An Automatic Dysarthric Speech Recognition Approach using Deep Neural Networks
Journal Title: International Journal of Advanced Computer Science & Applications - Year 2017, Vol 8, Issue 12
Abstract
Transcribing dysarthric speech into text is still a challenging problem for the state-of-the-art techniques or commercially available speech recognition systems. Improving the accuracy of dysarthric speech recognition, this paper adopts Deep Belief Neural Networks (DBNs) to model the distribution of dysarthric speech signal. A continuous dysarthric speech recognition system is produced, in which the DBNs are used to predict the posterior probabilities of the states in Hidden Markov Models (HMM) and the Weighted Finite State Transducers framework was utilized to build the speech decoder. Experimental results show that the proposed method provides better prediction of the probability distribution of the spectral representation of dysarthric speech that outperforms the existing methods, e.g., GMM-HMM based dysarthric speech recogniztion approaches. To the best of our knowledge, this work is the first time to build a continuous speech recognition system for dysarthric speech with deep neural network technique, which is a promising approach for improving the communication between those individuals with speech impediments and normal speakers.
Authors and Affiliations
Jun Ren, Mingzhe Liu
Keywords
Related Articles
- EP ID EP251594
- DOI 10.14569/IJACSA.2017.081207
- Views 110
- Downloads 0
Improving and Extending Indoor Connectivity Using Relay Nodes for 60 GHz Applications
A 60 GHz wireless system can provide very high data rates. However, it has a tremendous amount of both Free Space Path Loss (FSPL) and penetration loss. To mitigate these losses and extend the system range; we propose te...
Deep Learning Features Fusion with Classical Image Features for Image Access
Depending on the society, the access to the adult content can create social problems. This paper thus proposes a fusion approach for image based adult content filtering. The proposed approach merges the Deep Learning (DL...
Switched Control of a Time Delayed Compass Gait Robot
the analysis and control of delayed systems are becoming more and more research topics in progress. This is mainly due to the fact that the delay is frequently encountered in technological systems. Most control command l...
Connected Dominating Set based Optimized Routing Protocol for Wireless Sensor Networks
Wireless Sensor Networks(WSNs) have problem of energy starvation in their operations. This constraint demands that the topology of communicating nodes should be limited. One of the ways of restraining the communicating n...
Personalized E-Learning Recommender System using Multimedia Data
Due to the huge amounts of online learning materials, e-learning environments are becoming very popular as means of delivering lectures. One of the most common e-learning challenges is how to recommend quality learning m...