Sorry, you need to enable JavaScript to visit this website.

This paper proposes an unsupervised data selection method by using a submodular function based on contrastive loss ratios of target and training data sets. A model using a contrastive loss function is trained on both sets. Then the ratio of frame-level losses for each model is used by a submodular function. By using the submodular function, a training set for automatic speech recognition matching the target data set is selected.

In recent years, audio and video deepfake technology has advanced relentlessly, severely impacting people's reputation and reliability.
Several factors have facilitated the growing deepfake threat.
On the one hand, the hyper-connected society of social and mass media enables the spread of multimedia content worldwide in real-time, facilitating the dissemination of counterfeit material.

Feature selection has been explored in two ways, global feature selection and instance-wise feature selection. Global feature selection picks the same feature selector for the entire dataset, while instance-wise feature selection allows different feature selectors for different data instances. We propose group-wise feature selection, a new setting that sits between global and instance-wise feature selections.

While end-to-end models have shown great success on the Automatic Speech Recognition task, performance degrades severely when target sentences are long-form. The previous proposed methods, (partial) overlapping inference are shown to be effective on long-form decoding. For both methods, word error rate (WER) decreases monotonically when over- lapping percentage decreases. Setting aside computational cost, the setup with 50% overlapping during inference can achieve the best performance. However, a lower overlapping percentage has an advantage of fast inference speed.