Search results for: speech signal analysis

Authors: Nazaket Gazieva

Abstract:

Voice biometric data associated with physiological, psychological and other factors are widely used in forensic phonoscopy. There are various methods for identifying and verifying a person by voice. This article explores the minimum speech signal data as individual parameters of a speech signal. Monozygotic twins are believed to be genetically identical. Using the minimum data of the speech signal, we came to the conclusion that the voice imprint of monozygotic twins is individual. According to the conclusion of the experiment, we can conclude that the minimum indicators of the speech signal are more stable and reliable for phonoscopic examinations.

Keywords: phonogram, speech signal, temporal characteristics, fundamental frequency, biometric fingerprints

Procedia PDF Downloads 144

Authors: Eng. Alaa K. Satti Salih

Abstract:

Revolutions Applications such as telecommunications, hands-free communications, recording, etc. which need at least one microphone, the signal is usually infected by noise and echo. The important application is the speech enhancement, which is done to remove suppressed noises and echoes taken by a microphone, beside preferred speech. Accordingly, the microphone signal has to be cleaned using digital signal processing DSP tools before it is played out, transmitted, or stored. Engineers have so far tried different approaches to improving the speech by get back the desired speech signal from the noisy observations. Especially Mobile communication, so in this paper will do reconstruction of the speech signal, observed in additive background noise, using the Kalman filter technique to estimate the parameters of the Autoregressive Process (AR) in the state space model and the output speech signal obtained by the MATLAB. The accurate estimation by Kalman filter on speech would enhance and reduce the noise then compare and discuss the results between actual values and estimated values which produce the reconstructed signals.

Keywords: autoregressive process, Kalman filter, Matlab, noise speech

Procedia PDF Downloads 345

Authors: Monica Sharma, Harjit Pal Singh, Jasbinder Singh, Manju Bala

Abstract:

In Voice over Internet Protocol (VoIP) system, the speech signal is degraded when passed through the network layers. The speech signal is processed through the best effort policy based IP network, which leads to the network degradations including delay, packet loss and jitter. The packet loss is the major issue of the degradation in the VoIP signal quality; even a single lost packet may generate audible distortion in the decoded speech signal. In addition to these network degradations, the quality of the speech signal is also affected by the environmental noises and coder distortions. The signal quality of the VoIP system is improved through the interleaving technique. The performance of the system is evaluated for various types of noises at different network conditions. The performance of the enhanced VoIP signal is evaluated using perceptual evaluation of speech quality (PESQ) measurement for narrow band signal.

Keywords: VoIP, interleaving, packet loss, packet size, background noise

Procedia PDF Downloads 481

Authors: Sergei Aleinik, Yuri Matveev

Abstract:

In this paper a novel method for the detection of clipping in speech signals is described. It is shown that the new method has better performance than known clipping detection methods, is easy to implement, and is robust to changes in signal amplitude, size of data, etc. Statistical simulation results are presented.

Keywords: clipping, clipped signal, speech signal processing, digital signal processing

Procedia PDF Downloads 394

Authors: M. A. Ben Messaoud, A. Bouzid, N. Ellouze

Abstract:

Speech Segmentation is the measure of the change point detection for partitioning an input speech signal into regions each of which accords to only one speaker. In this paper, we apply two features based on multi-scale product (MP) of the clean speech, namely the spectral centroid of MP, and the zero crossings rate of MP. We focus on multi-scale product analysis as an important tool for segmentation extraction. The multi-scale product is based on making the product of the speech wavelet transform coefficients at three successive dyadic scales. We have evaluated our method on the Keele database. Experimental results show the effectiveness of our method presenting a good performance. It shows that the two simple features can find word boundaries, and extracted the segments of the clean speech.

