The Journal of the Acoustical Society of America · 2014 · 10 citations · 10 references
ComputingComputer ScienceDigital CommunicationNatural Language ProcessingStatistical LearningDigital Signal ProcessingHidden Markov ModelsPattern RecognitionSignal ProcessingSpeech RecognitionData ReconstructionError ConcealmentLinear InterpolationModel AdaptationSpeech Pattern RecognitionSpeech Recognition AccuracyState Transition ProbabilitiesTransition Probabilities
In distributed speech recognition (DSR), data packets may be lost over error prone channels. A commonly used approach to rectify this is to reconstruct a full frame rate data sequence for recognition using linear interpolation. In this study, an error-concealment decoding method that dynamically adapts the transition probabilities of hidden Markov models to match the frame loss observation sequence is proposed. Experimental results show that a DSR system using the proposed method can achieve the same level of accuracy as a data reconstruction method, is more robust against heavy frame loss, and significantly reduces the computation time.
10
Capacity of a Burst-Noise Channel
E. N. Gilbert · Bell System Technical Journal · 1960 · 2.3K citations