Abstract: In this era of technology, fueled by the pervasive usage of the internet, security is a prime concern. The number of new attacks by the so-called “bots", which are automated programs, is increasing at an alarming rate. They are most likely to attack online registration systems. Technology, called “CAPTCHA" (Completely Automated Public Turing test to tell Computers and Humans Apart) do exist, which can differentiate between automated programs and humans and prevent replay attacks. Traditionally CAPTCHA-s have been implemented with the challenge involved in recognizing textual images and reproducing the same. We propose an approach where the visual challenge has to be read out from which randomly selected keywords are used to verify the correctness of spoken text and in turn detect the presence of human. This is supplemented with a speaker recognition system which can identify the speaker also. Thus, this framework fulfills both the objectives – it can determine whether the user is a human or not and if it is a human, it can verify its identity.
Abstract: We report in this paper the procedure of a system of
automatic speech recognition based on techniques of the dynamic
programming. The technique of temporal retiming is a technique
used to synchronize between two forms to compare. We will see how
this technique is adapted to the field of the automatic speech
recognition. We will expose, in a first place, the theory of the
function of retiming which is used to compare and to adjust an
unknown form with a whole of forms of reference constituting the
vocabulary of the application. Then we will give, in the second place,
the various algorithms necessary to their implementation on machine.
The algorithms which we will present were tested on part of the
corpus of words in Arab language Arabdic-10 [4] and gave whole
satisfaction. These algorithms are effective insofar as we apply them
to the small ones or average vocabularies.
Abstract: A set of Artificial Neural Network (ANN) based methods
for the design of an effective system of speech recognition of
numerals of Assamese language captured under varied recording
conditions and moods is presented here. The work is related to
the formulation of several ANN models configured to use Linear
Predictive Code (LPC), Principal Component Analysis (PCA) and
other features to tackle mood and gender variations uttering numbers
as part of an Automatic Speech Recognition (ASR) system in
Assamese. The ANN models are designed using a combination of
Self Organizing Map (SOM) and Multi Layer Perceptron (MLP)
constituting a Learning Vector Quantization (LVQ) block trained in a
cooperative environment to handle male and female speech samples
of numerals of Assamese- a language spoken by a sizable population
in the North-Eastern part of India. The work provides a comparative
evaluation of several such combinations while subjected to handle
speech samples with gender based differences captured by a microphone
in four different conditions viz. noiseless, noise mixed, stressed
and stress-free.
Abstract: A combined three-microphone voice activity detector (VAD) and noise-canceling system is studied to enhance speech recognition in an automobile environment. A previous experiment clearly shows the ability of the composite system to cancel a single noise source outside of a defined zone. This paper investigates the performance of the composite system when there are frequently moving noise sources (noise sources are coming from different locations but are not always presented at the same time) e.g. there is other passenger speech or speech from a radio when a desired speech is presented. To work in a frequently moving noise sources environment, whilst a three-microphone voice activity detector (VAD) detects voice from a “VAD valid zone", the 3-microphone noise canceller uses a “noise canceller valid zone" defined in freespace around the users head. Therefore, a desired voice should be in the intersection of the noise canceller valid zone and VAD valid zone. Thus all noise is suppressed outside this intersection of area. Experiments are shown for a real environment e.g. all results were recorded in a car by omni-directional electret condenser microphones.
Abstract: Performance of any continuous speech recognition system is highly dependent on performance of the acoustic models. Generally, development of the robust spoken language technology relies on the availability of large amounts of data. Common way to cope with little data for training each state of Markov models is treebased state tying. This tying method applies contextual questions to tie states. Manual procedure for question generation suffers from human errors and is time consuming. Various automatically generated questions are used to construct decision tree. There are three approaches to generate questions to construct HMMs based on decision tree. One approach is based on misrecognized phonemes, another approach basically uses feature table and the other is based on state distributions corresponding to context-independent subword units. In this paper, all these methods of automatic question generation are applied to the decision tree on FARSDAT corpus in Persian language and their results are compared with those of manually generated questions. The results show that automatically generated questions yield much better results and can replace manually generated questions in Persian language.
