Evaluation of DNN-based Phoneme Estimation Approach on the NTCIR-12 SpokenQuery&Doc-2 SQ-STD Subtask

This paper proposes a correct phoneme sequence estimation method using a deep neural network (DNN)-based framework for spoken term detection (STD). We use a DNN architecture as a correct phoneme estimator. The DNN-based estimator estimates a correct phoneme sequence of an utterance from some sorts of phoneme-based transcriptions produced by multiple ASR systems in post-processing, for reducing phoneme errors. In the experimental evaluation on the NTCIR-12 SpokenQuery&Doc-2 SQ-STD test collection, our proposed approach defeated the baseline system prepared by the task organizers. However, our approach could not outperform our DTW-based STD method we previously proposed.