The field of speech processing has undergone a transformative shift with the advent of deep
learning. The use of multiple processing layers has enabled the creation of models capable …

Machine learning heavily relies on data, but real-world applications often encounter various
data-related issues. These include data of poor quality, insufficient data points leading to …

The virtual-to-real paradigm, ie, training models on virtual data and then applying them to
solve real-world problems, has attracted more and more attention from various domains by …

Pre-training is prevalent in nowadays deep learning to improve the learned model's
performance. However, in the literature on federated learning (FL), neural networks are …

Recently reported state-of-the-art results in visual speech recognition (VSR) often rely on
increasingly large amounts of video data, while the publicly available transcribed video …

The research community has long studied computer-assisted pronunciation training (CAPT)
methods in non-native speech. Researchers focused on studying various model …

While end-to-end automatic speech recognition (ASR) has achieved high performance, it
requires a huge amount of paired speech and transcription data for training. Recently, data …

Both unpaired speech and text have shown to be beneficial for low-resource automatic
speech recognition (ASR), which, however were either separately used for pre-training, self …

Neural transducer is now the most popular end-to-end model for speech recognition, due to
its naturally streaming ability. However, it is challenging to adapt it with text-only data …

Currently, a common approach in many speech processing tasks is to leverage large scale
pre-trained models by fine-tuning them on in-domain data for a particular application. Yet …