Wyniki wyszukiwania - BazTech

1

Genetic Algorithm for Combined Speaker and Speech Recognition using Deep Neural Networks

Kaur G., Srivastava M., Kumar A.

Journal of Telecommunications and Information Technology

|

2018

|

nr 2

23--31

EN

Huge growth is observed in the speech and speaker recognition ﬁeld due to many artiﬁcial intelligence algorithms being applied. Speech is used to convey messages via the language being spoken, emotions, gender and speaker identity. Many real applications in healthcare are based upon speech and speaker recognition, e.g. a voice-controlled wheelchair helps control the chair. In this paper, we use a genetic algorithm (GA) for combined speaker and speech recognition, relying on optimized Mel Frequency Cepstral Coeﬃcient (MFCC) speech features, and classiﬁcation is performed using a Deep Neural Network (DNN). In the ﬁrst phase, feature extraction using MFCC is executed. Then, feature optimization is performed using GA. In the second phase training is conducted using DNN. Evaluation and validation of the proposed work model is done by setting a real environment, and eﬃciency is calculated on the basis of such parameters as accuracy, precision rate, recall rate, sensitivity, and speciﬁcity. Also, this paper presents an evaluation of such feature extraction methods as linear predictive coding coefficient (LPCC), perceptual linear prediction (PLP), mel frequency cepstral coefﬁcients (MFCC) and relative spectra ﬁltering (RASTA), with all of them used for combined speaker and speech recognition systems. A comparison of diﬀerent methods based on existing techniques for both clean and noisy environments is made as well.

2

A class of neuro-computational methods for assamese fricative classification

Patgiri C., Sarma M., Sarma K. K.

Journal of Artificial Intelligence and Soft Computing Research

|

2015

|

Vol. 5, No. 1

59--70

EN

In this work, a class of neuro-computational classifiers are used for classification of fricative phonemes of Assamese language. Initially, a Recurrent Neural Network (RNN) based classifier is used for classification. Later, another neuro fuzzy classifier is used for classification. We have used two different feature sets for the work, one using the specific acoustic-phonetic characteristics and another temporal attributes using linear prediction cepstral coefficients (LPCC) and a Self Organizing Map (SOM). Here, we present the experimental details and performance difference obtained by replacing the RNN based classifier with an adaptive neuro fuzzy inference system (ANFIS) based block for both the feature sets to recognize Assamese fricative sounds.

3

Diagnostyka silnika synchronicznego oparta na rozpoznawaniu dźwięku z zastosowaniem LPCC i GSDM

Głowacz A., Głowacz W.

Pomiary Automatyka Kontrola

|

2010

|

R. 56, nr 5

479-482

PL

Zaprezentowano koncepcję badania sygnałów akustycznych stanów przedawaryjnych silnika synchronicznego. Oprogramowanie do rozpoznawania dźwięku zostało zaimplementowane. Algorytmy przetwarzania i analizy sygnałów akustycznych zostały zastosowane. System jest oparty na algorytmie LPCC (Współczynniki cepstralne liniowego kodowania) i GSDM (Genetyczna rozrzedzona pamięć rozproszona). Badania zostały przeprowadzone dla sygnałów akustycznych stanów przedawaryjnych. Zmiany w sygnale akustycznym spowodowane były przez zwarcia i przerwy w obwodzie stojana. Analiza wyników pokazuje wrażliwość metody opartej na LPCC i GSDM w zależności od danych wejściowych. Wyniki badań potwierdzają poprawne działanie systemu rozpoznawania dźwięku silnika synchronicznego.

EN

In recent years the methods of sound recognition have been de-veloped. Hence, there is an idea to use them in case of machines. The paper describes the concept of investigations of acoustic signals of synchronous motor imminent failure conditions. Measurements were taken with a recorder OLYMPUS WS-200S. Sound recognition software was implemented. Algorithms of signal processing and analysis were used. The system is based on the LPCC (Linear Predictive Cepstrum Coefficients) algorithm and GSDM (Genetic Sparse Distributed Memory). Investigations were carried out for acoustic signals of imminent failure conditions. The following plan of investigations of a synchronous motor acoustic signal was proposed: recording of audio track, sound track division, sampling, quantization, normalization, filtration, windowing, feature extraction, classification (Fig. 2). Figs. 3, 4, 5 and 6 show changes of the LPCC values for four types of the categories recognized. Changes in the acoustic signal were caused by short circuit and broken coils in the stator circuit. The sound recognition efficiency depending on the acoustic signal and the sample length is presented in Fig. 8. The sound recognition system was built for a synchronous motor. There were used 39 band-pass filters in investigations. Analysis of the results shows the sensitivity of the method based on LPCC and GSDM, depending on the input data. The results confirm correct operation of the synchronous motor sound recognition system. These studies can be used for diagnostics based on acoustic emission in electrical, mechanical, hydraulic and pneumatic machines.

4

Visualization of stages of determining cepstral factors in speech recognition systems

Proksa R.

Journal of Medical Informatics & Technologies

|

2009

|

Vol. 13

121--128

EN

The article presents two methods of determination of cepstral parameters commonly applied in digital signal processing, in particular in speech recognition systems. The solutions presented are part of a project aimed at developing applications allowing to control the Windows operating system with voice and the use of MSAA (Microsoft Active Accessibility). The analysed voice signal has been visually presented at each of the crucial stages of developing cepstral coefficients.