Keywords: multiscale product, spectral centroid, speech segmentation, zero crossings rate

Procedia PDF Downloads 501

Authors: Harisudha Kuresan, Dhanalakshmi Samiappan, T. Rama Rao

Abstract:

In modern telecommunication applications, Glottal Closure Instants location finding is important and is directly evaluated from the speech waveform. Here, we study the GCI using Speech Event Detection using Residual Excitation and the Mean Based Signal (SEDREAMS) algorithm. Speech coding uses parameter estimation using audio signal processing techniques to model the speech signal combined with generic data compression algorithms to represent the resulting modeled in a compact bit stream. This paper proposes a sub-band coder SBC, which is a type of transform coding and its performance for GCI detection using SEDREAMS are evaluated. In SBCs code in the speech signal is divided into two or more frequency bands and each of these sub-band signal is coded individually. The sub-bands after being processed are recombined to form the output signal, whose bandwidth covers the whole frequency spectrum. Then the signal is decomposed into low and high-frequency components and decimation and interpolation in frequency domain are performed. The proposed structure significantly reduces error, and precise locations of Glottal Closure Instants (GCIs) are found using SEDREAMS algorithm.

Keywords: SEDREAMS, GCI, SBC, GOI

Procedia PDF Downloads 358

Authors: Rajani S. Pujar, Pandurangarao N. Kulkarni

Abstract:

Increased temporal masking affects the speech perception in persons with sensorineural hearing impairment especially under adverse listening conditions. This paper presents a cascaded scheme, which employs a noise reduction algorithm as well as temporal splitting of the speech signal. Earlier investigations have shown that by splitting the speech temporally and presenting alternate segments to the two ears help in reducing the effect of temporal masking. In this technique, the speech signal is processed by two fading functions, complementary to each other, and presented to left and right ears for binaural dichotic presentation. In the present study, half cosine signal is used as a fading function with crossover gain of 6 dB for the perceptual balance of loudness. Temporal splitting is combined with noise reduction algorithm to improve speech perception in the background noise. Two noise reduction schemes, namely spectral subtraction and Wiener filter are used. Listening tests were conducted on six normal-hearing subjects, with sensorineural loss simulated by adding broadband noise to the speech signal at different signal-to-noise ratios (∞, 3, 0, and -3 dB). Objective evaluation using PESQ was also carried out. The MOS score for VCV syllable /asha/ for SNR values of ∞, 3, 0, and -3 dB were 5, 4.46, 4.4 and 4.05 respectively, while the corresponding MOS scores for unprocessed speech were 5, 1.2, 0.9 and 0.65, indicating significant improvement in the perceived speech quality for the proposed scheme compared to the unprocessed speech.

Keywords: MOS, PESQ, spectral subtraction, temporal splitting, wiener filter

Procedia PDF Downloads 328

Authors: Ajish K. Abraham

Abstract:

Hearing aid is the most fundamental technology employed towards rehabilitation of persons with sensory neural hearing impairment. Hearing in noise is still a matter of major concern for many hearing aid users and thus continues to be a challenging issue for the hearing aid designers. Several techniques are being currently used to enhance the speech at the hearing aid output. Most of these techniques, when implemented, result in reduction of intelligibility of the speech signal. Thus the dissatisfaction of the hearing aid user towards comprehending the desired speech amidst noise is prevailing. Multichannel Wiener Filter is widely implemented in binaural hearing aid technology for noise reduction. In this study, Wiener filter based noise reduction approach is experimented for a single microphone based hearing aid set up. This method checks the status of the input speech signal in each frequency band and then selects the relevant noise reduction procedure. Results showed that the Wiener filter based algorithm is capable of enhancing speech even when the input acoustic signal has a very low Signal to Noise Ratio (SNR). Performance of the algorithm was compared with other similar algorithms on the basis of improvement in intelligibility and SNR of the output, at different SNR levels of the input speech. Wiener filter based algorithm provided significant improvement in SNR and intelligibility compared to other techniques.

Keywords: hearing aid output speech, noise reduction, SNR improvement, Wiener filter, speech enhancement

Procedia PDF Downloads 247

Authors: Samba Raju, Chiluveru, Manoj Tripathy

Abstract:

Intelligibility is an essential characteristic of a speech signal, which is used to help in the understanding of information in speech signal. Background noise in the environment can deteriorate the intelligibility of a recorded speech. In this paper, we presented a simple variance subtracted - variable level discrete wavelet transform, which improve the intelligibility of speech. The proposed algorithm does not require an explicit estimation of noise, i.e., prior knowledge of the noise; hence, it is easy to implement, and it reduces the computational burden. The proposed algorithm decides a separate decomposition level for each frame based on signal dominant and dominant noise criteria. The performance of the proposed algorithm is evaluated with speech intelligibility measure (STOI), and results obtained are compared with Universal Discrete Wavelet Transform (DWT) thresholding and Minimum Mean Square Error (MMSE) methods. The experimental results revealed that the proposed scheme outperformed competing methods

