Abstract:The use of forward-backward (FB) computation based posterior probabilities as confidence measures (CMs) for all recognized candidates in a lattice seems to be common across various lattice-based audio indexing systems. However, a major limitation with this approach is that its performance for CMs cannot be improved easily, since it relies almost entirely on a single information source - the acoustic and language-model probabilities. In this paper, we propose to formulate computing CMs in the lattice case as a multi-class sequential labeling problem, using conditional random fields (CRFs) as the underlying model. In this approach, various relevant features including the FB posterior probabilities could be combined together. Note that CRFs are well suited to label sequence data and some features are defined over a word sequence. This paper presents how we resolve these two issues in the lattice case, beyond others' previous work in CRF-based CMs for the 1-best case. Once properly implemented, the proposed approach achieves significant performance improvements for both CMs in the lattice case and lattice-based audio indexing.

Study on Chinese speech retrieval based on posterior probability

Syllable Based Audio Search Using Confusion Network Arc As Indexing Unit

Improved lattice-based spoken document retrieval by directly learning from the evaluation measures

Improved Posterior Probability Estimation Methods for the Freely-Spoken Speech Evaluation

Lattice-based continuous Chinese speech indexing based on mutual information confidence measure

Fast Fuzzy Keyword Spotting Using Syllable Confusion Network Indexing

A Fast Fuzzy Keyword Spotting Algorithm Based on Syllable Confusion Network.

Lattice-based Indexing for Spontaneous Mandarin Speech

A k-climax neighbors policy based viterbi decoding for large vocabulary mandarin speech recognition

Phonetic Transcription Verification with Generalized Posterior Probability

Calibration of Word Posterior Estimation in Confusion Networks for Keyword Search

Keyword Spotting Based on Syllable Confusion Network.

CRF-based confidence measures of recognized candidates for lattice-based audio indexing

A Study of Lattice-Based Spoken Term Detection for Chinese Spontaneous Speech

Out-of-Vocabulary Issue in Chinese Spoken Term Detection and A Two-Stage Chinese Speech Retrieval Method

Posterior-based confidence measures for spoken term detection

Posteriori Confidence Measure of Keyword in Internet Phone Speech Recognition

Integrated Semantic and Phonetic Post-correction for Chinese Speech Recognition

Word Spotting Based on a Posterior Measure of Keyword Confidence

Joint sparse representation based cepstral-domain dereverberation for distant-talking speech recognition

Bayesian Estimation of Keyword Confidence in Chinese Continuous Speech Recognition