• Home /
  • Papers /
  • Deep Learning-Based Text-to-Speech Synthesis for Central Kurdish: A Transformer-Enhanced Approach

Abstract

A Central Kurdish speech synthesis study combining a CNN-Transformer encoder, Tacotron 2 autoregressive decoding, and WaveGlow. The experiment reuses Gigant-KTTS recordings and evaluates synthesized speech with native listeners and word error rate measurements.

Keywords

Central Kurdish Transformer Tacotron 2 WaveGlow Text-to-Speech

Related Datasets

Gigant-KTTS / Gigant Dataset

2024 6,078 utterances; 13.63 hours reported WAV, XLSX

A Central Kurdish speech corpus created for text-to-speech research by Hawraz A. Ahmad and Tarik A. Rashid. The release reports 6,078 utterances from one male speaker across twelve subject categories, …

Access Dataset

Citation

Ahmed, S. B., & Mahmood, S. A. (2026). Deep Learning-Based Text-to-Speech Synthesis for Central Kurdish: A Transformer-Enhanced Approach. Passer Journal of Basic and Applied Sciences, 8(2), 712–722. https://doi.org/10.24271/psr.2026.561069.2440

Publication Details

Authors 2 authors
Datasets 1 dataset
Read Full Paper View Project Back to Papers