Keywords: discrete wavelet transform, speech intelligibility, STOI, standard deviation

Procedia PDF Downloads 149

Authors: M. Kiran Reddy, K. Sreenivasa Rao

Abstract:

The conventional Hidden Markov Model (HMM)-based speech synthesis system (HTS) uses only a pulse excitation model, which significantly differs from natural excitation signal. Hence, buzziness can be perceived in the speech generated using HTS. This paper proposes an efficient excitation modeling method that can significantly reduce the buzziness, and improve the quality of HMM-based speech synthesis. The proposed approach models the pitch-synchronous residual frames extracted from the residual excitation signal. Each pitch synchronous residual frame is parameterized using 30 wavelet coefficients. These 30 wavelet coefficients are found to accurately capture the perceptually important information present in the residual waveform. In synthesis phase, the residual frames are reconstructed from the generated wavelet coefficients and are pitch-synchronously overlap-added to generate the excitation signal. The proposed excitation modeling method is integrated into HMM-based speech synthesis system. Evaluation results indicate that the speech synthesized by the proposed excitation model is significantly better than the speech generated using state-of-the-art excitation modeling methods.

Keywords: excitation modeling, hidden Markov models, pitch-synchronous frames, speech synthesis, wavelet coefficients

Procedia PDF Downloads 249

Authors: Ahmed Kamil Hasan Al-Ali, Bouchra Senadji, Ganesh Naik

Abstract:

We propose a system to real environmental noise and channel mismatch for forensic speaker verification systems. This method is based on suppressing various types of real environmental noise by using independent component analysis (ICA) algorithm. The enhanced speech signal is applied to mel frequency cepstral coefficients (MFCC) or MFCC feature warping to extract the essential characteristics of the speech signal. Channel effects are reduced using an intermediate vector (i-vector) and probabilistic linear discriminant analysis (PLDA) approach for classification. The proposed algorithm is evaluated by using an Australian forensic voice comparison database, combined with car, street and home noises from QUT-NOISE at a signal to noise ratio (SNR) ranging from -10 dB to 10 dB. Experimental results indicate that the MFCC feature warping-ICA achieves a reduction in equal error rate about (48.22%, 44.66%, and 50.07%) over using MFCC feature warping when the test speech signals are corrupted with random sessions of street, car, and home noises at -10 dB SNR.

Keywords: noisy forensic speaker verification, ICA algorithm, MFCC, MFCC feature warping

Procedia PDF Downloads 408

Authors: Fawaz S. Al-Anzi, Dia AbuZeina

Abstract:

Speech recognition is of an important contribution in promoting new technologies in human computer interaction. Today, there is a growing need to employ speech technology in daily life and business activities. However, speech recognition is a challenging task that requires different stages before obtaining the desired output. Among automatic speech recognition (ASR) components is the feature extraction process, which parameterizes the speech signal to produce the corresponding feature vectors. Feature extraction process aims at approximating the linguistic content that is conveyed by the input speech signal. In speech processing field, there are several methods to extract speech features, however, Mel Frequency Cepstral Coefficients (MFCC) is the popular technique. It has been long observed that the MFCC is dominantly used in the well-known recognizers such as the Carnegie Mellon University (CMU) Sphinx and the Markov Model Toolkit (HTK). Hence, this paper focuses on the MFCC method as the standard choice to identify the different speech segments in order to obtain the language phonemes for further training and decoding steps. Due to MFCC good performance, the previous studies show that the MFCC dominates the Arabic ASR research. In this paper, we demonstrate MFCC as well as the intermediate steps that are performed to get these coefficients using the HTK toolkit.

