[PhD position] Robust and Generalizable Deep Learning-based Audio-visual Speech Enhancement
The MULTISPEECH Team at Inria (Nancy, France), is seeking a PhD candidate to work on the topic “Robust and Generalizable Deep Learning-based Audio-visual Speech Enhancement“.
Context: Audio-visual speech enhancement (AVSE) refers to the task of improving the intelligibility and quality of a noisy speech utilizing the complementary information of visual modality (lips movements of the speaker) . Visual modality can help distinguish target speech from background sounds especially in highly noisy environments. Recently, and due to the great success and progress of deep neural network (DNN) architectures, AVSE has been extensively revisited. Existing DNN-based AVSE methods are categorized into supervised and unsupervised approaches. In the former category, a DNN is trained to map noisy speech and the associated video frames of the speaker into a clean estimate of the target speech. The unsupervised methods  follow a traditional maximum likelihood-based approach combined with the expressive power of DNNs. Specifically, the prior distribution of clean speech is learned using deep generative models such as variational autoencoders (VAEs) and combined with a likelihood function based on, e.g., non-negative matrix factorization (NMF), to estimate the clean speech in a probabilistic way. As there is no training on noisy speech, this approach is unsupervised.
Supervised methods require deep networks, with millions of parameters, as well as a large audio-visual dataset with diverse enough noise instances to be robust against acoustic noise. There is also no systematic way to achieve robustness to visual noise, e.g., head movements, face occlusions, changing illumination conditions, etc. Unsupervised methods, on the other hand, show a better generalization performance and can achieve robustness to visual noise thanks to their probabilistic nature . Nevertheless, their test phase involves a computationally demanding iterative process, hindering their practical use.
Objectives: Project description: In this PhD project, we are going to bridge the gap between supervised and unsupervised approaches, benefiting from both worlds. The central task of this project is to design and implement a unified AVSE framework having the following features: 1- Robustness to visual noise, 2- Good generalization to unseen noise environments, and 3- Computational efficiency at test time. To achieve the first objective, various techniques will be investigated, including probabilistic switching (gating) mechanisms , face frontalization , and data augmentation . The main idea is to adaptively lower bound the performance by that of audio-only speech enhancement when the visual modality is not reliable. To accomplish the second objective, we will explore techniques such as acoustic scene classification combined with noise modeling inspired by unsupervised AVSE, in order to adaptively switch to different noise models during speech enhancement. Finally, concerning the third objective, lightweight inference methods, as well as efficient generative models, will be developed. We will work with the AVSpeech  and TCD-TIMIT  audio-visual speech corpora.
 D. Michelsanti, Z. H. Tan, S. X. Zhang, Y. Xu, M. Yu, D. Yu, and J. Jensen, “An overview of deep-learning based audio-visual speech enhancement and separation,” arXiv:2008.09586, 2020.
 M. Sadeghi, S. Leglaive, X. Alameda-Pineda, L. Girin, and R. Horaud, “Audio-visual speech enhancement using conditional variational auto-encoders,” IEEE/ACM Transactions on Audio, Speech and Language Processing, vol. 28, pp. 1788 –1800, 2020.
 M. Sadeghi and X. Alameda-Pineda, “Switching variational autoencoders for noise-agnostic audio-visual speech enhancement,” in ICASSP, 2021.
 Z. Kang, M. Sadeghi, R. Horaud, “Face Frontalization Based on Robustly Fitting a Deformable Shape Model to 3D Landmarks,” arXiv:2010.13676, 2020.
 S. Cheng, P. Ma, G. Tzimiropoulos, S. Petridis, A. Bulat, J. Shen, M. Pantic, “Towards Pose-invariant Lip Reading,” in ICASSP, 2020.
 A. Ephrat, I. Mosseri, O. Lang, T. Dekel, K. Wilson, A. Hassidim, W.T. Freeman, M. Rubinstein, “Looking to Listen at the Cocktail Party: A Speaker-Independent Audio-Visual Model for Speech Separation,” SIGGRAPH 2018.
 N. Harte and E. Gillen, “TCD-TIMIT: An Audio-Visual Corpus of Continuous Speech,” IEEE Transactions on Multimedia, vol.17, no.5, pp.603-615, May 2015.
- Master’s degree, or equivalent, in the field of speech/audio processing, computer vision, machine learning, or in a related field,
- Ability to work independently as well as in a team,
- Solid programming skills (Python, PyTorch),
- A decent level of written and spoken English.
- Subsidized meals
- Partial reimbursement of public transport costs
- Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
- Possibility of teleworking (after 6 months of employment) and flexible organization of working hours
- Professional equipment available (videoconferencing, loan of computer equipment, etc.)
- Social, cultural, and sports events and activities
- Access to vocational training
- Social security coverage