Abstract | ||
---|---|---|
Recognising directory listings for national telephone number inquiry is slowly getting within r each for modern ASR tech- nology. Two key factors for a successful system design are (1) optimal extent of lexical modelling and (2) an effective utter- ance rejection method. In this paper we show how a choice for the first has consequences for the second. We have taken the approach of building a lexicon with multiword expressions for the most frequently requested tele- phone listings, stepwise extended with filler words and less frequently addressed listings. In doing so, we keep track of the consequences that different Out of Vocabulary (OOV) rates have on two diverging keyphrase rejection schemes. Ex- perimental results on field data clearly show that tasks with high OOV rates benefit most from acoustic confidence meas- ures, while tasks with low OOV rates are better off with N- best list-based rejection schemes. |
Year | Venue | Keywords |
---|---|---|
2001 | INTERSPEECH | system design |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
4 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Gies Bouwman | 1 | 1 | 1.03 |
Janienke Sturm | 2 | 356 | 36.54 |
Lou Boves | 3 | 565 | 93.68 |