Keywords: speech recognition, acoustic features, mel frequency, cepstral coefficients

Procedia PDF Downloads 260

Authors: Pogula Rakesh, T. Kishore Kumar

Abstract:

Speech enhancement is a long standing problem with numerous applications like teleconferencing, VoIP, hearing aids, and speech recognition. The motivation behind this research work is to obtain a clean speech signal of higher quality by applying the optimal noise cancellation technique. Real-time adaptive filtering algorithms seem to be the best candidate among all categories of the speech enhancement methods. In this paper, we propose a speech enhancement method based on Recursive Least Squares (RLS) adaptive filter of speech signals. Experiments were performed on noisy data which was prepared by adding AWGN, Babble and Pink noise to clean speech samples at -5dB, 0dB, 5dB, and 10dB SNR levels. We then compare the noise cancellation performance of proposed RLS algorithm with existing NLMS algorithm in terms of Mean Squared Error (MSE), Signal to Noise ratio (SNR), and SNR loss. Based on the performance evaluation, the proposed RLS algorithm was found to be a better optimal noise cancellation technique for speech signals.

Keywords: adaptive filter, adaptive noise canceller, mean squared error, noise reduction, NLMS, RLS, SNR, SNR loss

Procedia PDF Downloads 483

Authors: Hai Quang Hong Dam, Hai Ho, Minh Hoang Le Ngo

Abstract:

This paper investigates the problem of blind speech separation from the speech mixture of two speakers. A voice activity detector employing the Steered Response Power - Phase Transform (SRP-PHAT) is presented for detecting the activity information of speech sources and then the desired speech signals are extracted from the speech mixture by using an optimal beamformer. For evaluation, the algorithm effectiveness, a simulation using real speech recordings had been performed in a double-talk situation where two speakers are active all the time. Evaluations show that the proposed blind speech separation algorithm offers a good interference suppression level whilst maintaining a low distortion level of the desired signal.

Keywords: blind speech separation, voice activity detector, SRP-PHAT, optimal beamformer

Procedia PDF Downloads 283

Authors: Chafik Barnoussi, Mourad Talbi, Adnane Cherif

Abstract:

In this paper we propose a new speech compression system based on the application of the Bionic Wavelet Transform (BWT) combined with the psychoacoustic model. This compression system is a modified version of the compression system using a MDCT (Modified Discrete Cosine Transform) filter banks of 32 filters each and the psychoacoustic model. This modification consists in replacing the banks of the MDCT filter banks by the bionic wavelet coefficients which are obtained from the application of the BWT to the speech signal to be compressed. These two methods are evaluated and compared with each other by computing bits before and bits after compression. They are tested on different speech signals and the obtained simulation results show that the proposed technique outperforms the second technique and this in term of compressed file size. In term of SNR, PSNR and NRMSE, the outputs speech signals of the proposed compression system are with acceptable quality. In term of PESQ and speech signal intelligibility, the proposed speech compression technique permits to obtain reconstructed speech signals with good quality.

Keywords: speech compression, bionic wavelet transform, filterbanks, psychoacoustic model

Procedia PDF Downloads 384

Authors: M.R Asghari Bejestani, Gh. R. Mohammad Khani, V.R. Nafisi

Abstract:

Imagined speech recognition is one of the most interesting approaches to BCI development and a lot of works have been done in this area. Many different experiments have been designed and hundreds of combinations of feature extraction methods and classifiers have been examined. Reported classification accuracies range from the chance level to more than 90%. Based on non-stationary nature of brain signals, we have introduced 3 classification modes according to time difference in inter and intra-class samples. The modes can explain the diversity of reported results and predict the range of expected classification accuracies from the brain signal accusation procedure. In this paper, a few samples are illustrated by inspecting results of some previous works.

Keywords: brain computer interface, silent talk, imagined speech, classification, signal processing

Procedia PDF Downloads 153

Authors: Jing Wu, Wei Lv, Yibing Li, Yuanfan You

Abstract:

