대형 사전훈련 모델의 파인튜닝을 통한 강건한 한국어 음성인식 모델 구축
Building robust Korean speech recognition model by fine-tuning large pretrained model
0
조회
0
다운로드
9
인용
0
참고문헌
키워드
초록
Automatic speech recognition (ASR) has been revolutionized with deep learning-based approaches, among which self-supervised learning methods have proven to be particularly effective. In this study, we aim to enhance the performance of OpenAI’s Whisper model, a multilingual ASR system on the Korean language. Whisper was pretrained on a large corpus (around 680,000 hours) of web speech data and has demonstrated strong recognition performance for major languages. However, it faces challenges in recognizing languages such as Korean, which is not major language while training. We address this issue by fine-tuning the Whisper model with an additional dataset comprising about 1,000 hours of Korean speech. We also compare its performance against a Transformer model that was trained from scratch using the same dataset. Our results indicate that fine-tuning the Whisper model significantly improved its Korean speech recognition capabilities in terms of character error rate (CER). Specifically, the performance improved with increasing model size. However, the Whisper model’s performance on English deteriorated post fine-tuning, emphasizing the need for further research to develop robust multilingual models. Our study demonstrates the potential of utilizing a fine-tuned Whisper model for Korean ASR applications. Future work will focus on multilingual recognition and optimization for real-time inference.
Abstract
Automatic speech recognition (ASR) has been revolutionized with deep learning-based approaches, among which self-supervised learning methods have proven to be particularly effective. In this study, we aim to enhance the performance of OpenAI’s Whisper model, a multilingual ASR system on the Korean language. Whisper was pretrained on a large corpus (around 680,000 hours) of web speech data and has demonstrated strong recognition performance for major languages. However, it faces challenges in recognizing languages such as Korean, which is not major language while training. We address this issue by fine-tuning the Whisper model with an additional dataset comprising about 1,000 hours of Korean speech. We also compare its performance against a Transformer model that was trained from scratch using the same dataset. Our results indicate that fine-tuning the Whisper model significantly improved its Korean speech recognition capabilities in terms of character error rate (CER). Specifically, the performance improved with increasing model size. However, the Whisper model’s performance on English deteriorated post fine-tuning, emphasizing the need for further research to develop robust multilingual models. Our study demonstrates the potential of utilizing a fine-tuned Whisper model for Korean ASR applications. Future work will focus on multilingual recognition and optimization for real-time inference.
참고문헌 (20)
- "Aihub broadcast content korean speech recognition data" (2021)[URL]
- Baevski, A., "wav2vec 2.0: A framework for self-supervised learning of speech representations", Proceedings of theAdvances in Neural Information Processing Systems, 2, pp.12449~12460 (2020)
- Bang, J. U., "KsponSpeech : Korean spontaneous speech corpus for automatic speech recognition", Applied Sciences, 10(19), pp.6936~ (2020)
- Chan, W., "Listen, attend and spell: A neural network for large vocabulary conversational speech recognition", Proceedings of the 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.4960~4964 (2016)
- Chen, T., "A simple framework for contrastive learning of visual representations", Proceedings of the 37th International Conference on Machine Learning, pp.1597~1607 (2020)
- "Bert: Pre-training of deep bidirectional transformers for language understanding" (2018)[URL]
- Graves, A., "Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks", Proceedings of the 23rd International Conference on Machine Learning, pp.369~376 (2006)
- Graves, A., "Towards end-to-end speech recognition with recurrent neural networks", Proceedings of the 31st International Conference on Machine Learning, pp.1764~1772 (2014)
- Hadsell, R., "Dimensionality reduction by learning an invariant mapping", Proceedings of the 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'06), pp.1735~1742 (2006)
- 남기훈, "한글 단어의 음성 인식 처리에 관한 연구", 문화기술의 융합, 5(4), pp.407~412 (2019), 국제문화기술진흥원
- 오유리, "Fast offline transformer-based end-to-end automatic speech recognition for real-world applications", ETRI Journal, 44(3), pp.476~490 (2022), 한국전자통신연구원[DOI]
- "Openai/whisper" (2023)[URL]
- Panayotov, V., "Librispeech: An ASR corpus based on public domain audio books", Proceedings of the 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp.5206~5210 (2015)
- Radford, A., "Robust speech recognition via large-scale weak supervision", Proceedings of the 40th International Conference on Machine Learning, pp.28492~28518 (2023)
- Schneider, S., "wav2vec : Unsupervised pre-training for speech recognition", Proceedings of the Interspeech 2019, pp.3465~3469 (2019)
- Tsai, Y. H. H., "Multimodal transformer for unaligned multimodal language sequences", Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp.6558~6569 (2019)
- "Representation learning with contrastive predictive coding" (2018)[URL]
- Vaswani, A., "Attention is all you need", Proceedings of the Advances in Neural Information Processing Systems (2017)
- "ESPnet : End-to-end speech processing toolkit" (2018)[URL]
- Yadav, H., "A survey of multilingual models for automatic speech recognition", Proceedings of the Thirteenth Language Resources and Evaluation Conference, pp.5071~5079 (2022)