CTC-Segmentation of Large Corpora for German End-to-end Speech Recognition

TL;DR

This paper presents a method to leverage unlabeled German speech data using CTC-based segmentation to improve end-to-end speech recognition, resulting in a significant WER reduction.

Contribution

It introduces a two-stage data preparation approach using CTC to extract segments from unlabeled data, enhancing training data for end-to-end ASR models.

Findings

Achieved 12.8% WER on Tuda-DE test set.

Created a dataset of over 1700 hours of German speech data.

Surpassed previous hybrid ASR baseline performance.

Abstract

Recent end-to-end Automatic Speech Recognition (ASR) systems demonstrated the ability to outperform conventional hybrid DNN/ HMM ASR. Aside from architectural improvements in those systems, those models grew in terms of depth, parameters and model capacity. However, these models also require more training data to achieve comparable performance. In this work, we combine freely available corpora for German speech recognition, including yet unlabeled speech data, to a big dataset of over $1700$h of speech data. For data preparation, we propose a two-stage approach that uses an ASR model pre-trained with Connectionist Temporal Classification (CTC) to boot-strap more training data from unsegmented or unlabeled training data. Utterances are then extracted from label probabilities obtained from the network trained with CTC to determine segment alignments. With this training data, we trained a hybrid CTC/attention Transformer model that achieves $12.8\%$ WER on the Tuda-DE test set, surpassing the previous baseline of $14.4\%$ of conventional hybrid DNN/HMM ASR.