Abstract: In this paper, an algorithm for detecting and attenuating
puff noises frequently generated under the mobile environment is
proposed. As a baseline system, puff detection system is designed
based on Gaussian Mixture Model (GMM), and 39th Mel Frequency
Cepstral Coefficient (MFCC) is extracted as feature parameters. To
improve the detection performance, effective acoustic features for puff
detection are proposed. In addition, detected puff intervals are
attenuated by high-pass filtering. The speech recognition rate was
measured for evaluation and confusion matrix and ROC curve are used
to confirm the validity of the proposed system.
Abstract: Computerized lip reading has been one of the most
actively researched areas of computer vision in recent past because
of its crime fighting potential and invariance to acoustic environment.
However, several factors like fast speech, bad pronunciation,
poor illumination, movement of face, moustaches and beards make
lip reading difficult. In present work, we propose a solution for
automatic lip contour tracking and recognizing letters of English
language spoken by speakers using the information available from
lip movements. Level set method is used for tracking lip contour
using a contour velocity model and a feature vector of lip movements
is then obtained. Character recognition is performed using modified
k nearest neighbor algorithm which assigns more weight to nearer
neighbors. The proposed system has been found to have accuracy
of 73.3% for character recognition with speaker lip movements as
the only input and without using any speech recognition system in
parallel. The approach used in this work is found to significantly
solve the purpose of lip reading when size of database is small.
Abstract: Despite the fact that Arabic language is currently one
of the most common languages worldwide, there has been only a
little research on Arabic speech recognition relative to other
languages such as English and Japanese. Generally, digital speech
processing and voice recognition algorithms are of special
importance for designing efficient, accurate, as well as fast automatic
speech recognition systems. However, the speech recognition process
carried out in this paper is divided into three stages as follows: firstly,
the signal is preprocessed to reduce noise effects. After that, the
signal is digitized and hearingized. Consequently, the voice activity
regions are segmented using voice activity detection (VAD)
algorithm. Secondly, features are extracted from the speech signal
using Mel-frequency cepstral coefficients (MFCC) algorithm.
Moreover, delta and acceleration (delta-delta) coefficients have been
added for the reason of improving the recognition accuracy. Finally,
each test word-s features are compared to the training database using
dynamic time warping (DTW) algorithm. Utilizing the best set up
made for all affected parameters to the aforementioned techniques,
the proposed system achieved a recognition rate of about 98.5%
which outperformed other HMM and ANN-based approaches
available in the literature.
Abstract: An emotional speech recognition system for the
applications on smart phones was proposed in this study to combine
with 3G mobile communications and social networks to provide users
and their groups with more interaction and care. This study developed
a mechanism using the support vector machines (SVM) to recognize
the emotions of speech such as happiness, anger, sadness and normal.
The mechanism uses a hierarchical classifier to adjust the weights of
acoustic features and divides various parameters into the categories of
energy and frequency for training. In this study, 28 commonly used
acoustic features including pitch and volume were proposed for
training. In addition, a time-frequency parameter obtained by
continuous wavelet transforms was also used to identify the accent and
intonation in a sentence during the recognition process. The Berlin
Database of Emotional Speech was used by dividing the speech into
male and female data sets for training. According to the experimental
results, the accuracies of male and female test sets were increased by
4.6% and 5.2% respectively after using the time-frequency parameter
for classifying happy and angry emotions. For the classification of all
emotions, the average accuracy, including male and female data, was
63.5% for the test set and 90.9% for the whole data set.
Abstract: Automated operations based on voice commands will become more and more important in many applications, including robotics, maintenance operations, etc. However, voice command recognition rates drop quite a lot under non-stationary and chaotic noise environments. In this paper, we tried to significantly improve the speech recognition rates under non-stationary noise environments. First, 298 Navy acronyms have been selected for automatic speech recognition. Data sets were collected under 4 types of noisy environments: factory, buccaneer jet, babble noise in a canteen, and destroyer. Within each noisy environment, 4 levels (5 dB, 15 dB, 25 dB, and clean) of Signal-to-Noise Ratio (SNR) were introduced to corrupt the speech. Second, a new algorithm to estimate speech or no speech regions has been developed, implemented, and evaluated. Third, extensive simulations were carried out. It was found that the combination of the new algorithm, the proper selection of language model and a customized training of the speech recognizer based on clean speech yielded very high recognition rates, which are between 80% and 90% for the four different noisy conditions. Fourth, extensive comparative studies have also been carried out.
