Aims
The aim of this module is to introduce the issues in speech recognition and discuss the statistical and deep learning approaches used to build automatic speech recognition (ASR) systems.
Objectives:
On completion of this module, students should understand:
- hidden Markov acoustic models, N-gram language models, and their use in speech recognition
- the use of various neural network acoustic models
- how large vocabulary speech recognition operates
- connectionist temporal classification and how to build models for the TIMIT corpus
- end-to-end trainable speech recognition approaches.