Southern Kurdish ASR Benchmark
Year: 2026
Size: Approximately 30 hours of validated speech
License:
Domain: Automatic Speech Recognition
Languages:
Southern Kurdish
Formats:
Audio
Text Transcriptions
Resource Description
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 Whisper-turbo and wav2vec-BERT CTC baselines.
Data Structure
Validated Southern Kurdish read speech with corresponding transcriptions for ASR benchmarking.
Primary Publication
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 …
Related Publications
Dataset Information
Total Size
Approximately 30 hours of validated speech
Languages
1 language
Formats
2 formats
Related Papers
1 publication
Data Access
Licensed under