Utterance verification using modified segmental probability model
نویسندگان
چکیده
Today speech recognition is requested not only to decode utterances into transcriptions, but also to determine the reliabilities of the result, by Utterance Verification (UV). With the conventional HMM, the measure of reliabilities can not be determined directly by the likelihoods of models. Whereas, Modified Segmental Probability Model (MSPM), suggested in this paper, with its normalized likelihood, facilitates rendering UV and speech recognition at the same time and as a whole. In the paper, Integrated Anti-word Model (IAM) is suggested, which is used to advance the measure of UV likelihood of MSPM. Some experiments show high perfor-mance and moderate computation with IAM.
منابع مشابه
Hybrid Utterance Verification based on N-best Models and Model derived from Kullback-Leibler Divergence
In this paper, utterance verification based on hybrid scores obtained from three pairs of models is investigated. The three models considered are the on-line garbage model, the antiword function model and a model derived using Kullback-Leibler divergence. The performance of utterance verification algorithm using hypothesis testing depends on the accuracy of the estimate of the alternative hypot...
متن کاملHybrid utterance verification based on n-best models and model derived from kulback-leibler divergence
In this paper, utterance verification based on hybrid scores obtained from three pairs of models is investigated. The three models considered are the on-line garbage model, the antiword function model and a model derived using Kullback-Leibler divergence. The performance of utterance verification algorithm using hypothesis testing depends on the accuracy of the estimate of the alternative hypot...
متن کاملSpeaker Verification Using Gaussian Com Trajectory Spa
In this paper, a novel approach is presented for text-independent speaker verification. A quantized acoustic trajectory is constructed for each utterance based on Universal Background Model (UBM). Analysis of the speaker entropy in the trajectory space demonstrates that the segmental trajectory catches some speaker-specific information, which can be used to discriminate different speakers. Gaus...
متن کاملSegmental Normalization for Robust Speaker Verification
For the task of speaker verification, similarity measure normalization methods are relevant to cope with variability problems and with data and/or decision fusion problems. The aim of this paper is to suggest a new method of normalization which combines classical world model based normalization techniques with ones based on a posteriori probability. This original method presents the well-known ...
متن کاملA segmental HMM based on a modified emission probability
In this paper, a novel segmental Hidden Markov Model (HMM) is proposed. The model is based on a modified emission density where additional statistical dependencies between subsequent frames of the speech signal are considered. In the following we derive an effective search strategy for the modified statistical model. Further an approach to parameter reduction is introduced. Experiments were car...
متن کامل