concept LibriSpeech corpus in category machine learning

This is an excerpt from Manning's book Machine Learning with TensorFlow, Second Edition MEAP V08. Login to get full access to this book.

One common source of training for ASR networks is audio books. They are useful in that they typically have both a parallel corpus of sound along with transcripts mapping to the spoken words as embodied in it. The LibriSpeech corpus is a part of the Open Speech and Language Resources (OpenSLR) project and can be used to train the deep-speech model. As you know by now, however, you’ll have to take care of some data cleaning to get the information ready for training.

Figure 17.3 MFCC features for five samples of an audio file from the LibriSpeech corpus.