Thank you to the reviewers for the feedback. MUiR outperforms the existing methods10 (Table 1). Interestingly, the existing methods each outperforms single task learning (STL) on two out of three tasks. Thismethod (Table1: MUiR+Hierarchical Init.) stilloutperforms theprevious methods on15 all tasks, but may be better or worse than MUiR for a given task. This result confirms the value of MUiR as a framework, and16 indicates thatexploring initialization schemes isapromising areaoffuture work.

Despite rapid progress in the recent past, current speech recognition systems still require labeled training data which limits this technology toasmall fraction of the languages spoken around the globe. This paper describes wav2vec-U, short for wav2vec Unsupervised, a method to train speech recognition models without any labeled data.

Deep reinforcement learning has shown great success in automatic policy learning for various sequential decision-making problems, such as training AI agents to defeat professional players in sophisticated games [74, 65, 24, 37] and controlling robots to accomplish complicated tasks [33, 38].