The separation of speech signals has become a research hotspot in the field of signal processing in recent years. It has many applications and influences in teleconferencing, hearing aids, speech recognition of machines and so on. The sounds received are usually noisy. The issue of identifying the sounds of interest and obtaining clear sounds in such an environment becomes a problem worth exploring, that is, the problem of blind source separation. This paper focuses on the under-determined blind source separation (UBSS). Sparse component analysis is generally used for the problem of under-determined blind source separation. The method is mainly divided into two parts. Firstly, the clustering algorithm is used to estimate the mixing matrix according to the observed signals. Then the signal is separated based on the known mixing matrix. In this paper, the problem of mixing matrix estimation is studied. This paper proposes an improved algorithm to estimate the mixing matrix for speech signals in the UBSS model. The traditional potential algorithm is not accurate for the mixing matrix estimation, especially for low signal-to noise ratio (SNR).In response to this problem, this paper considers the idea of an improved potential function method to estimate the mixing matrix. The algorithm not only avoids the inuence of insufficient prior information in traditional clustering algorithm, but also improves the estimation accuracy of mixing matrix. This paper takes the mixing of four speech signals into two channels as an example. The results of simulations show that the approach in this paper not only improves the accuracy of estimation, but also applies to any mixing matrix.

Keywords: DBSCAN, potential function, speech signal, the UBSS model

Procedia PDF Downloads 135

Authors: Mahgfirah Firdaus Soberatta

Abstract:

Language cannot be separated from human life. Humans use language to convey ideas, thoughts, and feelings. We can use words for different things for example like asserted, advising, promise, give opinions, hopes, etc. Propaganda is an attempt which seeks to obtain stable behavior to adopt everyone to his everyday life. It also controls the thoughts and attitudes of individuals in social settings permanent. In this research, the writer will discuss about the speech act in a propaganda speech delivered by Martin Luther King Jr. in Washington at Lincoln Memorial on August 28, 1963. 'I Have a Dream' is a public speech delivered by American civil rights activist MLK, he calls from an end to racism in USA. In this research, the writer uses Searle theory to analyze the types of illocutionary speech act that used by Martin Luther King Jr. in his propaganda speech. In this research, the writer uses a qualitative method described in descriptive, because the research wants to describe and explain the types of illocutionary speech acts used by Martin Luther King Jr. in his propaganda speech. The findings indicate that there are five types of speech acts in Martin Luther King Jr. speech. MLK also used direct speech and indirect speech in his propaganda speech. However, direct speech is the dominant speech act that MLK used in his propaganda speech. It is hoped that this research is useful for the readers to enrich their knowledge in a particular field of pragmatic speech acts.

Keywords: speech act, propaganda, Martin Luther King Jr., speech

Procedia PDF Downloads 443

Authors: Fathi Kallel, Ahmed Ben Hamida, Christian Berger-Vachon

Abstract:

In this paper, a Speech Enhancement Algorithm based on Multi-Band Spectral Subtraction (MBSS) principle is evaluated for Bilateral Cochlear Implant (BCI) users. Specifically, dual-channel noise power spectral estimation algorithm using Power Spectral Densities (PSD) and Cross Power Spectral Densities (CPSD) of the observed signals is studied. The enhanced speech signal is obtained using Dual-Channel Multi-Band Spectral Subtraction ‘DC-MBSS’ algorithm. For performance evaluation, objective speech assessment test relying on Perceptual Evaluation of Speech Quality (PESQ) score is performed to fix the optimal number of frequency bands needed in DC-MBSS algorithm. In order to evaluate the speech intelligibility, subjective listening tests are assessed with 3 deafened BCI patients. Experimental results obtained using French Lafon database corrupted by an additive babble noise at different Signal-to-Noise Ratios (SNR) showed that DC-MBSS algorithm improves speech understanding for single and multiple interfering noise sources.

Keywords: speech enhancement, spectral substracion, noise estimation, cochlear impalnt

Procedia PDF Downloads 549

Authors: Panprae Bunyapukkna

Abstract:

This study investigated figures of speech used in fragrance advertising captions on the Internet. The objectives of the study were to find out the frequencies of figures of speech in fragrance advertising captions and the types of figures of speech most commonly applied in captions. The relation between figures of speech and fragrance was also examined in order to analyze how figures of speech were used to represent fragrance. Thirty-five fragrance advertisements were randomly selected from the Internet. Content analysis was applied in order to consider the relation between figures of speech and fragrance. The results showed that figures of speech were found in almost every fragrance advertisement except one advertisement of Lancôme. Thirty-four fragrance advertising captions used at least one kind of figure of speech. Metaphor was most frequently found and also most frequently applied in fragrance advertising captions, followed by alliteration, rhyme, simile and personification, and hyperbole respectively.

