Evaluation of Hybrid Vector Quantization and Hidden Markov Model Methods in Noisy Environments
نویسندگان
چکیده
In this paper, we presents a comparison between Hidden Markov Model (HMM) and an approach using a hybrid of Vector Quantization (VQ) with HMM methods. The aim of combination scheme used is to improve the standalone HMM performance. A Malay spoken digit database is used for the testing and validation modules. It is shown that, in clean environments, a total success rate (TSR) of 99.97% is achieved using this hybrid approach. For speaker verification, the true speaker rejection rate is 0.06% while the impostor acceptance rate is 0.03% and the equal error rate (EER) is 11.72%. Meanwhile, in noisy environments, TSRs of between 62.57%-76.80% are achieved for SNRs of 0-30 dBs. Key-Words: Speaker Verification, Speech Recognition, Vector Quantization, Hidden Markov Model
منابع مشابه
New Ways to Use Lvq - Codebookstogether with Hidden Markov
We introduce a novel way to employ codebooks trained by Learning Vector Quantization together with hidden Markov models. In previous work, LVQ-codebooks have been used as frame labelers. The resulting label stream has been modeled and decoded by discrete observation HMMs. We present a way to extract more information out of the LVQ stage. This is accomplished by modeling the class-wise quantizat...
متن کاملA Comparative Study of Techniques for Hmm-based Speech Recognition in Noisy Car Environment
The performance of existing speech recognition systems degrades rapidly in the presence of background noise when training and testing cannot be done under the same ambient conditions. The aim of this paper is to report the application of several robust techniques on a system based on the HMM (Hidden Markov Models) and VQ (Vector Quantization) approaches for speech recognition in noisy car envir...
متن کاملUsing a New Hybrid Models for Speech and Medical Pattern Classification
The main goal of this paper is to compare the performance which can be achieved by two different hybrid approaches analyzing their applications’ potentiality on real world paradigms (speech recognition and medical diagnosis). We compare the performance obtained with (1) Multinetwork RBF/LVQ structure, we use involves Learning Vector Quantization (LVQ) as a competitive decision processor and Rad...
متن کاملHybrid System of Optimal Self Organizing Maps and Hidden Markov Model for Arabic Digits Recognition
Thanks to Automatic Speech Recognition (ASR), a lot of machines can nowadays emulate human being ability to understand and speak natural language. However, ASR problematic could be as interesting as it is difficult. Its difficulty is precisely due to the complexity of speech processing, which takes into consideration many aspects: acoustic, phonetic, syntactic, etc. Thus, the most commonly used...
متن کاملIndex assignment in vector quantisation for channels with memory - Vision, Image and Signal Processing, IEE Proceedings-
The study prcsents a formulation of a Hadaniard framework for analysing the vector quantisation aver channels with memory. In seeking faster response, classes of index assignments are defined in terms of the Hadamard transform of channel transition probabilities, An index assignment algorithm is developed that achieves high robustness against channel errors, and its performance i n vector quant...
متن کامل