INTERSPEECH 2004 - ICSLP
This paper proposes a two-level schema for detecting recognition errors. Given the recognition hypothesis of an utterance, the first level in our schema applies an utterance classifier (UC) to decide if the hypothesis is error-free or erroneous. In the latter case, the utterance is passed on to the second level in our schema for further processing. A word classifier (WC) is applied to each of the word hypotheses in the utterance to decide whether or not it is a misrecognition. Hence the two-level schema can locate error-containing regions in the recognition hypotheses. These are the target regions to which we can apply more sophisticated and expensive language models for error correction as a next step. Experiments on Mandarin Chinese speech recognition showed that the UC has a detection error rate of 16.5% for misrecognized utterances; the WC has a detection error rate of 19.8% for erroneous word hypotheses.
Bibliographic reference. Zhou, Zhengyu / Meng, Helen (2004): "A two-level schema for detecting recognition errors", In INTERSPEECH-2004, 449-452.