Издательство John Wiley, 2013, -355 pp.
This book came about as a result of the standing-room-only special session on crowdsourcing for speech processing at Interspeech 2011. There has been a great amount of interest in this new technique as a means to solve some persistent issues. Some researchers dived in head first and have been using crowdsourcing for a few years by now. Others waited to see if it was reliable, and yet others waited for some service to exist in their country. The first results are very encouraging: crowdsourcing can be a solution that approaches expert results. However, it also comes with warnings: the incoming data must go through quality assessment.
This book is a hands-on, how-to manual that is directed at several groups of readers:
Experienced users: Those who have already used crowdsourcing for speech processing should find a good set of references to the literature as well as some novel approaches that they may not be familiar with.
Speech processing users who have not yet used crowdsourcing: The information in this book should help you get up to speed rapidly and avoid reinventing the wheel for common interface and assessment issues.
Users who are not speech processing experts who also need to use crowdsourcing for their speech data: This book should also help you get started since you will have many of the same issues in dealing with your data.
We start the book with an overview of the principles of crowdsourcing. This is followed by some basic concepts and an overview of research in the area. The following chapters in the book cover most of the present types of speech processing. Chapter 3 covers the acquisition of speech. Chapter 4 covers speech labeling. Chapter 5 covers the variability of crowd speech and how to acquire and label speech in one effort. Chapter 6 explains how to run perception experiments using crowdsourcing. Chapter 7 explains how to use crowdsourcing for speech synthesis. Chapter 8 describes how to use crowdsourcing for assessment of spoken dialog systems. Chapter 9 covers the variety of platforms that are used for crowdsourcing and how they work. Chapter 10 covers industrial applications of crowdsourcing for speech processing. Finally, Chapter 11 covers the legal and ethical issues surrounding the use of crowdsourcing.
An Overview.
The Basics.
Collecting Speech from Crowds.
Crowdsourcing for Speech Transcription.
How to Control and Utilize Crowd-Collected Speech.
Crowdsourcing in Speech Perception.
Crowdsourced Assessment of Speech Synthesis.
Crowdsourcing for Spoken Dialog System Evaluation.
Interfaces for Crowdsourcing Platforms.
Crowdsourcing for Industrial Spoken Dialog Systems.
Economic and Ethical Background of Crowdsourcing for Speech.