S
S.M. Mirrezaie
Researcher at Amirkabir University of Technology
Publications - 6
Citations - 18
S.M. Mirrezaie is an academic researcher from Amirkabir University of Technology. The author has contributed to research in topics: Speaker recognition & Speaker diarisation. The author has an hindex of 3, co-authored 6 publications receiving 18 citations.
Papers
More filters
Proceedings ArticleDOI
Weighting of Mel Sub-bands Based on SNR/Entropy for Robust ASR
TL;DR: Experimental results indicate that the proposed set of noise-robust features based on conventional MFCC feature extraction method leads to improved ASR performance in noisy environments and its computational overhead is quite small.
Proceedings ArticleDOI
Spoken Language Identification Using a New Sequence Kernel-based SVM Back-end Classifier
TL;DR: This paper presents a new back-end classifier for GMM-LM based language identification systems, consisting of two main parts, mapping matrix and bank of SVMs, and shows that the new sequence kernel-based SVMs separate languages more efficiently than common Gaussian mixture and GLDS SVM back- end classifiers.
Proceedings ArticleDOI
Speaker diarization in a multi-speaker environment using particle swarm optimization and mutual information
TL;DR: An approach comprising of PSO (particle swarm optimization) algorithm, which encodes possible segmentations of an audio record by measuring mutual information between these segments and the audio data which is used as the fitness function for the PSO.
Proceedings ArticleDOI
Robust Speaker Diarization in a Multi-Speaker Environment Using Autocorrelation-based Noise Subtraction
TL;DR: This paper addresses the robustness issue by using a method already successful in speech recognition application, ANS (Autocorrelation-Based Noise Subtraction) for robust genetic algorithm-based speaker diarization, and compares the results with the baseline MFCC-based system in clean and noisy conditions.
Weighting ofMelSub-bands BasedonSNR/Entropy
TL;DR: This paper proposes a set of within particular frequency- bands bysplitting thefrequency noise-robust features based on conventional MFCC featurespectrum intosub-bands by splitting thefrequency Noise-Robust featuresbased on conventional cepstral coefficients, and proposes aframework tocompensate additive noise effects on I.