About Me
My name is Heming Wang, and I am currently a research scientist with the Meta GenAI LlaMA Speech team. I hold a Ph.D. in Computer Science and Engineering from [The Ohio State University] (https://www.osu.edu/), where I had the privilege of being mentored by Professor DeLiang Wang. I also hold a master’s degree in Applied Mathematics from the University of Waterloo.
My research primarily focuses on advancing auditory perception, with a particular emphasis on speech enhancement, audio super-resolution, and machine learning applications that aim to enhance human listening experiences in complex environments.
Interests
- Speech enhancement
- Generative AI
- Self-supervised learning
Education
-
PhD in Computer Science and Engineering, in progress
The Ohio State University
-
MMath in Applied Mathematics, 2018
University of Waterloo
-
BSc in Physics and Computer Science Minor, 2016
University of Waterloo
Experience
Research Intern
Tencent AI Lab
Nearfield sound resynthesis with vocoder /codec / self-supervised learning models.
Research Intern
Microsoft
Use self-supervised learning to improve generative performance of speech representations.
Research Intern
Microsoft
Use self-supervised learning to improve robust automatic speech recognition.
Recent Publications
(2024).
Combined Generative and Predictive Modeling for Speech Super-resolution.
arXiv preprint arXiv:2401.14269.
(2023).
uSee: Unified Speech Enhancement and Editing with Conditional Diffusion Models.
ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2023).
Diffusion Conditional Expectation Model for Efficient and Robust Target Speech Extraction.
arXiv preprint arXiv:2309.13874.
(2023).
Unifying Robustness and Fidelity: A Comprehensive Study of Pretrained Generative Methods for Speech Enhancement in Adverse Conditions.
arXiv preprint arXiv:2309.09028.
(2023).
SpatialCodec: Neural Spatial Speech Coding.
ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2023).
$ F0 $ Estimation and Voicing Detection With Cascade Architecture in Noisy Speech.
IEEE/ACM Transactions on Audio, Speech, and Language Processing.
(2023).
Cross-Domain Diffusion Based Speech Enhancement for Very Noisy Speech.
ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2023).
DATA2VEC-SG: Improving Self-Supervised Learning Representations for Speech Generation Tasks.
ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2022).
Wav2vec-switch: Contrastive Learning from Original-noisy Speech Pairs for Robust Speech Recognition.
ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2022).
Attention-Based Fusion for Bone-Conducted and Air-Conducted Speech Enhancement in the Complex Domain.
ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2022).
Densely-connected Convolutional Recurrent Network for Fundamental Frequency Estimation in Noisy Speech.
Proc. Interspeech 2022.
(2022).
Fusing Bone-conduction and Air-conduction Sensors for Complex-Domain Speech Enhancement.
IEEE/ACM Transactions on Audio, Speech, and Language Processing.
(2022).
Improving Noise Robustness of Contrastive Speech Representation Learning with Speech Reconstruction.
ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2022).
Cross-Domain Speech Enhancement with a Neural Cascade Architecture.
ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2021).
Neural Cascade Architecture with Triple-domain Loss for Speech Enhancement.
IEEE/ACM Transactions on Audio, Speech, and Language Processing.
(2021).
Towards Robust Speech Super-resolution.
IEEE/ACM transactions on audio, speech, and language processing.
(2020).
Time-frequency Loss for CNN based Speech Super-resolution.
ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).
(2018).
A Diffusion-based Two-dimensional Empirical Mode Decomposition (EMD) Algorithm for Image Analysis.
International Conference Image Analysis and Recognition.
(2018).
A Novel Foward-PDE Approach as an Alternative to Empirical Mode Decomposition.
arXiv preprint arXiv:1802.00835.