A NEW CONFIDENCE MEASURE BASED ON RANK-ORDERING SUBPHONE SCORES

Abstract

This paper describes a new confidence measure based on rank-ordering subphone likelihood scores. The approach consists of three major steps. The first step is standard decoding which hypothesizes a word string given an utterance. Then forced Viterbi alignment is made at the second step. (This step is of course not needed for a Viterbi decoder.) From the aligned sentence, the third step computes likelihood scores of the hypothesized subphone and all other competing subphones to generate a list of subphones in the descending order of the likelihood scores. A rank is assigned to the hypothesized subphone according to its positioning in the list. The rank value is then merged to obtain the corresponding rank at the phone level. After the merge, selective weighting is applied such that contribution of phones having large acoustic variations is de-emphasized. Additional upper-bound limiting is also made to guarantee a rank computation not to be contaminated by a very bad segment. The new confidence measure has been favorably evaluated on word confirmation/rejection experiments with a small vocabulary of many confusable and short words. More specifically, experimental results show that the new approach outperforms other measures such as whole-word scores by reducing the equal error rate from 32% to 20%.