STD Method Based on Hash Function for NTCIR11 SpokenQuery&Doc Task.
Satoru Tsuge, Norihide Kitaoka, Kazuya Takeda, Kenji Kita · NTCIR · 2014
In this paper, we describe a spoken term detection (STD) method which is used in Spoken Query and Documents task of NTCIR-11 meeting. Our STD method extracts sub-sequences from the syllable-based speech recognition candidates of the target speech and converts them into bit sequences using a hash function. The query is also converted into a bit sequence in the same way. Term detection candidates are detected by calculating the hamming distance between the bit sequence of the query and those of the target documents. Then, our method calculates the distances between the query and these candidates using DP (Dynamic Programming) matching. At the same time, our method based on the suffix array searches the query term from the wordbased speech recognition candidates. Finally, our method detects the query term by combining these results. Using this method, we submitted the results for the SQ-STD (Spoken Query Spoken Term Detection) task at NTCIR-11.