< Index page

4th ISCA Tutorial and Research Workshop on

Atholl Palace Hotel, Perthshire, Scotland

August 29th - September 1st, 2001

Welcome from the chair

It gives me great pleasure to introduce the proceedings of the fourth ISCA Speech Synthesis Workshop, which was held at the Atholl Palace hotel in Perthshire, Scotland. This workshop was held in an excellent location which maintained the theme of previous workshops in being situated away from the noise and bustle of the city.

It is certainly an exciting time for the speech synthesis community. In the last few years synthesis systems have made incredible progress and now produce speech quality that many would have previously thought impossible. I'm sure I speak for many at the workshop when I say we have witnessed a seismic shift in users' attitudes towards text-to-speech. Whereas before the common impression was frequently of robotic sounding systems, now we see that many users are very happy with the quality.

Hand in hand with this, we have also seen an explosion in commercial activity with regard to speech synthesis, with several top quality synthesizers on the market, all making healthy sales. Whereas in the past TTS was often seen as a last resort, now it is common to hear of companies using TTS in email reading, talking heads, voice portals, directories enquiries and even horoscope reading! The perceived importance of TTS is certainly increasing if registrations to this workshop were anything to go by: the workshop was completely full with a total of 120 people attending.

Underlying all this success, is of course the efforts of our research community, which is the focus of this workshop. The ``sudden'' improvement in perceived quality is of course the work of many individuals in many laboratories over many years. While great progress has certainly been made in recent years, the quality is far from perfect and I expect to see challenging research for many years yet.

In terms of workshop format, a number of notable differences from previous workshops were made. Firstly, as with this year's Eurospeech, authors were required to submit full and final papers rather than summary abstracts. While this resulted in fewer submissions than in the immediately preceding workshop, I am happy to say that the quality of the submitted papers was judged to be extremely high by the review panels and I hope any reduction in quantity is more than made up by quality. Each paper was independently reviewed by 3 members of the scientific community and 41 papers in total were accepted. These were presented in 7 scientific sessions; 6 oral and 1 poster.

A particularly exciting aspect of the last workshop was the evaluation session. We aimed for something similar this time, and some months ago calls for participation were sent out. The response was excellent and has resulted in 42 demonstration systems from 18 participants in 20 languages. In a departure from last time, the emphasis this time is very much on demonstration rather than evaluation. In practice this means that the demonstration comprises two components. In the first part, the same specially chosen set of text sentences was synthesized by each system (These were created off-line prior to the workshop). The second component was a free for all, where participants presented their system live on a laptop for attendees to type in anything they wanted. In addition to these, each participant was asked to present a poster giving a system overview.

The workshop had two keynote speakers. The first, Gérard Bailly, presented an excellent review paper on visual synthesis. Gérard was asked to present on this subject to help inform the workshop of the exciting work currently being carried out in the related field of Visual Synthesis. Our second speaker, Yoshinori Sagisaka, presented a review paper on unit selection. This was felt to be a great opportunity to hear a unique perspective on the developments in this field by one of its first pioneers.

Another feature of the workshop was the two discussion sessions. Kevin Lenzo and Gérard Bailly chaired two animtaed debates on Open Source Software for Synthesis, and Evaluation.

On a sadder note, many of you will no doubt now know that a pivotal member of our community, Mike Macon, tragically died earlier this year. Mike was a faculty member at OGI, were he had worked since completing his PhD in 1996. Mike was the centre of synthesis activities at OGI and made a real impact through his research. Those that knew him will remember him as an energetic and fun figure, always the life and soul of the party. I know he would have loved to have come to this workshop, and on behalf of the scientific and organising committee, it gives me great honour to dedicate this conference to his memory.

It now remains for me to thank the many people who have helped in the organisation of the workshop. These include Eva Steel who took care of the admin, Inez Woldman who helped with graphics and web pages, Rob Clark who managed the paper submission and review system. I would like to thank the scientific committee for giving up their valuable time to review the submitted papers. A special thanks also to Alan Black for taking on the arduous task of organising the demonstration session.

Paul Taylor,
Workshop Chair.

On behalf of the organising committee:
Andy Breen
Bob Damper
John Local