AUTOMATIC TRANSCRIPTION AND CLASSIFICATION OF CUSTOMER CALL RECORDINGS FOR CRM ANALYTICS: A WHISPER-BASED WEB SYSTEM
Scientific works/Elmi eserler · 22.07.2026 · 132
DOI: 10.58225/mpmma.2026.132 ↗Kurz gefasst
Fachartikel in der Zeitschrift Scientific works/Elmi eserler, online erschienen am 22.06.2026. Verfasst von Ali Khalilov und 1 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben.
Abstract (Originalsprache)
This paper presents a web-based system that transcribes customer call-centre recordings and classifies them into CRM categories. Audio is stored in a relational database and transcribed into Azerbaijani text with Open AI Whisper; each transcript is then assigned a category by a supervised text classifier. Four classifiers — Multinomial Naive Bayes, Logistic Regression, Random Forest, and a linear Support Vector Machine (SVM) — are compared on TF-IDF features over 1,480 calls in five categories. The linear SVM performs best (accuracy = 91.2%, macro-F1 = 0.89) and Whisper reaches a 14.8% word error rate (WER) with the large-v3 model. Deployed as a Flask application with sub-50 ms classification latency, the pipeline offers Azerbaijani call centres a low-cost route to automated call analytics.
Zitieren
APA
Khalilov, A., & Jafarova, H. (2026). AUTOMATIC TRANSCRIPTION AND CLASSIFICATION OF CUSTOMER CALL RECORDINGS FOR CRM ANALYTICS: A WHISPER-BASED WEB SYSTEM. Scientific works/Elmi eserler, 132. https://doi.org/10.58225/mpmma.2026.132
Vancouver
Khalilov A, Jafarova H. AUTOMATIC TRANSCRIPTION AND CLASSIFICATION OF CUSTOMER CALL RECORDINGS FOR CRM ANALYTICS: A WHISPER-BASED WEB SYSTEM. Scientific works/Elmi eserler. 2026:132. doi:10.58225/mpmma.2026.132
BibTeX
@article{Khalilov2026,
author = {Khalilov, Ali and Jafarova, Hilala},
title = {AUTOMATIC TRANSCRIPTION AND CLASSIFICATION OF CUSTOMER CALL RECORDINGS FOR CRM ANALYTICS: A WHISPER-BASED WEB SYSTEM},
journal = {Scientific works/Elmi eserler},
year = {2026},
pages = {132},
doi = {10.58225/mpmma.2026.132}
}
EUVIMED