WHERE COMPUTING, DESIGN, SCIENCE AND SOCIETY MEET...

School of Computing

University of Dundee

HAMLET - synthetic speech with emotion


HAMLET - synthetic speech with emotion

(© Iain Murray, 2008 - last update 21st June 2008)

The Helpful Automatic Machine for Language and Emotional Talk

Introduction

HAMLET is a system which adds simulated vocal emotion effects into synthetic speech generated by rule. The system runs on a standard PC and controls a DECtalk speech synthesiser (originally the DECtalk hardware version, but the software version DECtalk for Windows NTis also supported).

HAMLET works by altering the voice quality of the synthesiser, as well as the pitch contour and speed of the speech; together these effects simulate emotion effects in the voice.

See my publications list for papers describing the system and background research.

Here are some fun things we've done with DECtalk.


Funding

HAMLET is the result of an SERC postgraduate studentship (1986-1989) and an SERC Research Grant (1990-1993); an equipment grant from the Nuffield Foundation (1994-1995) is also acknowledged. The HAMLET prototype won 2nd prize in the Toshiba/Design Council "Year of Invention" competition in 1989. 

Demonstration

Try listening to some samples of HAMLET for yourself ... here are two phrases, each spoken with five HAMLET emotions, plus unemotional for comparison (all are coded as WAV files, so your web browser should be able to play them directly):

Phrase 1 - "This is not what I expected"

Phrase 2 - "You have asked me that question so many times"


Emotions in Laureate

Emotion effects have also been used with BT's Laureate speech synthesiser, as part of a Short-term Research Fellowship in 1997.

Hear some of the results of this work.

© School of Computing