Skip to content

Latest commit

 

History

History
 
 

KsponSpeech

README.md

KsponSpeech

About the corpus

KsponSpeech is a large-scale spontaneous speech corpus of Korean. This corpus contains 965.2 hours of general open-domain dialog utterances with transcriptions.

More information can be found at https://www.mdpi.com/2076-3417/10/19/6936

Training and evaluation transcripts were generated the same way as in the above paper.

Prepare data

You can download KsponSpeech at Link

Run following script to unzip the downloaded file

./unzip_ksponspeech.sh "PATH_TO_DOWNLOADED/한국어 음성" PATH_TO_UNZIPPED

pcm files need to be converted to wav for training. Run following script for the conversion.

python convert_to_wav.py --dirpath PATH_TO_UNZIPPED --nj num_processes

Installing Extra Dependencies

Before proceeding, ensure you have installed the necessary additional dependencies. To do this, simply run the following command in your terminal:

pip install -r extra_requirements.txt

How to run an ASR experiment with KsponSpeech

To train a full speech recognition system the pipeline is the following:

  1. Train a tokenizer. The tokenizer takes in input the training transcripts and determines the subword units that will be used for both acoustic and language model training. Training a tokenizer before the language and acoustic model is necessary. Indeed, both of them will reuse this tokenizer to map the output tokens.
  2. Train a Language Model (LM). The language model takes in input long texts from available books. We have recipes with RNN and transformer-based LMs. In both cases, the LM is used during beam search to assign different weights to different hypotheses generated by the acoustic model.
  3. Train an acoustic model (AM). The acoustic model maps the input speech into a set of sub-words units. The current repository contains recipes for transformer-based systems.

The results obtained with the different models can be found in the corresponding sub-directories!

How to simply use pretrained models to transcribe my audio file?

SpeechBrain provides a simple interface to transcribe audio files with pretrained models. All the necessary information can be found on the HuggingFace repositories corresponding to our models for KsponSpeech:

Citing SpeechBrain

Please, cite SpeechBrain if you use it for your research or business.

@misc{speechbrainV1,
  title={Open-Source Conversational AI with SpeechBrain 1.0},
  author={Mirco Ravanelli and Titouan Parcollet and Adel Moumen and Sylvain de Langen and Cem Subakan and Peter Plantinga and Yingzhi Wang and Pooneh Mousavi and Luca Della Libera and Artem Ploujnikov and Francesco Paissan and Davide Borra and Salah Zaiem and Zeyu Zhao and Shucong Zhang and Georgios Karakasidis and Sung-Lin Yeh and Pierre Champion and Aku Rouhe and Rudolf Braun and Florian Mai and Juan Zuluaga-Gomez and Seyed Mahed Mousavi and Andreas Nautsch and Xuechen Liu and Sangeet Sagar and Jarod Duret and Salima Mdhaffar and Gaelle Laperriere and Mickael Rouvier and Renato De Mori and Yannick Esteve},
  year={2024},
  eprint={2407.00463},
  archivePrefix={arXiv},
  primaryClass={cs.LG},
  url={https://arxiv.org/abs/2407.00463},
}
@misc{speechbrain,
  title={{SpeechBrain}: A General-Purpose Speech Toolkit},
  author={Mirco Ravanelli and Titouan Parcollet and Peter Plantinga and Aku Rouhe and Samuele Cornell and Loren Lugosch and Cem Subakan and Nauman Dawalatabad and Abdelwahab Heba and Jianyuan Zhong and Ju-Chieh Chou and Sung-Lin Yeh and Szu-Wei Fu and Chien-Feng Liao and Elena Rastorgueva and François Grondin and William Aris and Hwidong Na and Yan Gao and Renato De Mori and Yoshua Bengio},
  year={2021},
  eprint={2106.04624},
  archivePrefix={arXiv},
  primaryClass={eess.AS},
  note={arXiv:2106.04624}
}

Citing this recipe

@misc{returnzero,
  title = {ReturnZero Conformer Korean ASR model},
  author = {Dongwon Kim and Dongwoo Kim and Jeongkyu Roh},
  year = {2021},
  howpublished = {\url{https://huggingface.co/ddwkim/asr-conformer-transformerlm-ksponspeech}},
}

Citing KsponSpeech dataset

@Article{app10196936,
AUTHOR = {Bang, Jeong-Uk and Yun, Seung and Kim, Seung-Hi and Choi, Mu-Yeol and Lee, Min-Kyu and Kim, Yeo-Jeong and Kim, Dong-Hyun and Park, Jun and Lee, Young-Jik and Kim, Sang-Hun},
TITLE = {KsponSpeech: Korean Spontaneous Speech Corpus for Automatic Speech Recognition},
JOURNAL = {Applied Sciences},
VOLUME = {10},
YEAR = {2020},
NUMBER = {19},
ARTICLE-NUMBER = {6936},
URL = {https://www.mdpi.com/2076-3417/10/19/6936},
ISSN = {2076-3417},
DOI = {10.3390/app10196936}
}