PARSA-Bench: A Comprehensive Persian Audio-Language Model Benchmark
In preparation for submission to EACL 2027
NLP × Neural Signals × Speech
Parmis Bathayan
M.Sc. Candidate in Computer Engineering · University of Tehran
I work on multimodal learning at the intersection of language and neural signals — bringing together EEG, speech, and NLP for cross-lingual and cross-cultural understanding.
Please feel free to contact me if you have research interests similar to mine.
Focus Areas
Multimodal NLP
Integrating EEG signals with textual representations for sentiment and meaning across languages.
Neural Signals & Cognition
EEG feature pipelines, neural decoding, and computational neuro-linguistics.
Speech & Audio
Persian ASR, speaker-attribute modeling, and audio-language model benchmarking.
Selected Publications
Education
M.Sc. in Computer Engineering
University of Tehran · GPA 19.21 / 20
Thesis: Leveraging EEG signals from Persian and English speakers to enhance multimodal sentiment analysis.
B.Sc. in Computer Engineering Ranked 1st of 96
Shahrood University of Technology · GPA 18.84 / 20
Experience
Research Assistant — IIS Lab & BCoL Lab
University of Tehran · EEG-based multimodal NLP & sentiment analysis · Supervisors: Dr. Azadeh Shakery & Dr. Mohammad-Reza A. Dehaqani
Chief Teaching Assistant — Deep Learning
University of Tehran · Coordinated a 16-member TA team; reviewed assignments and course operations.
Teaching Assistant — University of Tehran
NLP, Deep Learning, Statistical Inference, Advanced Data Mining & Advanced Algorithm Design · 5 courses across 7 semesters.
Teaching Assistant — Shahrood University of Technology
Information Retrieval from Web, Data Mining & Algorithm Design · 3 courses across 4 semesters.
Co-founder & Executive Coordinator — University of Tehran AI Club
Lead the organization and running of AI-focused talks and student events.