking number is XYZ123.
customer: Thanks! Can you confirm the expected delivery date?
agent: Yes, it should arrive by next Tuesday.
"""

# Get Summary
summary = summarizer(call_transcript, max_length=100, min_length=20, do_sample=False)
print("Summary:", summary[0]['summary_text'])