A Study on Speech Classification Based on Deep Neural Network under Adverse Environments
Abstract
Recently, speech classification plays an important role in real-world speech application such as mobile devices and voice control system of a car.However, additive noise and reverberation under adverse environments can degrade the performance of speech classification.In this thesis, author focuses on improving speech classification considering additive noise and reverberation under adverse environments by using deep neural network (DNN).Two speech classification tasks: accent recognition and voice activity detection (VAD) are considered.The accent recognition task is investigated under reverberant environment while the VAD task will be examined under noisy environment.For accent recognition task, DNN-based accent recognition is proposed for modeling the reverberation over multi-frame.Moreover, the combination of scores of DNN-based accent recognition and Gaussian Mixture Model (GMM)-based accent recognition is also proposed to utilize complementary characteristics of these two models.In reverberant environment, our proposed method outperformed conventional GMM-based accent recognition.The accent recognition rate was improved from 90.7 % with GMM to 93.0 % with DNN.The combination of GMM and DNN achieved recognition rate of 97.5 %, which outperformed than the individual GMM and DNN because the complementation of GMM and DNN.At first, I would like to express my gratitude to my supervisor Professor Longbiao WANG and Professor
Keywords
Not available
Specialisations
Similar theses
- Single-Channel Speech Enhancement Based on Deep Neural Networks Zhiheng Ouyang · Concordia University · 2019
- Training Methods for Deep Neural Network-Based Acoustic Models in Speech Recognition Tamás Grósz · University of Szeged · 2018
- Robust speech recognition in noisy and reverberant environments using deep neural network-based systems Novoa Ilic ; José Eduardo · Universidad de Chile · 2018
- Automatic speech recognition and machine translation with deep neural networks for open educational resources, parliamentary contents and broadcast media Gonzalo Vicente Garcés Díaz-Munío · Universitat Politècnica de València · 2024
- A Deep Neural Network Approach Integrating CNN and BiLSTM-Transformer Architectures for Emotion Recognition from Speech Divyansh Anand · 2024
- Streaming Automatic Speech Recognition with Hybrid Architectures and Deep Neural Network Models Javier Jorge Cano · Universitat Politècnica de València · 2022
