Efficient Nn-based Search Space Reduction in a Large Vocabulary Speech Recognition System
نویسندگان
چکیده
In very large vocabulary speech recognition systems using the hypothesis-verification paradigm, the verification stage is usually the most time consuming. State of the art systems combine fixed size hypothesized search spaces with advanced pruning techniques. In this paper we propose a novel strategy to dynamically calculate the hypothesized search space, using neural networks as the estimation module and designing the input feature set with a careful greedy-based selection approach. The main achievement has been a statistically significant relative decrease in error rate of 33.53%, while getting a relative decrease in average computational demands of up to 19.40%.
منابع مشابه
Spoken Term Detection for Persian News of Islamic Republic of Iran Broadcasting
Islamic Republic of Iran Broadcasting (IRIB) as one of the biggest broadcasting organizations, produces thousands of hours of media content daily. Accordingly, the IRIBchr('39')s archive is one of the richest archives in Iran containing a huge amount of multimedia data. Monitoring this massive volume of data, and brows and retrieval of this archive is one of the key issues for this broadcasting...
متن کاملDynamic beam pruning strategy using adaptive control
In large vocabulary continuous speech recognition, huge search space results in vast computational cost. While most pruning search strategies can reduce the computation, but the recognition rate often decreases. This paper aims to reduce the computation time without any sacrifice of the recognition rate. By means of the adaptive control theory, a novel pruning method is presented. It can automa...
متن کاملVector Quantizer Signal Transform
This paper deals with the problem of combination of Neural Networks (NN) and traditional statistical pattern classiiers. It is shown that a Neural Network can be used to replace the vector quantizer (VQ) and some feature extraction and feature reduction modules in a discrete pattern recognition system. A criterion for training the NN-weights and the classiier jointly is derived, leading to the ...
متن کاملDecoder Technology for Connectionist Large Vocabulary Speech Recognition
The search problem in large vocabulary continuous speech recognition (LVCSR) is to locate the most probable string of words for a spoken utterance given the acoustic signal and a set of sentence models. Searching the space of possible utterances is difficult because of the large vocabulary size and the complexity imposed when long-span language models are used. This report describes an efficien...
متن کاملAn efficient search method for large-vocabulary continuous-speech recognition
This paper proposes an efficient method for largevocabulary continuous-speech recognition, using a compact data structure and an efficient search algorithm. We introduce a very compact data structure DAWG as a lexicon to reduce the search space. We also propose a search algorithm to obtain the N-best hypotheses using the DAWG structure. This search algorithm is composed of two phases: “forward ...
متن کامل