University Homepage School Homepage School Contacts School Search

Lecture 1 - Introduction

This lecture was an introduction to the ASR course.

• Basic logistics for the course
• Necessary background
• The problem of automatic speech recognition, and why it is difficult
• Different causes of variability - vocabulary size, speaker variation, acoustic environment, speaking style
• Hierarchical modelling of speech and the basic components of an ASR system - signal analysis, acoustic modelling, lexicon, language modelling, search.
• Frame-based representation of recorded speech
• Labelling recorded speech
• Phones and phonemes
• The TIMIT corpus (covered in the recordings not in this year's lecture)
• Evaluating speech recognition using word error rate - substitutions, insertions, deletions (covered in the recordings not in this year's lecture)

---

Copyright (c) University of Edinburgh 2015-2017
The ASR course material is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International Public License
licence.txt
This page maintained by Steve Renals.
Last updated: 2018/01/15 18:50:06UTC