In this paper, we study training of automatic speech recognition system in a weakly supervised setting where the order of words in transcript labels of the audio training data is not known. We train a word-level acoustic model which aggregates the distribution of all output frames using LogSumExp operation and uses a cross-entropy loss to match with the ground-truth words distribution. Using the pseudo-labels generated from this model on the training set, we then train a letter-based acoustic model using Connectionist Temporal Classification loss. Our system achieves 2.3%/4.6% on test-clean/test-other subsets of LibriSpeech, which closely matches with the supervised baseline's performance.
@article{arxiv.2110.05994,
title = {Word Order Does Not Matter For Speech Recognition},
author = {Vineel Pratap and Qiantong Xu and Tatiana Likhomanenko and Gabriel Synnaeve and Ronan Collobert},
journal= {arXiv preprint arXiv:2110.05994},
year = {2021}
}