Hao Tang 唐顥
Hi, I'm Hao. I am a Lecturer in the Institute for Language, Cognition and Computation at the University of Edinburgh. I work in the field of speech and language processing, studying models that can tell who says what at when and where.
For prospective students
If you want to pursue a PhD and are interested in working with me, please get in touch and I'm happy to hear about your research interest. I can take students from both the ILCC PhD program and the CDT in NLP.
Research
My research, often drawing ideas from signal processing and machine learning (particularly deep learning), involves designing experiments that reveal and verify properties of the models. For example,
- we analyze vector representations of speech learned from self-supervised learning (Chung et al., 2019; Chung et al., 2020);
- we show a deep residual network is able to localize sound events in time even though the model is not trained to do so (Ford et al., 2019);
- we study how speech enhancement interacts with speaker identification in a noisy environment (Shon et al., 2019);
- we analyze the phonetic cues a model picks up when performing speaker identification (Shon et al., 2019);
- we analyze how different training strategies affect the long-term memory of recurrent neural networks (Tang and Glass, 2018);
- we look at distant speech recognition from a domain adaptation viewpoint (Tang et al., 2018);
- we study end-to-end training with different graph topologies for automatic speech recognition (Tang, 2017).
About me
 |
 |
 |
| 2010–2017 | 2017–2020 | 2020– |
I started working in speech and language as a master student with Lin-Shan Lee at National Taiwan University, and have stayed in this field since then. I spent wonderful years doing PhD with Karen Livescu at Toyota Technological Institute at Chicago. Before coming to Edinburgh, I was a post-doctoral associate at MIT in the Spoken Language Systems group led by Jim Glass.
Feel free to reach out if you have any questions about me and my work. I am always happy to hear from students and colleagues who are interested in speech and language.
