Towards Pretraining Robust ASR Foundation Model with Acoustic-Aware Data Augmentation

Open in new window