no code implementations • Findings (EMNLP) 2021 • Wei-Tsung Kao, Hung-Yi Lee
This paper investigates whether the power of the models pre-trained on text data, such as BERT, can be transferred to general token sequence classification applications.
no code implementations • 7 Apr 2022 • Wei-Cheng Tseng, Wei-Tsung Kao, Hung-Yi Lee
Mean opinion score (MOS) is a typical subjective evaluation metric for speech synthesis systems.
no code implementations • 1 Apr 2022 • Wei-Tsung Kao, Yuan-Kuei Wu, Chia-Ping Chen, Zhi-Sheng Chen, Yu-Pao Tsai, Hung-Yi Lee
User-defined keyword spotting is a task to detect new spoken terms defined by users.
1 code implementation • 9 Nov 2021 • Wei-Cheng Tseng, Wei-Tsung Kao, Hung-Yi Lee
Recently, adapting the idea of self-supervised learning (SSL) on continuous speech has started gaining attention.
6 code implementations • 7 Apr 2021 • Wei-Cheng Tseng, Chien-yu Huang, Wei-Tsung Kao, Yist Y. Lin, Hung-Yi Lee
In this paper, we use self-supervised pre-trained models for MOS prediction.
no code implementations • 12 Mar 2021 • Wei-Tsung Kao, Hung-Yi Lee
This paper investigates whether the power of the models pre-trained on text data, such as BERT, can be transferred to general token sequence classification applications.
no code implementations • 25 Jan 2020 • Wei-Tsung Kao, Tsung-Han Wu, Po-Han Chi, Chun-Cheng Hsieh, Hung-Yi Lee
Although Bidirectional Encoder Representations from Transformers (BERT) have achieved tremendous success in many natural language processing (NLP) tasks, it remains a black box.