Project Overview
A curated overview of automatic speech recognition research for Kurdish, covering Central Kurdish (Sorani), Northern Kurdish (Kurmanji), Southern Kurdish, and Garrusi. The collected work spans early statistical and CMUSphinx-based systems through deep-learning, wav2vec 2.0 XLS-R, Whisper, and other Transformer-based approaches, together with the speech corpora and evaluation resources used to develop and benchmark these systems.
Technologies & Methods
Applications
- Kurdish Speech Transcription
- Speech Interfaces
- Accessibility Technology
- Speech Resource Development
- Low-Resource Language Technology
Related Publications
Jira: a Central Kurdish speech recognition system, designing and building speech corpus and pronunciation lexicon
Hadi Veisi , Hawre Hosseini , Mohammad Mohammadamini , Wirya Fathy , Aso Mahmudi (2022)
Jira is a large-vocabulary speech recognition system for Central Kurdish (Sorani), developed together with a purpose-designed speech corpus and pronunciation lexicon. The work uses corpus design based …
Central Kurdish Automatic Speech Recognition using Deep Learning
Abdulhady Abas Abdullah , Hadi Veisi (2022)
Develops a Central Kurdish automatic speech recognition system using deep learning and transfer learning. The approach combines MFCC speech features with LSTM networks and Connectionist Temporal …
Breaking Walls: Pioneering Automatic Speech Recognition for Central Kurdish: End-to-End Transformer Paradigm
Abdulhady Abas Abdullah , Hadi Veisi , Tarik A. Rashid (2024)
Develops end-to-end Transformer-based automatic speech recognition for Central Kurdish using a newly constructed 100-hour speech corpus. The study fine-tunes wav2vec 2.0 XLS-R models and combines them …
End-to-End Transformer-based Automatic Speech Recognition for Northern Kurdish: A Pioneering Approach
Abdulhady Abas Abdullah , Shima Tabibian , Hadi Veisi , Aso Mahmudi , Tarik A. Rashid (2024)
Investigates end-to-end automatic speech recognition for Northern Kurdish (Kurmanji) by fine-tuning OpenAI Whisper v3 using multiple adaptation strategies. The work uses the Northern Kurdish portion …
Kurdish (Sorani) Speech to Text: Presenting an Experimental Dataset
Akam Qader , Hossein Hassani (2019)
Presents an early experimental speech-to-text resource for Sorani Kurdish using CMUSphinx. The accompanying BD-4SK-ASR dataset contains 2,000 utterances derived from vocabulary and sentences in …
Southern Kurdish Speech Recognition Resources and Benchmarking
Mohammad Mohammadamini , Marie Tahon (2026)
Introduces speech-recognition resources and an evaluation benchmark for Southern Kurdish, including approximately 30 hours of validated read speech. Whisper-turbo and wav2vec-BERT CTC baselines are …
FLEURS-Kobani: Extending the FLEURS Dataset for Northern Kurdish
Daban Q. Jaff , Mohammad Mohammadamini (2026)
Extends the multilingual FLEURS benchmark to Northern Kurdish with 5,162 validated utterances totaling 18 hours and 24 minutes from 31 native speakers. The resource supports ASR and speech-translation …
Unadapted Multilingual ASR on a Garrusi Kurdish Evaluation Set: A Common-Reference Staged Normalization Analysis
Hiwa Asadpour (2026)
Evaluates unadapted multilingual automatic speech recognition systems on a newly constructed Garrusi Kurdish evaluation set. The study uses staged text normalization to enable more consistent CER and …
Related Datasets
AsoSoft Speech Corpus
Published: January 1, 2021 | Size: 43.68 hours (full corpus); public subset approximately 30 hours
Central Kurdish (Sorani) speech corpus developed for automatic speech recognition and related speech-processing research. The full corpus reported in the Jira study contains 43.68 hours of speech from …
Central Kurdish 100-Hour ASR Corpus
Published: January 1, 2024 | Size: Approximately 100 hours
A Central Kurdish (Sorani) speech corpus constructed for end-to-end automatic speech recognition research. The Breaking Walls study reports approximately 100 hours of speech and uses the corpus to …
Mozilla Common Voice 18.0 - Northern Kurdish
Published: January 1, 2024 | Size: Approximately 101 recorded hours; approximately 68 validated hours
Northern Kurdish (Kurmanji) speech data from Mozilla Common Voice 18.0 used for end-to-end Transformer-based ASR research. The referenced study reports approximately 101 hours of recorded speech and …
BD-4SK-ASR
Published: January 1, 2019 | Size: 2,000 utterances
An experimental Sorani Kurdish speech-to-text dataset developed for early Kurdish automatic speech recognition research using CMUSphinx. The dataset contains 2,000 utterances based on vocabulary and …
Southern Kurdish ASR Benchmark
Published: January 1, 2026 | Size: Approximately 30 hours of validated speech
A Southern Kurdish speech-recognition resource and benchmark introduced for low-resource ASR evaluation. The associated study reports approximately 30 hours of validated read speech and evaluates …
FLEURS-Kobani
Published: January 1, 2026 | Size: 5,162 validated utterances; 18 hours 24 minutes; 31 speakers
An extension of the FLEURS multilingual speech benchmark for Northern Kurdish. The resource contains 5,162 validated utterances totaling 18 hours and 24 minutes of speech from 31 native speakers and …
Garrusi Kurdish ASR Evaluation Set
Published: January 1, 2026 | Size:
A Garrusi Kurdish evaluation set constructed for assessing unadapted multilingual automatic speech recognition systems. The associated study applies common-reference staged normalization to compare …