Abstract Decadt
In this paper, we describe a method to enhance the readability of the textual output of a large vocabulary speech recognizer when out-of-vocabulary (OOV) words occur. The basic idea is (i) to mark, in the transcriptions of the speech recognizer, the words which are uncertain and (ii) to replace these words with phoneme strings generated by a phoneme recognizer and to convert the phonemes to graphemes.This paper concentrates on the second step: we show that the phoneme strings can be reasonably reliably transcribed orthographically using machine learning techniques. More specifically: (i) we show baseline performance of a machine learning approach to phoneme-to-grapheme conversion when different levels of artificial noise are added (simulating phoneme recognizer errors), (ii) we provide results on real phoneme recognition data, and (iii) we provide a detailed error analysis.
The experiments show that, even when the grapheme strings are not fully correct, the resulting transcriptions are more easily readable than the original ones.
Last modified $Date: 2001/10/04 13:39:44 $ by Parlevink Webmaster