Abstract: Heart sound is an acoustic signal and many techniques
used nowadays for human recognition tasks borrow speech recognition
techniques. One popular choice for feature extraction of accoustic
signals is the Mel Frequency Cepstral Coefficients (MFCC) which
maps the signal onto a non-linear Mel-Scale that mimics the human
hearing. However the Mel-Scale is almost linear in the frequency
region of heart sounds and thus should produce similar results with
the standard cepstral coefficients (CC). In this paper, MFCC is
investigated to see if it produces superior results for PCG based
human identification system compared to CC. Results show that the
MFCC system is still superior to CC despite linear filter-banks in
the lower frequency range, giving up to 95% correct recognition rate
for MFCC and 90% for CC. Further experiments show that the high
recognition rate is due to the implementation of filter-banks and not
from Mel-Scaling.
Abstract: We report in this paper the model adopted by our
system of continuous speech recognition in Arab language SySRA
and the results obtained until now. This system uses the database
Arabdic-10 which is a corpus of word for the Arab language and
which was manually segmented. Phonetic decoding is represented
by an expert system where the knowledge base is translated in the
form of production rules. This expert system transforms a vocal
signal into a phonetic lattice. The higher level of the system takes
care of the recognition of the lattice thus obtained by deferring it in
the form of written sentences (orthographical Form). This level
contains initially the lexical analyzer which is not other than the
module of recognition. We subjected this analyzer to a set of
spectrograms obtained by dictating a score of sentences in Arab
language. The rate of recognition of these sentences is about 70%
which is, to our knowledge, the best result for the recognition of the
Arab language. The test set consists of twenty sentences from four
speakers not having taken part in the training.
Abstract: An automatic speech recognition system for the
formal Arabic language is needed. The Quran is the most formal
spoken book in Arabic, it is spoken all over the world. In this
research, an automatic speech recognizer for Quranic based speakerindependent
was developed and tested. The system was developed
based on the tri-phone Hidden Markov Model and Maximum
Likelihood Linear Regression (MLLR). The MLLR computes a set
of transformations which reduces the mismatch between an initial
model set and the adaptation data. It uses the regression class tree, as
well as, estimates a set of linear transformations for the mean and
variance parameters of a Gaussian mixture HMM system. The 30th
Chapter of the Quran, with five of the most famous readers of the
Quran, was used for the training and testing of the data. The chapter
includes about 2000 distinct words. The advantages of using the
Quranic verses as the database in this developed recognizer are the
uniqueness of the words and the high level of orderliness between
verses. The level of accuracy from the tested data ranged 68 to 85%.
Abstract: This paper discusses the cued speech recognition
methods in videoconference. Cued speech is a specific gesture
language that is used for communication between deaf people. We
define the criteria for sentence intelligibility according to answers of
testing subjects (deaf people). In our tests we use 30 sample videos
coded by H.264 codec with various bit-rates and various speed of
cued speech. Additionally, we define the criteria for consonant sign
recognizability in single-handed finger alphabet (dactyl) analogically
to acoustics. We use another 12 sample videos coded by H.264 codec
with various bit-rates in four different video formats. To interpret the
results we apply the standard scale for subjective video quality
evaluation and the percentual evaluation of intelligibility as in
acoustics. From the results we construct the minimum coded bit-rate
recommendations for every spatial resolution.
Abstract: This research work is aimed at speech recognition
using scaly neural networks. A small vocabulary of 11 words were
established first, these words are “word, file, open, print, exit, edit,
cut, copy, paste, doc1, doc2". These chosen words involved with
executing some computer functions such as opening a file, print
certain text document, cutting, copying, pasting, editing and exit.
It introduced to the computer then subjected to feature extraction
process using LPC (linear prediction coefficients). These features are
used as input to an artificial neural network in speaker dependent
mode. Half of the words are used for training the artificial neural
network and the other half are used for testing the system; those are
used for information retrieval.
The system components are consist of three parts, speech
processing and feature extraction, training and testing by using neural
networks and information retrieval.
The retrieve process proved to be 79.5-88% successful, which is
quite acceptable, considering the variation to surrounding, state of
the person, and the microphone type.