Keywords: advertising speech, fragrance advertisements, figures of speech, metaphor

Procedia PDF Downloads 241

Authors: Tusar Kanti Dash, Ganapati Panda

Abstract:

The evaluation of speech quality and intelligence is critical to the overall effectiveness of the Speech Enhancement Algorithms. Several intrusive and non-intrusive measures are employed to calculate these parameters. Non-Intrusive Evaluation is most challenging as, very often, the reference clean speech data is not available. In this paper, a novel non-intrusive speech evaluation measure is proposed using audio features derived from the Stockwell transform. These features are used with the Light Gradient Boosting Machine for the effective prediction of speech quality and intelligibility. The proposed model is analyzed using noisy and reverberant speech from four databases, and the results are compared with the standard Intrusive Evaluation Measures. It is observed from the comparative analysis that the proposed model is performing better than the standard Non-Intrusive models.

Keywords: non-Intrusive speech evaluation, S-transform, light GBM, speech quality, and intelligibility

Procedia PDF Downloads 260

Authors: Jalal Karam

Abstract:

The methods of Fourier, Laplace, and Wavelet Transforms provide transfer functions and relationships between the input and the output signals in linear time invariant systems. This paper shows the equivalence among these three methods and in each case presenting an application of the appropriate (Fourier, Laplace or Wavelet) to the convolution theorem. In addition, it is shown that the same holds for a direct integration method. The Biorthogonal wavelets Bior3.5 and Bior3.9 are examined and the zeros distribution of their polynomials associated filters are located. This paper also presents the significance of utilizing wavelets as effective tools in processing speech signals for common multimedia applications in general, and for recognition and compression in particular. Theoretically and practically, wavelets have proved to be effective and competitive. The practical use of the Continuous Wavelet Transform (CWT) in processing and analysis of speech is then presented along with explanations of how the human ear can be thought of as a natural wavelet transformer of speech. This generates a variety of approaches for applying the (CWT) to many paradigms analysing speech, sound and music. For perception, the flexibility of implementation of this transform allows the construction of numerous scales and we include two of them. Results for speech recognition and speech compression are then included.

Keywords: continuous wavelet transform, biorthogonal wavelets, speech perception, recognition and compression

Procedia PDF Downloads 417

Authors: Yunbin Deng

Abstract:

Most existing applications of automatic speech recognition relies on cooperative subjects at a short distance to a microphone. Standoff speech recognition using microphone arrays can extend the subject to sensor distance somewhat, but it is still limited to only a few feet. As such, most deployed applications of standoff speech recognitions are limited to indoor use at short range. Moreover, these applications require air passway between the subject and the sensor to achieve reasonable signal to noise ratio. This study reports long range (50 feet) automatic speech recognition experiments using a Laser Doppler Vibrometer (LDV) sensor. This study shows that the LDV sensor modality can extend the speech acquisition standoff distance far beyond microphone arrays to hundreds of feet. In addition, LDV enables 'listening' through the windows for uncooperative subjects. This enables new capabilities in automatic audio and speech intelligence, surveillance, and reconnaissance (ISR) for law enforcement, homeland security and counter terrorism applications. The Polytec LDV model OFV-505 is used in this study. To investigate the impact of different vibrating materials, five parallel LDV speech corpora, each consisting of 630 speakers, are collected from the vibrations of a glass window, a metal plate, a plastic box, a wood slate, and a concrete wall. These are the common materials the application could encounter in a daily life. These data were compared with the microphone counterpart to manifest the impact of various materials on the spectrum of the LDV speech signal. State of the art deep neural network modeling approaches is used to conduct continuous speaker independent speech recognition on these LDV speech datasets. Preliminary phoneme recognition results using time-delay neural network, bi-directional long short term memory, and model fusion shows great promise of using LDV for long range speech recognition. To author’s best knowledge, this is the first time an LDV is reported for long distance speech recognition application.

Keywords: covert speech acquisition, distant speech recognition, DSR, laser Doppler vibrometer, LDV, speech intelligence surveillance and reconnaissance, ISR

Procedia PDF Downloads 180

Authors: Punnoose A. K.

