Skip to main content
STSoheil Tehranipour

Research

Research

Research in Persian automatic speech recognition, language models, and model evaluation and optimization, focused on real industrial data and conditions.

Research interests

  • 01Automatic Speech Recognition
  • 02Persian NLP
  • 03Language Models
  • 04Deep Learning
  • 05Speech Processing
  • 06AI Evaluation
  • 07Model Optimization

Publications

Only publications with confirmed metadata are listed. The full list will be completed once metadata (DOI and links) is confirmed.

  1. ParsAirCall: Automated Conversational IVR in Airport Call Center Using Deep Transfer Learning

    S. Tehranipour, M. Manthouri, S. Yazdani · 2023

    Intelligent Multimedia Processing and Communication Systems (IMPCS), vol. 4, no. 2, pp. 11–23

    DOI & link pendingProject details
  2. ParsCallScribe: Persian speech recognition for contact-center conversations

    S. Tehranipour, M. Manthouri · 2026

    —

    In preparationProject details

Academic background

2017 – 2026

Ph.D., Artificial Intelligence

Islamic Azad University, North Tehran Branch

Defended; degree issuance in progress

2013 – 2015

M.Sc., Artificial Intelligence

Islamic Azad University, Qazvin

2006 – 2011

B.Sc., Computer Engineering

K. N. Toosi University of Technology

Research projects

Datasets & technical contributions

A 512-hour Persian speech corpus was assembled for ParsCallScribe from three sources: Common Voice, survey recordings (about 85 hours, 480 speakers) and real contact-center calls (about 30 hours). Whether this data will be released publicly has not been decided.

Research profiles

Pending confirmation Google Scholar, ORCID, Semantic Scholar, ResearchGate and GitHub links will be added once confirmed.