mask-Net: Learning Context Aware Invariant Features using Adversarial Forgetting (Student Abstract)
This work addresses the problem of improving generalization for STT systems by making features invariant to nuisances like accents, which could benefit users in diverse linguistic environments.
This paper proposes a novel approach using adversarial forgetting to induce invariance in features for Speech-to-Text (STT) tasks. The method achieves better generalization, showing an absolute improvement of 2.2% in word error rate (WER) on out-of-distribution test sets and 1.3% on in-distribution test sets.
Training a robust system, e.g.,Speech to Text (STT), requires large datasets. Variability present in the dataset such as unwanted nuisances and biases are the reason for the need of large datasets to learn general representations. In this work, we propose a novel approach to induce invariance using adversarial forgetting (AF). Our initial experiments on learning invariant features such as accent on the STT task achieve better generalizations in terms of word error rate (WER) compared to the traditional models. We observe an absolute improvement of 2.2% and 1.3% on out-of-distribution and in-distribution test sets, respectively.