Greg Kochanski

New Tools and Methods for Very-Large-Scale Phonetics Research

These papers were presented at the "New Tools and Methods for Very-Large-Scale Phonetics Research" workshop at the University of Pennsylvania, January 28-31, 2011. Archival copies may be found (via searching) at http://ora.ouls.ox.ac.uk .

• This paper, by Greg P. Kochanski, Chilin Shih, Ryan Shosted, "Should Corpora be Big, Rich, or Dense?", discusses how one can compute whether or not an ASR system has done a good job of aligning your text to your audio.
• This paper, by Ladan Baghai-Ravary, Sergio Grau, and Greg Kochanski, "Detecting gross alignment errors in the Spoken British National Corpus", discusses how one can compute whether or not an ASR system has done a good job of aligning your text to your audio. It is also available at http://arxiv.org . It was presented at the workshop as a poster .
• This paper, by John Coleman, Mark Liberman, Greg Kochanski, Lou Burnard and Jaahong Yuan, "Mining a Year of Speech", describes the Spoken British National Corpus, its uses, techniques used, and lessons learned in a project to align a large speech corpus with its transcription.

Our papers at this workshop were funded by:

• The UK's Economic and Social Research Council (the ESRC) via project RES-062-23-1172. More information on that project can be found at http://kochanski.org/gpk/papers/2007/exemplar2.
• The UK's Economic and Social Research Council (the ESRC) via project RES-062-23-1323. More information on that project can be found at http://kochanski.org/gpk/papers/2008/aesopus2.
• The UK's Economic and Social Research Council (the ESRC) via project RES-062-23-2566.
• The "Digging into Data" collaboration between the UK's JISC and the US's National Science Foundation, via the "Mining a Year of Speech" project. For more information, see http://www.phon.ox.ac.uk/mining_speech.