Abstract:

This paper presents an approach for identifying noisy speech recording using a multi-layer perception (MLP) trained to predict phonemes from acoustic features. Characteristics of the MLP posteriors are explored for clean speech and noisy speech at the frame level. Appropriate density functions are used to fit the softmax probability of the clean and noisy speech. A function that takes into account the ratio of the softmax probability density of noisy speech to clean speech is formulated. These phoneme independent scoring is weighted using a phoneme-specific weightage to make the scoring more robust. Simple thresholding is used to identify the noisy speech recording from the clean speech recordings. The approach is benchmarked on standard databases, with a focus on precision.

Keywords: noisy speech identification, speech pre-processing, noise robustness, feature engineering

Procedia PDF Downloads 128

Authors: Maha Saliba Foster

Abstract:

Speech perception has been recognized as a bi-sensory process involving the auditory and visual channels. Compared to the auditory modality, the contribution of the visual signal to speech perception is not very well understood. Studying how the visual modality affects speech recognition can have pedagogical implications in second language learning, as well as clinical application in speech therapy. The current investigation explores the potential effect of speech visual cues on the perception of Arabic emphatics (AEs). The corpus consists of 36 minimal pairs each containing two contrasting consonants, an AE versus a non-emphatic (NE). Movies of four Lebanese speakers were edited to allow perceivers to have partial view of facial regions: lips only, lips-cheeks, lips-chin, lips-cheeks-chin, lips-cheeks-chin-neck. In the absence of any auditory information and relying solely on visual speech, perceivers were above chance at correctly identifying AEs or NEs across vowel contexts; moreover, the models were able to predict the probability of perceivers’ accuracy in identifying some of the COIs produced by certain speakers; additionally, results showed an overlap between the measurements selected by the computer and those selected by human perceivers. The lack of significant face effect on the perception of AEs seems to point to the lips, present in all of the videos, as the most important and often sufficient facial feature for emphasis recognition. Future investigations will aim at refining the analyses of visual cues used by perceivers by using Principal Component Analysis and including time evolution of facial feature measurements.

Keywords: Arabic emphatics, machine learning, speech perception, visual speech perception

Procedia PDF Downloads 307

Authors: Negar Riazifar, Nigel G. Stocks

Abstract:

This paper proposes strategies in level crossing (LC) sampling and reconstruction that provide high fidelity signal reconstruction for speech signals; these strategies circumvent the problem of exponentially increasing number of samples as the bit-depth is increased and hence are highly efficient. Specifically, the results indicate that the distribution of the intervals between samples is one of the key factors in the quality of signal reconstruction; including samples with short intervals do not improve the accuracy of the signal reconstruction, whilst samples with large intervals lead to numerical instability. The proposed sampling method, termed reduced conventional level crossing (RCLC) sampling, exploits redundancy between samples to improve the efficiency of the sampling without compromising performance. A reconstruction technique is also proposed that enhances the numerical stability through linear interpolation of samples separated by large intervals. Interpolation is demonstrated to improve the accuracy of the signal reconstruction in addition to the numerical stability. We further demonstrate that the RCLC and interpolation methods can give useful levels of signal recovery even if the average sampling rate is less than the Nyquist rate.

Keywords: level crossing sampling, numerical stability, speech processing, trigonometric polynomial

Procedia PDF Downloads 146

Authors: Mumtaz Begum Mustafa, Siti Salwah Salim, Feizal Dani Rahman

Abstract:

Recently, Automatic Speech Recognition (ASR) systems were used to assist children in language acquisition as it has the ability to detect human speech signal. Despite the benefits offered by the ASR system, there is a lack of ASR systems for Malay-speaking children. One of the contributing factors for this is the lack of continuous speech database for the target users. Though cross-lingual adaptation is a common solution for developing ASR systems for under-resourced language, it is not viable for children as there are very limited speech databases as a source model. In this research, we propose a two-stage adaptation for the development of ASR system for Malay-speaking children using a very limited database. The two stage adaptation comprises the cross-lingual adaptation (first stage) and cross-age adaptation. For the first stage, a well-known speech database that is phonetically rich and balanced, is adapted to the medium-sized Malay adults using supervised MLLR. The second stage adaptation uses the speech acoustic model generated from the first adaptation, and the target database is a small-sized database of the target users. We have measured the performance of the proposed technique using word error rate, and then compare them with the conventional benchmark adaptation. The two stage adaptation proposed in this research has better recognition accuracy as compared to the benchmark adaptation in recognizing children’s speech.

