Kmr Tatoeba: Northern Kurdish Speech Corpus for ASR Evaluation

A Tatoeba-based speech corpus for Northern Kurdish (Kurmanji) containing only a test split. It is intended for automatic speech recognition (ASR) and speech-to-text translation (S2TT) evaluation. The dataset was created by aranemini and last updated on August 13, 2025.

Use Cases

• Evaluate automatic speech recognition (ASR) models based on recorded sentence audio.
• Benchmark speech-to-text translation (S2TT) systems based on Northern Kurdish speech samples.
• Test model performance on native speaker audio from approximately 30 speakers.

Strengths

• Contains audio recordings in .wav format at 48 kHz mono, which suggests high-quality audio suitable for speech processing.
• Includes anonymized speaker IDs for approximately 30 native speakers, allowing for speaker-diversity analysis.

Limitations

• The dataset contains only a test split, limiting its use for training or validation tasks.
• Row count, column definitions, and license information are unknown, which complicates suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Sentences from the Tatoeba Project.

Collection Method

Likely contains recorded sentences by native speakers.

Freshness

Last updated 2025-08-13 13:09:53; freshness should be verified.