Dataset · farabi-lab
Kazakh speech dataset for ASR
farabi-lab/kazakh-stt
High-quality open-source Kazakh speech corpus with a duration of 554 hours for ASR systems.
What's inside
The dataset was developed by the Department of Artificial Intelligence and Big Data at KazNU and contains 554 hours of recorded Kazakh speech. 873 speakers participated in the collection, with a total of 204,250 utterances. Audio materials are presented in the .wav format.
What you can do with it
- Development and fine-tuning of automatic speech recognition (ASR) systems
- Training audio-oriented models for the Kazakh language
- Languages
- Kazakh
- Size
- 100K–1M rows
- Downloads this month
- 205
- Likes
- 12
- Access
- on request on Hugging Face
Models trained on it
- shyngys879/kazakh-whisper-large-v3-turboAutomatic speech recognition 2.6K
- mradermacher/qwen3-asr-1.7b-kazakh-GGUFAutomatic speech recognition 735
- mradermacher/qwen3-asr-0.6b-kazakh-GGUF 605
- ai4kazakh/qwen3-asr-1.7b-kazakhAutomatic speech recognition 72
- ai4kazakh/qwen3-asr-0.6b-kazakhAutomatic speech recognition 48
- Musa505/kazakh-tts 16
- nur-dev/ait-asr 0
How to load it
from datasets import load_dataset
ds = load_dataset("farabi-lab/kazakh-stt")