Abstract: The practical implementation of audio-video coupled speech recognition systems is mainly limited by the hardware complexity to integrate two radically different information capturing devices with good temporal synchronisation. In this paper, we propose a solution based on a smart CMOS image sensor in order to simplify the hardware integration difficulties. By using on-chip image processing, this smart sensor can calculate in real time the X/Y projections of the captured image. This on-chip projection reduces considerably the volume of the output data. This data-volume reduction permits a transmission of the condensed visual information via the same audio channel by using a stereophonic input available on most of the standard computation devices such as PC, PDA and mobile phones. A prototype called VMIKE (Visio-Microphone) has been designed and realised by using standard 0.35um CMOS technology. A preliminary experiment gives encouraged results. Its efficiency will be further investigated in a large variety of applications such as biometrics, speech recognition in noisy environments, and vocal control for military or disabled persons, etc.
Abstract: The Automatic Speech Recognition (ASR) applied to
Arabic language is a challenging task. This is mainly related to the
language specificities which make the researchers facing multiple
difficulties such as the insufficient linguistic resources and the very
limited number of available transcribed Arabic speech corpora. In
this paper, we are interested in the development of a HMM-based
ASR system for Standard Arabic (SA) language. Our fundamental
research goal is to select the most appropriate acoustic parameters
describing each audio frame, acoustic models and speech recognition
unit. To achieve this purpose, we analyze the effect of varying frame
windowing (size and period), acoustic parameter number resulting
from features extraction methods traditionally used in ASR, speech
recognition unit, Gaussian number per HMM state and number of
embedded re-estimations of the Baum-Welch Algorithm. To evaluate
the proposed ASR system, a multi-speaker SA connected-digits
corpus is collected, transcribed and used throughout all experiments.
A further evaluation is conducted on a speaker-independent continue
SA speech corpus. The phonemes recognition rate is 94.02% which is
relatively high when comparing it with another ASR system
evaluated on the same corpus.
Abstract: Cluster analysis is the name given to a diverse collection of techniques that can be used to classify objects (e.g. individuals, quadrats, species etc). While Kohonen's Self-Organizing Feature Map (SOFM) or Self-Organizing Map (SOM) networks have been successfully applied as a classification tool to various problem domains, including speech recognition, image data compression, image or character recognition, robot control and medical diagnosis, its potential as a robust substitute for clustering analysis remains relatively unresearched. SOM networks combine competitive learning with dimensionality reduction by smoothing the clusters with respect to an a priori grid and provide a powerful tool for data visualization. In this paper, SOM is used for creating a toroidal mapping of two-dimensional lattice to perform cluster analysis on results of a chemical analysis of wines produced in the same region in Italy but derived from three different cultivators, referred to as the “wine recognition data" located in the University of California-Irvine database. The results are encouraging and it is believed that SOM would make an appealing and powerful decision-support system tool for clustering tasks and for data visualization.
Abstract: In this paper we present an enhanced noise reduction method for robust speech recognition using Adaptive Gain Equalizer with Non linear Spectral Subtraction. In Adaptive Gain Equalizer method (AGE), the input signal is divided into a number of subbands that are individually weighed in time domain, in accordance to the short time Signal-to-Noise Ratio (SNR) in each subband estimation at every time instant. Instead of focusing on suppression the noise on speech enhancement is focused. When analysis was done under various noise conditions for speech recognition, it was found that Adaptive Gain Equalizer method algorithm has an obvious failing point for a SNR of -5 dB, with inadequate levels of noise suppression for SNR less than this point. This work proposes the implementation of AGE when coupled with Non linear Spectral Subtraction (AGE-NSS) for robust speech recognition. The experimental result shows that out AGE-NSS performs the AGE when SNR drops below -5db level.
Abstract: This paper proposes evaluation of sound parameterization methods in recognizing some spoken Arabic words, namely digits from zero to nine. Each isolated spoken word is represented by a single template based on a specific recognition feature, and the recognition is based on the Euclidean distance from those templates. The performance analysis of recognition is based on four parameterization features: the Burg Spectrum Analysis, the Walsh Spectrum Analysis, the Thomson Multitaper Spectrum Analysis and the Mel Frequency Cepstral Coefficients (MFCC) features. The main aim of this paper was to compare, analyze, and discuss the outcomes of spoken Arabic digits recognition systems based on the selected recognition features. The results acqired confirm that the use of MFCC features is a very promising method in recognizing Spoken Arabic digits.