Skip to main content

Trinity College Dublin, The University of Dublin

Trinity Menu Trinity Search

School of Linguistic, Speech and Communication Sciences

You are here Our Research > Current Research Projects

FASTNET (SFI)

Focus on Actions in Social Talk: Network-Enabling Technology

FASTNET (SFI)

Fastnet@Trinity: An Charraig Aonair is a symbol of Ireland, standing isolated off its coast in much the same way that Ireland itself stands isolated from mainland Europe, but serving as its westernmost component. This project, codenamed FASTNET, will burn a light from these shores that will be seen across the world. It will take the lead in introducing a paradigm shift in speech processing that will help ordinary people to make use of advanced technology in a simple and natural way.

Team

  • Prof. Nick Campbell (Lead PI)
  • Prof. Ailbhe Ní Chasaide (PI)
  • Dr. Christer Gobl (PI)
  • PostDocs and Research Assistants

Academic Partners

  • Prof. Kashioka Hideki (NAIST, Japan): statistical modelling
  • Dr. Joaquim Gustafson (KTH, Sweden)
  • Dr. Jens Edlund (KTH, Sweden)
  • Prof. Julie Berndsen (UCD): phonetic feature recognition tools
  • Dr. Fred Cummins (UCD): data collection and annotation
  • Prof. Petra Wagner (Bielefeld): data collection and annotation
  • Professor Toshiyuki Sadanobu (Kobe University, Japan): assistance with ontology development
  • Prof. Anton Nijholt (Twente, Netherlands)
  • Prof. Anna Esposito (COST 2012, EU)

Advisors

  • Professor John Laver (Queen Margaret College, Edinburgh): auditory classification of voice quality
  • Dr. Janet Beck (Queen Margaret College, Edinburgh): auditory classification of voice quality
  • Professor James Mahshie (George Washington University): testing of the voice parameterisation system
  • Professor Francis Nolan (Cambridge University): voice and prosody modelling
  • Dr. John McKenna (Dublin City University): assistance with data collection
  • Dr. Ronan Scaife (Dublin City University): assistance with audio technology

Project Overview

Networking is an essential component of human interaction, and the content of a spoken conversation has as much to do with social bonding as with the transfer of propositional meaning. Recognising that social interactions are the essential components of vocal communication, and that actions, rather than words, are the prime units to be processed in a discourse, we aim for a paradigm shift in the way computers process speech, to incorporate speaking-style information alongside message content to provide a richer expression (or understanding) of an utterance.

This research addresses the question of how speech technology should be produced which is capable of processing not only the lexical content of a spoken utterance, but also its underlying communicative intentions. People intuitively interpret the nonverbal information in speech and tone-of-voice to aid in the interpretation of each utterance in context. We aim to produce a machine that can be programmed to perform similar interpretation of speech utterances. More ...

Further information: