The adoption of high-accuracy speech recognition algorithms without an effective evaluation of their impact on the target computational resource is impractical for mobile and embedded systems In this paper, techniques are adopted to minimise the required computational resource for an effective mobile-based speech recognition system A Dynamic Multi-Layer Perceptron speech recognition technique, capable of running in real time on a state-of-the-art mobile device, has been introduced Even though a conventional hidden Markov model when applied to the same dataset slightly outperformed our approach, its processing time is much higher The Dynamic Multi-layer Perceptron presented here has an accuracy level of 9694% and runs significantly faster than similar techniques

/pdf/a-comparative-review-of-dynamic-neural-networks-and-hidden-4lloojt2n4.pdf

A comparative review of dynamic neural networks and hidden Markov model methods for mobile on-device speech recognition

This paper explains the concept of Automatic speech recognition (ASR) from the view point of pattern recognition. An ASR system can be broadly classified into two parts: front end and the back end which are responsible for feature extraction and acoustic modelling respectively. The presented paper elaborates and compares all popular feature extraction and acoustic modelling techniques along with the challenges and advancements in the field of ASR.

http://ijarcsee.org/index.php/IJARCSEE/article/view/440/407

Automatic Speech Recognition: A Survey

This paper proposes a new method to embed data in compressed binary images (JPEG), including scanned text, figures, and signatures. The method manipulates "fiippable" pixels to embed data without distorting image. Shuffling is applied before embedding to provide more security. The hidden data can be extracted without using the original image. The proposed method uses look up table which is to be used for extraction of data at other end is made more compact, secure. After applying proposed method the image will be distorted no drought but human eye will not recognize the change in the image. The proposed data embedding method can be used to detect unauthorized use of a digitized signature, and annotate or authenticate binary documents.

A Novel Scheme of Data Hiding in Binary Images

Accurate vowel recognition forms the backbone of most successful speech recognition systems. A collection of techniques exists to extract the relevant features from the steady-state regions of the vowels both in time as well as in frequency domains. In this paper we present a novel and accurate feature extraction technique for recognizing Malayalam spoken vowels based on Linear Predictive Coding method and compared the result with wavelet packet decomposition method. Recognition is performed using k-NN pattern classifier. The classification is conducted for 5 Malayalam vowel sounds using training and test set consisting of 50 ( 10 from each class) samples each. The overall recognition accuracy obtained for the vowel using LPC feature extraction method is 94%. The proposed method is efficient and computationally less expensive. The experimental results demonstrate the efficiency of the proposed algorithm

Malayalam Vowel Recognition Based on Linear Predictive Coding Parameters and k-NN Algorithm

Mutual-learning sequence-level knowledge distillation for automatic speech recognition

In this paper, the improved linear predictive coding (LPC) coefficients of the frame are employed in the feature extraction method. In the proposed speech recognition system, the static LPC coefficients + dynamic LPC coefficients of the frame were employed as a basic feature. The framework of linear discriminant analysis (LDA) is used to derive an efficient and reduced-dimension speech parametric speech vector space for the speech recognition system. Using the continuous hidden Markov model (HMM) as the speech recognition model, the speech recognition system was successfully constructed. Experiments are performed on the isolated-word speech recognition task. It is found that the improved LPC feature extraction method is quite efficient.

Er Meng Joo

Papers

Improved linear predictive coding method for speech recognition