SAEDNEWS: A researcher at the University of Tehran has successfully developed a homegrown artificial intelligence (AI) system capable of recognizing Persian speech and converting it into text during phone calls.
According to Saednews, A newly developed Iranian speech recognition technology has outperformed leading international services, including Google’s speech recognition system, in comparative evaluations.
Designed specifically to address the challenges of recognizing Persian speech in telephone conversations, the technology was developed by Hadi Vaisi, a faculty member at the University of Tehran’s Faculty of Intelligent Systems.
According to Vaisi, the system was trained using thousands of hours of real telephone conversations collected from contact centers across Iran.
He explained that using locally sourced data, including speech samples reflecting different Iranian dialects and accents, has enabled the technology to better adapt to the linguistic diversity and real-world conditions of telephone conversations in Iran.
“This local data gives the system a significant advantage in understanding how Persian is actually spoken across different regions of Iran,” Vaisi said.
The technology is already being used by several organizations in smartphone-based assistant systems and intelligent customer service solutions.
Its development marks another step toward strengthening Iran’s domestic artificial intelligence infrastructure and reducing dependence on foreign technologies. By relying on an independent system specifically tailored to the Persian language, regional dialects, and local communication patterns, the technology could help expand the development of Persian-language AI services.
