Bridging Speech and Textual Pre-Trained Models With Unsupervised ASR | IEEE Conference Publication | IEEE Xplore