Kazakh Speech Corpus 2

This repository provides the recipe for the paper KSC2: An Industrial-Scale Open-Source Kazakh Speech Corpus.

Pre-trained models

You can download the best performing model HERE.

Inference

To convert your audio file to text, please make sure it follows a wav format with sample rate of 16k. Unzip the pre-trained model in the current directory, and make sure that asr_model_path & lm_model_path refer to valid directories. Install the necessary packages by running pip install -r requirements.txt. To perform the evaluation please run:

python recognize.py -f <path_to_your_wav>

Dataset

Download ISSAI_KSC2 dataset on HuggingFace form, and untar tar -xvf ISSAI_KSC2.tar.gz in the directory of your choice. Specify the path to the dataset inside espnet/egs2/Kazakh_ASR/asr1/run.sh file:

dataset_path=specify_path

Training

Our code builds upon ESPnet, and requires prior installation of the framework for DNN training. Please follow the installation guide and put the TurkicASR folder inside espnet/egs2/ directory. Run the traning scripts with ./run.sh

Citation

@inproceedings{mussakhojayeva22,
  author={Saida Mussakhojayeva, Yerbolat Khassanov, Huseyin Atakan Varol},
  title={KSC2: An Industrial-Scale Open-Source Kazakh Speech Corpus},
  year=2022,
  booktitle={Interspeech},
}