Keywords: Automatic Speech Recognition System, children speech, adaptation, Malay

Procedia PDF Downloads 398

Authors: Shibo Wei, Ting Jiang

Abstract:

Convolutional-recurrent neural networks (CRN) have achieved much success recently in the speech enhancement field. The common processing method is to use the convolution layer to compress the feature space by multiple upsampling and then model the compressed features with the LSTM layer. At last, the enhanced speech is obtained by deconvolution operation to integrate the global information of the speech sequence. However, the feature space compression process may cause the loss of information, so we propose to model the upsampling result of each step with the residual LSTM layer, then join it with the output of the deconvolution layer and input them to the next deconvolution layer, by this way, we want to integrate the global information of speech sequence better. The experimental results show the network model (RES-CRN) we introduce can achieve better performance than LSTM without residual and overlaying LSTM simply in the original CRN in terms of scale-invariant signal-to-distortion ratio (SI-SNR), speech quality (PESQ), and intelligibility (STOI).

Keywords: convolutional-recurrent neural networks, speech enhancement, residual LSTM, SI-SNR

Procedia PDF Downloads 202

Authors: Ana Londral, Burcu Demiray, Marcus Cheetham

Abstract:

Speech recording is a methodology used in many different studies related to cognitive and behaviour research. Modern advances in digital equipment brought the possibility of continuously recording hours of speech in naturalistic environments and building rich sets of sound files. Speech analysis can then extract from these files multiple features for different scopes of research in Language and Communication. However, tools for analysing a large set of sound files and automatically extract relevant features from these files are often inaccessible to researchers that are not familiar with programming languages. Manual analysis is a common alternative, with a high time and efficiency cost. In the analysis of long sound files, the first step is the voice segmentation, i.e. to detect and label segments containing speech. We present a comprehensive methodology aiming to support researchers on voice segmentation, as the first step for data analysis of a big set of sound files. Praat, an open source software, is suggested as a tool to run a voice detection algorithm, label segments and files and extract other quantitative features on a structure of folders containing a large number of sound files. We present the validation of our methodology with a set of 5000 sound files that were collected in the daily life of a group of voluntary participants with age over 65. A smartphone device was used to collect sound using the Electronically Activated Recorder (EAR): an app programmed to record 30-second sound samples that were randomly distributed throughout the day. Results demonstrated that automatic segmentation and labelling of files containing speech segments was 74% faster when compared to a manual analysis performed with two independent coders. Furthermore, the methodology presented allows manual adjustments of voiced segments with visualisation of the sound signal and the automatic extraction of quantitative information on speech. In conclusion, we propose a comprehensive methodology for voice segmentation, to be used by researchers that have to work with large sets of sound files and are not familiar with programming tools.

Keywords: automatic speech analysis, behavior analysis, naturalistic environments, voice segmentation

Procedia PDF Downloads 282

Authors: Van Nhan Nguyen, Harald Holone

Abstract:

Over the past few years, a lot of research has been conducted to bring Automatic Speech Recognition (ASR) into various areas of Air Traffic Control (ATC), such as air traffic control simulation and training, monitoring live operators for with the aim of safety improvements, air traffic controller workload measurement and conducting analysis on large quantities controller-pilot speech. Due to the high accuracy requirements of the ATC context and its unique challenges, automatic speech recognition has not been widely adopted in this field. With the aim of providing a good starting point for researchers who are interested bringing automatic speech recognition into ATC, this paper gives an overview of possibilities and challenges of applying automatic speech recognition in air traffic control. To provide this overview, we present an updated literature review of speech recognition technologies in general, as well as specific approaches relevant to the ATC context. Based on this literature review, criteria for selecting speech recognition approaches for the ATC domain are presented, and remaining challenges and possible solutions are discussed.

Keywords: automatic speech recognition, asr, air traffic control, atc

Procedia PDF Downloads 400