You are here:
Peter Cahill
Peter Cahill
Biography
Publications
Conference Publications:
- Cahill, P. and Du, J. and Way, A. and Carson-Berndsen, J. (2009) Using Same-Language Machine Translation to Create Alternative Target Sequences for Text-To-Speech Synthesis Interspeech [Details]
- The Jess Blizzard Challenge 2007 Entry (2007) Cahill, P. and Carson-Berndsen, J Blizzard Challenge 2007 [Details]
- Cahill, P. and Aioanei, D. and Carson-Berndsen, J. (2007) Articulatory Acoustic Feature Applications in Speech Synthesis Interspeech
[Details] - Cahill, P. and Macek, J. and Carson-Berndsen, J. (2007) SVM Based Feature Extraction in Speech Synthesis SSW6 [Details]
- Cahill, P. and Carson-Berndsen, J. (2006) The Jess Blizzard Challenge 2006 Entry Blizzard Challenge 2006 [Details]
Research
Research Interests:
- Methods to automatically learn language pronunciation rules, analyse speech for both acoustic and prosodic information, and build voices entirely automatically off an unannotated collection of speech recordings.
- Identify synergies between speech synthesis and automatic speech recognition.
- Automatic learning of phonetic features and applications of them.
- Unit selection speech synthesis methodologies.
- Stochastic modelling of speech, for both synthesis and recognition.
My current work primarily focuses on multi-lingual personalised unit selection speech synthesis. My aim is to have a real time TTS system which can speak in a variety of European languages, and also some Asian languages, with multiple voices. The system I created to achieve this is deployed over web services, where any text in any supported language can be synthesised on-the-fly, while taking user preferences into account.
My research interests are: